Fakultät für Mathematik und Informatik
Bereich Elektrotechnik und Informationstechnik Elektrotechnik und Informationstechnik Elektrotechnik und Informationstechnik Elektrotechnik und Informationstechnik Elektrotechnik und Informationstechnik Jahresbericht 2008
Annual Report
Table of Contents Vorwort . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
Datenverarbeitungstechnik/Data Processing Technology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 (Prof. Dr. Bernd J. Kr¨amer) Elektrische Energietechnik/Electrical Power Engineering . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 (Prof. Dr. Detlev Hackstein) AG Leistungsbauelemente und Sensorik/Power Devices and Sensors . . . . . . . . . . . . . . . . . 19 (apl. Prof. Dr. rer. nat. Reinhart Job) Elektronische Schaltungen/Electronics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 Informationstechnik/Computer Engineering, esp. Real Time Systems . . . . . . . . . . . . . . . . 31 (Prof. Dr. Dr. Wolfgang A. Halang) Juniorprofessur f¨ ur Eingebettete Systeme/Embedded Systems . . . . . . . . . . . . . . . . . . . . . . . . 41 (Dr. Zhong Li) Kommunikationsnetze/Communication Networks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 (Prof. Dr.-Ing Herwig Unger) Focus Group on Photonics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 (Allgemeine und Theoretische Elektrotechnik, Prof. Dr. Reinhold Pregla) (Optische Nachrichtentechnik, Prof. Dr. J¨ urgen Jahns) (Juniorprofessur f¨ ur Optische Mikrosysteme, Jun.-Prof Dr. Matthias Gruber) Prozeßsteuerung und Regelungstechnik/Control Systems Engineering . . . . . . . . . . . . . . . . 77 (Prof. Dr. Helmut Hoyer, Dr.-Ing. Michael Gerke)
Vorwort - Ist Ingenieursausbildung wirtschaftlich? Jeder, der sich mit mathematischen Methoden in Naturwissenschaften und Technik besch¨aftigt hat, weiss, dass die Optimierung eines aus mehreren Teilsystemen bestehenden Systems die Anwendung globaler“ und nicht nur lokaler“ Optimierungskriterien erfor” ” dert. Leider gibt es auch viele, die dies nicht wissen. Seit Jahren werden die Universit¨aten lokal“ optimiert, man nennt dies Autonomie der Universit¨aten“. Gerne werden dabei ” ” Kriterien in Anspruch genommen, die mit dem Begriff Wirtschaftlichkeit“ in Verbindung ” gebracht werden: Anzahl der Studenten (sollte hoch sein), Kosten der Studienpl¨atze (sollten gering sein). Wer in der ¨okonomisierten Gesellschaft k¨onnte sich dem entziehen? Die ersten Ergebnisse sind schon sichtbar: die Fachbereiche der Ingenieurs- und Naturwissenschaften sind teilweise unter Druck oder gar ins Hintertreffen geraten - nicht nur an der FernUniversit¨at in Hagen - haben sie doch tendenziell weniger Studierende und kosten mehr Geld als andere Bereiche. Ist diese Entwicklung aber wirklich wirtschaftlich“? Dass sie vielleicht zumindest nicht ” ganz im Sinne des Gesamtsystems ist, scheinen in Zeiten der Systemkrise auch diejenigen zu ahnen, welche bisher die Bedeutung des Ingenieursmangels f¨ ur Deutschland u ¨bersehen haben oder sich keine Gedanken dazu gemacht haben. Deutlicher wird das Problem jedoch von Leuten gesehen, deren Wirtschaftlichkeitsorientierung kaum in Frage gestellt werden d¨ urfte. Diese sitzen nicht in Fakult¨aten und Hochschulbeir¨aten, sondern in deutschen Unternehmen und antworten z.B. auf die Frage Welche Ver¨anderung w¨ urde die ” Wettbewerbsf¨ahigkeit Deutschlands am meisten f¨ordern?“ am h¨aufigsten: Ausbildung ” einer h¨oheren Anzahl von Ingenieuren“. So eine aktuelle Studie der Boston Consulting Group [1]. Der seit kurzem vernehmbare Ruf nach verst¨arkter Ingenieursausbildung ist in Zeiten der Krise jedoch eher ein Notruf. Vermutlich d¨ urfte eine sinnvolle Trendwende mehr Maßnahmen in Anspruch nehmen und wohl auch etwas l¨anger dauern als mancher Finanzbuchungstrick bei einer Investmentbank. Das Bem¨ uhen um das Gesamtsystem (oder in hehrer Formulierung: um die gesellschaftliche Funktion der Universit¨aten) muss sich wohl erst wieder Platz im ¨offentlichen Bewusstsein schaffen. Hagen stellt nur einen Teil des Gesamtsystems dar, wir wollen unsere Rolle nicht u ¨bertreiben. Im Bereich der Aus- und Weiterbildung in technischen F¨achern kann die FernUniversit¨at jedoch eine sinnvolle Rolle f¨ ur das Gesamtsystem spielen, insbesondere durch aktuelle Masterangebote. Dass dies m¨oglich ist, zeigt die Entwicklung des Masterstudiengangs Elektrotechnik und Informationstechnik [2]. Dieser 2004 eingerichtete Studiengang w¨achst seit Jahren kontinuierlich und hat derzeit ca. 650 eingeschriebene Studierende sowie eine Abschlussquote von weit u ur einen eher kleinen ¨ber 50% (Stand: WS 2008/09). F¨ Ingenieursbereich sind dies sinnvolle und erfreuliche Zahlen. Soviel zur aktuellen Situation des Bereichs ETIT in Hagen. Erfreulich auch der - zwischenzeitlich angenommene - Ruf an Wilfrid Pascher (vormals Lehrgebiet f¨ ur Allgemeine und Theoretische Elektrotechnik) auf eine Professur f¨ ur Theoretische Elektrotechnik an der Universit¨at der Bundeswehr in M¨ unchen. [1] The Boston Consulting Group, 2009, Perspektiven zum Wirtschaftsstandort Deutschland, Ergebnisse des VI. AmCham Business Barometer. (http://www.amcham.de/publications/studies.html) [2] www.fernuni-hagen.de/mathinf/studium/studiengaenge/master/etit/index.shtml
Hagen, Februar 2009
J¨ urgen Jahns
Chair of
Data Processing Technology (Datenverarbeitungstechnik)
Prof. Dr.-Ing. Bernd J. Krämer
Universitätsstr. 27 58084 Hagen Tel.: +49 2331 987-371 Fax: +49 2331 987-375 E-Mail:
[email protected] http://www.dvt.fernuni-hagen.de
Sta:
Ext.
E-Mail:
Thorsten Blum
4539
[email protected]
Dr.-Ing. Klaus Gotthardt
4537
[email protected]
Steen Groÿ Dr. Peng Han Zhiyun Li Daniel Rudolph Silvia Schreier Daniel Schulte Dr. Uwe Steinmann
03643 582102
[email protected]
4536
[email protected]
1198
[email protected]
03643 582102 4541 1197
[email protected] [email protected] [email protected]
03643 840446
[email protected]
Xia Wang
4540
[email protected]
Stefan Winkler
4314
[email protected]
Volker Winkler
1190
[email protected]
4536
[email protected]
Dr.-Ing. Fan Yang Annett Zobel
03643 582102
[email protected]
Secretary: Renate Zielinski
371
[email protected]
Guest Professor: Prof. Dr. Yusuf Ozturk, San Diego State University, USA (09/08 10/08)
Formal Methods and Prospects of Semantic Computing in Software Engineering Bernd J. Krämer New research activities sail under fashionable brands like semantic Web, semantic Web service, semantic computing, or semantic software engineering. In this research, we investigate the concepts behind these buzzwords and relate them to the classical meaning of the term semantics as the software engineering community established it in the last century. A contemporary interpretation of the concept "semantic software engineering" suggests the application and evolution of semantic computing technologies to all kinds of software artifacts being produced and consumed in a software development process.
Semantic
computing designates an area of research that attempts to unite a range of computational techniques that allow a) the extraction of information from various types of multimedia data, including text, speech, image or video, and b) the meaningful correlation of information fragments originating from disparate sources. Semantic computing has a broader scope than Semantic Web research, both in breadth and depth. Semantic Web focuses on the augmentation of Web documents through machinereadable metadata and related processing methods.
Semantic Web promotes a formal
data representation, thus allowing software agents to grasp some of the intended meaning of data on the Web and meaningfully integrate data from disparate Web resources. To achieve these objectives, Semantic Web technologies are based on model theory to provide semantics of knowledge representation techniques and logics to perform inferences on Web data. Semantic computing addresses the information content of all possible data sources occurring in IT applications, not just Web pages. It builds on a range of base technologies including natural language understanding and processing, speech recognition, text and data mining techniques, or formal concept analysis. Software engineering involves deep problem solv-
ing knowledge.
Unfortunately, a great deal of
problem solving knowledge acquired from software projects only exists in the heads of people,
!"#$
in the form of personal (e-mail) communication, internal memos or otherwise vaguely-formulated ideas and goals.
Thus, a lot of important in-
( $ '#$ ) $#*
formation is eeting or forms an inaccessible part of the archive of project documents. Some software organizations have therefore introduced
% & '#$ % &'#$ + '$& !'#$
knowledge management in software engineering projects. We identi ed a number of opportunities and initial attempts to bring semantic computing techniques, in particular natural language
Figure 1: Model-based software devel-
recognition and processing and knowledge repre-
opment
sentation and reasoning facilities, into software engineering. Our current conclusion is, however, that except for very few initial attempts, semantic computing did not yet penetrate into mainstream software engineering.
[1] B.J. Krämer, T. Margaria: A Hindsight on Formal Methods and Prospects of Semantic Computing in Software Engineering, International Journal of Semantic Computing , Aug 2009
Requipse a research platform for trace retrieval experiments Stefan Winkler Requirements traceability is the ability to follow the life of requirements through the software development lifecycle. Traceability is the basis for many other software development tasks, such as change impact analysis, veri cation, validation system understanding, mainenance, etc. [1]. In practice, traceability can only be achieved by precisely recording each and every step and dependency during development. This is a manual, work-intensive, and error-prone task.
Recent research has shown promising results of a new method of automated as-
sistance for this task: trace retrieval.
Trace retrieval is the application of information
retrieval (IR) methods to the problem of traceability. Using this techniques, interrelated requirements can be identi ed based on a latent semantic analysis: requirements using related terms have a greater chance of being related to each other. Current trace retrieval implementations developed by other researchers have several drawbacks:
•
They are designed to do experiments in a limited area: They are limited to certain algorithms and to certain artifacts or artifact formats
•
They are designed with end-user usability in mind (at the cost of lower extensibility)
•
They are not available for public download.
To overcome these de cits and enable the analysis of dierent aspects of trace retrieval and the in uence of dierent parameters on it, we developed a research platform based on Eclipse. The research platform addresses the issues mentioned above in the following way:
•
it incorporates a basic structural model as a conversion target for theoretically any hierarchically structured artefact format
•
it provides a exible plug-in mechanism based on eclipse to quickly contribute algorithms and to extend the system in most aspects
•
it provides several interfaces that are scriptable at runtime using the Groovy programming language and its seamless Java integration
•
it concentrates on being a research platform, so its strengths do not lie in (industrial) end-user friendliness, but in ease of extensibility
•
it is available for download freely.
The current implementation consists of the basic platform which includes artifact management and a basic set of trace retrieval algorithms, components to import artifacts from DOORS, database tables and OpenO ce text documents, and a component to produce reports that can be used for evaluating the performance of the trace retrieval algorithms. In the coming weeks to months, the platform will be used to perform various experiments on an industrial dataset kindly provided by the Robert Bosch GmbH. This empirical research will also look into various changes to the algorithms like preprocessing using partof-speech-tagging and the application of dierent strategies to identify and use heigher weights for domain-speci c terminology.
[1] Stefan Winkler: On Usability in Requirements Trace Visualizations 3rd Int'l Workshop on Requirements Engineering Visualization , 2008.
Concept Similarity in Heterogeneous Ontologies X. Wang and Y. Zhao To enable interoperability in the semantic web, ontologies are developed by dierent organisations at a large scale, also in overlapping areas. Therefore, ontology mapping has become important to achieve knowledge sharing, semantic integration, information retrieval, query mediation, or web service mapping in an environment where knowledge and information are represented by dierent underlying ontologies. The ontology mapping problem can be de ned as acquiring the relationships holding between the entities of two ontologies. The proposed method to map concepts and properties between ontologies rst applies syntactic analysis based on token strings, and then executes semantic analysis according to WordNet and tree-like graphs representing ontological structures [1]. Experimental results indicate that this algorithm nds mappings with high precision. A weighted rough ontology mapping method based on rough set theory and formal concept analysis rst transforms two source ontologies into formal contexts with linguistic processing techniques [2]. Then the two formal contexts are merged to obtain a single concept lattice and, nally, a rough similarity measure is introduced to produce the ontology mapping results (see Fig. 1).
Figure 2: Mapping results with ontologies Figure 1: Principle of the method proposed
Russia
Determining concept similarity in heterogeneous ontologies is a vital problem for the semantic web. Current approaches only consider single hierarchial concept relations failing to express rich and implied information. In addition, concept similarity under multiple relations as required in many application scenarios of semantic web services has not been investigated, yet. Relying on a method to merge heterogenous ontologies into an application ontology, in [3] a representation model for application ontologies in the form of a semantic net with multiple weighted concept relations was introduced. This approach extends multiple concept relations to measure ontological concept similarity, measures concept similarity by employing concept knowledge from ontology structure and concept information content, and solves the problem of calculating similarity of compound concepts.
[1] Y. Zhao and W.A. Halang: Mapping Ontologies by Utilising Their Semantic Structure, in Encyclopedia of Arti cial Intelligence, J.R. Rabuñal Dopico, J. Dorado de la Calle and A. Pazos Sierra (Eds.), pp. 1049 1055, Hershey-New York: Information Science Reference 2008, ISBN 978-1-59904-849-9 [2] Y. Zhao, X. Wang and W.A. Halang: A Rough Similarity Measure for Ontology Mapping, Proc. 3rd IEEE Intl. Conf. on Internet and Web Applications and Services, A. Mellouk, J. Bi, G. Ortiz, D.K.W. Chu and M. Popescu (Eds.), pp. 136 141, Los Alamitos: IEEE Computer Society Press 2008, ISBN 978-0-7695-3163-2 [3] X. Wang, Y. Zhao and W.A. Halang: Concept Similarity in Multirelational Application Ontologies, in Computational Intelligence in Decision and Control, D. Ruan, J. Montero, J. Lu, L. Martínez, P. D'hondt and E.E. Kerre (Eds.), pp. 169 174, Singapore: World Scienti c 2008, 978-981-279-946-3
oWSD
: A Tool for Word Sense Disambiguation of Ontologies Xia Wang
Word sense disambiguation (WSD) is very important to the Semantic Web/Web 2.0. So far, there is still no easy-to-use tool available. e cient tool called
oWSD
As a remedy, in [1] a simple and very
is demonstrated. It disambiguates the senses of words in their
ontological contexts, and obtains the right word senses from WordNet. It is very helpful to natural language processing and applications involving ontologies.
Figure 1: WSD workbench
The
oWSD
tool was developed as an Eclipse plug-in in Java. It has ve features. For
the rst feature, Importing Ontology, the input is currently limited to ontologies meeting the WSMO standard in the WSML Ontology language. After importing an ontology, it is parsed and all its ontological concepts are extracted. This ontology serves as text context of its concepts. The feature Clean Ontological Concepts is for splitting compound words, clearing their delimiters, and replacing abbreviations by normal words. For the feature
Load Dictionary, WordNet as default dictionary will be loaded. JWord 2.0 is the Java API that is used to access WordNet. The fourth feature is Execute Word Sense Disambiguation. It has been further elaborated into three algorithms: (1) Analysing all concepts in WordNet in order to nd out how many polysemous words, single words, and exceptional words (which cannot be dealt with by WordNet) there are.
The disambiguation algorithm is only executed for the
polysemous words; (2) Generating word windows. The principle is to use the single words in each word window as often as possible in order to improve e ciency; (3) Calculating the concept density by inspecting every subhierarchy of word sense branches retrieved from WordNet. The sense with the highest concept density is taken as the result sense. Finally, there is the feature Data Analysis. It is used to validate this tool by presenting the statistics about results provided by the concept cleaning method and the WSD algorithm. More details illustrated by real examples can be found in [1].
[1] X. Wang and V. Peristeras, oWSD: A Tool for Word Sense Disambiguation in Its Ontology Context, 7th International Semantic Web Conference (ISWC08), Karlsruhe, 2630 October 2008.
soMERGE
: A Tool to Merge Service Ontologies for Semantic Web Services Xia Wang
The existence of an increasing number of dierent ontologies used in the eld of Semantic Web Services (SWS) raises numerous problems of interoperation and communication with respect to service discovery, composition, and execution. Current approaches of ontology mediation often fail due to their lack of su cient semantic expressiveness and reasoning capability.
Merging service ontologies to build an application domain ontology for a
common kind of semantic web services has proven to be an eective approach for this challenge. In this project, the core technology of
W SAO
1
Stuido , the
soMERGE
tool,
was worked out. The mechanism selected for ontology merging is rule-based, and eight rules have been de ned with detailed examples in [1]. In a nutshell, during the process of merging service ontologies, it considers the ontological concepts, concept data types, attributes and instances, and especially it extends concept relations from single to multiple ones, which are able to express potential semantics. A merging example is illustrated by Fig. 1. There, two service ontologies (say so1 and so2 ) in WSML language are imported, merged into 0 a new ontology (say so ) and, nally, cleaned of any con icts, such as name con icts or taxonomy con icts.
Figure 1: WSAO Studio Ontology Merge
Another important feature of the
soMERGE
tool is that it can calculate the ontologi-
cal concept similarity of two service ontologies by fully taking into account their lexical information (a set of labels denoting concepts of the two ontologies) and conceptual information (the taxonomic structures and the relations in the ontologies). This kind of work has laid a solid foundation for other applications of SWSs. For example, it has been proved that semantic service discovery can be improved based on the calculated concept similarity (cp. [2]).
[1] X. Wang, Building Application Ontologies from Descriptions of Semantic Web, Web Intelligence and Agent Systems (WIAS) Journal, accepted 19 October 2008. [2] X. Wang, M. Hauswirth, T. Vitvar and M. Zaremba, Semantic Web Services Selection Improved by Application Ontology with Multiple Concept Relations, The 23rd Annual ACM Symposium on Applied Computing, Fortaleza, Brazil, 1620 March 2008. 1 http://waso.deri.ie
Practices on Construction and Reuse of Process Oriented Service Component Zhiyun Li Service component architecture (SCA) [1] provides a programming model for building applications and systems based on service oriented architecture (SOA). It is based on the idea that business function is provided as a collection of services, which are assembled together to create solutions that serve a particular business need. These composite applications can contain both new services created speci cally for the application and also business function from existing systems and applications, reused as part of the composition. Based on this service component-based model we reconstructed the registration module of the conference management system (CMS) Conftool
2
for the purpose of a case
study [2].
Figure 1: Registration module framework
First we separated dierent functionalities required for a registration process, such as validation of email addresses and duplication check for user names, which have already been implemented in the legacy PHP code, into individual service components.
Then
we constructed a BPEL process that uses methods from these service components. This BPEL process is then encapsulated as a service component, which can be further reused in the PHP-based CMS. Finally, we equipped the service component provided by the BPEL processes with a user interfaces so as to provide a front-end to end users. During an execution of the process, it will collaborate with users through an interface called User Inbox, a concept provided by BPEL4People and WS-HumanTask. The overall architecture of our solution is illustrated as Figure 1. Human interactions are fundamental to many business processes, either for process coordination or task accomplishment. During the execution of a process, end users are often required to take on various activities according to their roles. As in our example, each registrant is required to submit his or her personal information and later con rm the registration data. To enable this man-machine interaction, so we built a customized web-based User Inbox as a companion of process-oriented service component. This way the developed processes component can then be transparently integrated into new applications as long as an entry point to the customized inbox is maintained.
[1] Bernd J. Krämer: Component meets service: what does the mongrel look like?. Innovations in Systems and Software Engineering, Vol. 4, Number 4, 385 394, Spinger 2008. [2] Zhiyun Li, Peng Han: A Preliminary Study on the Construction and Reuse of Process Oriented Service Component. Submitted for publication 2009. 2 http://www.conftool.net
Human-supported Services Daniel Schulte Service-oriented computing (SOC) deals with ideas of how to develop reusable software artifacts, called services (e. g.
web services) and how to build software in-the-large by
orchestrating these artifacts (e. g., by means of BPEL). SOC focuses not only on technical solutions but also on organizational and architectural aspects to bridge the gap between business and IT, and improve business agility among other things. The interaction of users with (orchestrated) services, in particular with business processes, was rst considered by IBM and SAP in 2005 [1] leading to proposals on Human Tasks and BPEL4People. Although solutions based on these proposals are widely-used in industry, this proposals focus on technical XML-based solutions largely omitting questions on architectural styles, particularly with regard to answers enabling people to work with several business systems from dierent organizations comfortably. Moreover, current industrial realizations of these standards limit the implementation options for human tasks. In 2008, Schall et al. introduced the description of services provided by humans comparable to conventional web services, making them e. g. searchable in UDDI like directories [2]. But the simple interaction of humans with services from several independent organizations written in any programming languages and having dierent complexity (from simple approval services to SaaS solutions), is not reached or addressed by current research. An architectural style inspired by emails and depicted in Fig.
1 will allow users
to easily access and implement human-
supported services (i. e., services of any complexity which need the intervention of at least one human): Humans become addressable entities by an unique URI with an assignment list service listening for new tasks (similar to email addresses and mailboxes).
Human-provided services inform
humans via their assignment list service about outstanding (also uniquely addressable) tasks and deliver representations of
Figure 1: Human-provided service based ar-
requested tasks including a solution for re-
chitecture style
porting task progress. Additional identity
management services alleviate the provision of user data and allow the usage of role-based and even workload-based selection of potential task owners. A human assistant may help to receive task assignments, to sort, schedule, access and perform tasks, and to instantiate new processes. A prototype implementation will serve to show the exibility, elegance and simplicity of this architecture style, and will allow us to explore security features and framework development support. It will also be used to study mechanisms for humans to keep track of the execution of their processes andhaving su cient authorizationsto manipulate process instances on the y putting them in control of their processes.
[1] Kloppmann, M.; Koenig, D.; Leymann, F.; Pfau, G.; Rickayzen, A.; von Riegen, C.; Schmidt, P. & Trickovic, I.: WS-BPEL Extension for People BPEL4People. IBM and SAP, 2005. [2] Schall, D.; Truong, H. & Dustdar, S.: Unifying Human and Software Services in Web-Scale Collaborations. IEEE Internet Computing. 2008 May/Jun; 12(3), 62 8.
Location Based Collaborative Information Sharing in Web2.0 Mode Fan Yang The popularization of the global positioning system (GPS) and the ongoing trend of its integration into mobile devices such as smartphones and personal digital assistants (PDAs) has enabled the provision of innovative mobile services. How to provide personalized information sharing and query services according to users' locations has attracted increasing attention, which led to an independent research and application area called location-based service (LBS). In an LBS, location is used as the main criteria to index and organize interesting information for potential users. Web 2.0 can be viewed as a new mode of web applications, in which users are not just information consumers, but also providers. Through a dedicated platform (Web 2.0 applications such as blogs, Web bookmarks and albums), users populate the Internet with contents with more or less personal tastes or preferences, which can then be conveniently shared between users with similar interests. LBS and Web 2.0 complement each other in a way that on one hand, the Web2.0 mode provides an e cient and scalable way to create large amount of interesting and personalized location-aware content for an LBS; on the other hand, LBS also provides a new dimension for Web2.0 applications, through which users can explore an information space and be connected to other similar users.
Figure 1: Reference framework for scalable reuse of animated learning objects
In our work, we propose a location-based collaborative information sharing platform using the Web 2.0 application mode, through which users can exchange useful location-aware information such as sight-seeing tour, local shops and restaurants or public transportation schedule with each other. As illustrated in gure 1, by using a service-oriented computing (SOC) based data and service interface, we aim at providing a uni ed and exible method to integrate existing information and services platform into high-level composite location based services.
Social software techniques, such as collaborative ltering and
self-organizing community construction, will also be used to identify user groups with similar information interests so as to provide better personalized information ltering.
CampusContent: An Infrastructure for Sharing and Reuse of Teaching Experience and e-Content Bernd J. Krämer Investments in the development of high-quality learning content are often lost and successful applications of pedagogical designs are di cult to nd and adapt to other contexts because individuals and groups work in isolation. They also lack eective means to codify and share their assets with peers teaching in the same or a similar subject area. CampusContent is both the name of a project and a DFG-funded center of excellence for e-learning that aimed to evolve the state-of-the-art and provide innovative methods and technology supporting instructors in sharing and reusing high quality learning content and codi ed pedagogical models [1]. It will provide a work environment including community services, authoring tools and learning management systems that operate on top of a content repository. A QTI-compatible o ine editor, which instructors or course designers can use to create question sheets whilst on the move or at home, is also made available for exercises and self-tests. The portal also contains an editor helping in the formulation of learning objectives based on the learning taxonomy embedded (which can be replaced). Open source components, such as the enterprise content management system Alfresco and the enterprise portal and social collaboration software Liferay are used as base components of CampusContent. Figure 1 illustrates the main components included in the portal (delineated by a dashed line).
The
network in the upper right corner suggests that a portal installation's repository can be seamlessly connected to the repositories of other CampusContent portals to form a distributed content repository. Besides content management, networking and other community services, the project has also developed innovative methods to improve ing the
the
reusability
objects inherent
(Los)
and
of
learn-
mitigate
contradiction
be-
tween context-neutral content and
Figure 1: CampusContent portal technology
the necessity of tailoring LOs to the needs of the learner.
These
methods have been demonstrated for selecetd types of LOs that proved to be useful in higher education (see article on the next page). End of March 2009, when the funding period will end, the portal CampusContent will be launched. We will then run pilot applications with four to six pilot institutions to n-tune usability issues and customize the portal interface to dierent user types.
[1] B.J. Krämer, A. Zobel: Rollout of CampusContent: DFG-Competence Centre for eLearning, elearning and education , July 2008, http://eleed.campussource.de/archive/4/1417/
A Methodology of Developing Reusable Animated Learning Objects Peng Han The concept of learning objects (LOs) arose in the early nineties driven by the motivation to reduce the development and maintenance cost of digital learning resources through modularization and reuse. For this purpose, we propose to mitigate the inherent contradiction between context-neutral content and the necessity of tailoring LOs to the needs of the learner by a heuristic principle. It suggests developing content, called information objects (IOs), independently from pedagogical context, such as learning objectives and scenarios, and connecting both ad hoc at reuse time to form meaningful learning objects. In the CampusContent repository we have recently built, all bindings between IOs and pedagogical context that were ever established by CampusContent users are maintained sustainably. Hence, they can be retrieved and inspire new combinations. We illustrate the implementation of this principle through four types of LOs that proved to be useful in higher education. We evolved these resources into generic objects from which custom-designed objects can be generated through combination, parameter con guration, and adaptation [1]. Our rst example, nite automata, is implemented in Java and serves to demonstrate the exibility in combinations we can achieve through the separation of content and pedagogical context. Three further examples of generic objects, interactive concept classi cation, sorting and graph-based search algorithms, are implemented in Adobe's Flash format. Both Java and Flash are two media formats that are operational under a range of operating systems and browsers. From our experience we found that the more exibility a reuse approach achieves, the more requirements it imposes on re-users.
In fact, learning object development with a
high demand in pedagogy and media design is a complex process.
It involves dierent
competencies such as instructional design, media design, programming, and domain expertise. It seems unreasonable for one person to have all these skills. Therefore we think that an eective LO reuse paradigm should be leveraged to a higher degree of productivity by using the best tting technology and exibly organizing the cooperation of necessary competence holders.
Based on his or her own expertise, a re-user can choose the cor-
responding level of customizing learning objects. To support such processes, we aim to provide a collaborative software environment in which re-users with dierent expertise can work together seamlessly.
Figure 1: Reference framework for scalable reuse of animated learning objects
[1] Peng Han, Bernd J. Krämer: Generating Interactive Learning Objects from Con gurable Samples, In: International Conference on Mobile, Hybrid, and On-line Learning , Jan. 2009.
Chair of
Electrical Power Engineering Prof. Dr.-Ing. Detlev Hackstein Universitätsstr. 27 (PRG) D-58084 Hagen Tel. +49 (0)2331 987-1180 Fax: +49 (0)2331 987-357 E-Mail:
[email protected] www.fernuni-hagen.de/eet/
Secretary: Marianne Lehmann,
+49 (0)2331 987-1181,
[email protected]
Henning Bohn,
+49 (0)2331 987-1182,
[email protected]
Ralf Busch,
+49 (0)2331 987-1184,
[email protected]
+49 (0)2331 987-1185,
[email protected]
Scientific Staff:
Technical Staff: Wolfgang Köhler,
Lithium-Ion Batteries – Energy Storage for Extended-Range Electric and Hybrid Vehicles? W. Köhler Department of Mathematics and Computer Science, LG EET, University of Hagen, Germany In 2008 the Chair of Electrical Power Engineering was responsible for the technical analyses of solar vehicles participating at the Tour de Ruhr (June 26th – 29th, 2008), an annually conducted rally for solar vehicles. In frame of the Tour de Ruhr a symposium focusing on new and innovative ideas concerning public mobility and solar vehicles was organized by the NRW Energy Agency (http://www.energieagentur.nrw.de/). Electric vehicles with lithium-ion batteries for energy storage were the highlights of the Tour de Ruhr. One of the vehicles – a 450 kg Twike™ with equipped with a lithium-ion battery and a power density of 7 kW – drove a distance of about 160 km (∼ 100 miles). State of the hybrid vehicles (e. g. the Toyota Prius) exhibit significant lower power densities; and hence, the mobility range of such systems is much smaller. But not only the power densities are important, also the weight of the battery pack is a crucial parameter. With this regard lithium-ion batteries have a much better performance than standard battery systems like metal-hydride batteries. In addition lithium-ion batteries have no memory effect.
Fig. 1: Twike electric vehicle © twike.de
Fig. 2: Lithium-ion battery power pack (20Ah / 353V), © twike.de
Concerning environmental and security aspects, the demands on lithium-ion batteries in vehicles is very high. For everyday suitability it is important that technical problems concerning the lifetime of the batteries and especially security problems – recently lithium-ion batteries tended to inflame during operation – have to be solved. In a car the lithium-ion batteries have to sustain a large range of ambient temperatures, thousand of reload cycles without significant loss of performance, mechanical vibrations and so on. The final goal is that vehicles for private transportation powered by lithium-ion batteries should attain a range of approximately 400 km. Actually the automobile industry focuses on mild-hybrid systems powered by with lithiumion batteries. However, this can be only an interim solution.
Development of a Model for the Analyses and Control of the Thermal Heating of Buildings * R. Busch, R. Job, D. Hackstein Department of Mathematics and Computer Science, LG EET, University of Hagen, Germany Recently, man-made climate change as a matter of fact became obvious. During the 21st century the climate will be significantly changed by greenhouse emission gases – predominantly CO2 – due to the combustion of fossil fuels. Power generation, world wide transportation and mobility as well as heating can be counted to the main cause of climate damage. With regard to a minimization of upcoming impacts caused by climate change, innovative measures have to be taken. Based on this background, in frame of a master-thesis thermal heating of buildings was modeled and analyzed by means of Matlab/SimulinkTM simulations. A complex solar radiation model considering the impact of real weather conditions and the geographic situation was developed for the conducted analyses.
Fig. 1: Simulated profiles of radiation power and temperature (environmental temperature, room temperature, cellar temperature, wall temperatures, etc.) of a test bungalow over a 30 days period in September and October – geographic situation: 52.315° North, 9.432° East
In particular the developed model was employed for the analyses of a simple bungalow test building. To reduce the calculation time, the hyperbolic heat equations were linearized and expressed in the state-space model. Typical simulated temperature and radiation power profiles are shown in Fig. 1. Based on the employed solar radiation model and on accessible weather data of the German Weather Bureau – Deutscher Wetterdienst – realistic radiation and temperature profiles for the course of the day and the year were generated and their impact on the heating requirements was analyzed. The underlying model assumptions and simulations provide significant predictions for the heating necessities of buildings with regard to efficient energy consumption; i. e. based on the investigations discussed in the thesis, it is possible ● to develop a control unit for the heating and hot water supply of a building supported by solar power, which is optimized to a minimization of energy consumption ● to predict alterations of temperature profiles within the building after gentrification ● to ascertain the most appropriate photovoltaic or photo-thermal systems, their location and orientation ● to evaluate energy efficiency based on realistic climatic conditions ● to analyze the costs and benefits of photovoltaic and/or photo-thermal systems * Master thesis of R. Busch (completed in Jan. 2009)
Power Devices and Sensors * apl. Professur für Technologie elektronischer Bauelemente (AG Leistungsbauelemente und Sensorik) apl. Prof. Dr. rer. nat. Reinhart Job Universitätsstr. 27 (PRG) D-58084 Hagen Tel. +49 (0)2331 987-1183 Fax: +49 (0)2331 987-4105 E-Mail:
[email protected] www.fernuni-hagen.de/ag_job/ Secretary: Diana Schulz,
Tel.:
+49 (0)2331 987-4101,
[email protected]
Master-Students: Ralf Busch,
Barsinghausen,
[email protected]
Dirk Falk,
Bad Schönborn,
[email protected]
Alexander Seybold, Elthagen,
[email protected]
Thierry Tchoumi,
Denzlingen,
[email protected]
Heidelberg,
[email protected]
Fredenbeck,
[email protected]
Bachelor-Student: Thomas Kissinger, Diploma-Student: Maik Wolfkühler,
* The work group Power Devices and Sensors is affiliated to the Chair of Electrical Power Engineering (Prof. Dr.-Ing. Detlev Hackstein)
Development of a Laboratory Experiment of a Smoke Sensor System for Master Program Lectures * D. Falk, R. Job Department of Mathematics and Computer Science, AG LS, University of Hagen, Germany In frame of the master program Electrical Engineering and Information Technology at the University of Hagen, the course Sensor Technology actually lectures mainly in the basic detection principles of the most important physically measured quantities and the corresponding sensor devices. A detailed analysis of operation and measuring circuits is up to now not embedded into these lessons. Without such knowledge the state of the art professional application of sensors is not possible, especially concerning mechatronical concepts. Therefore, within a master thesis a laboratory experiment based on a smoke sensor system was developed, which should partly fill this gap in the students knowledge. In detail, the general operation principles as well as the architecture of carrier frequency amplification for the operation of active sensor elements can be studied with the developed laboratory experiment. Phase shift
Measurement Channel Photodiode: scattered light detection Wave generator
Phase sensitive rectifier
Low pass filter
Phase sensitive rectifier
Low pass filter
Controlled light source (LED) Photodiode: direct light detection
Reference Channel
Phase shift
Fig. 1: Block diagram of the developed laboratory experiment; the measurement line delivers a voltage signal, which is dependent on the scattered light (red arrows: optical measuring paths)
The operation principle of the sensor system is based on intensity measurements of the scattered light depending on the particle density of smoke within the optical measuring path. In the block diagram of the laboratory experiment – shown in Fig. 1 – the sensor element consists of the LED light source, the optical measuring path, where the scattered light originating from smoke is created and the photodiode detector for scattered light detection. The carrier frequency amplifier consists of the wave generator, phase shift, phase sensitive rectifier and the low pass filter. With this part of the setup (upper part of Fig. 1), the students should find out, which disturbances can effect the measuring path, and which of the disturbances can be eliminated by the amplifier. Mainly the impact of electronic noise and interfering light (neon light) will be studied. Moreover, the students can also study, which disturbances can be compensated by a reference channel (lower part of Fig. 1). The architecture of the reference line is very similar to the measurement channel; however, the light of the LED is directly measured – without the alteration by smoke. With the reference line the impact of dirtiness (for instance in an industrial environment) or the aging of the diode can be faded out. The laboratory experiment should be executed at the university or at home, for the latter case an experimentation box can be mailed to the students. Within the thesis, for each component prototype cards were developed and tested. Depending on the future application (experiments at the university or at home) appropriate housing of these components is still necessary. With regard to home experiments, each card is configurated with an USB port, which simultaneously records two voltage signals; hence, it is not necessary to employ an oscilloscope. * Master thesis of D. Falk (completed in Feb. 2009)
Implementation of a phase control circuit for an ultrasonic motor * T. Kissinger1, 2, R. Job2 1
2
ISIS sentronics GmbH, Mannheim, Germany Department of Mathematics and Computer Science, AG LS, University of Hagen, Germany
Working principle of an ultrasonic motor: In ultrasonic motors, the periodical movement of a tip mounted on an oscillating piezoelectric crystal drives a sliding rail which carries the motor load. Resonant mechanical oscillations of the crystal at frequencies of around 200 kHz periodically modulate the force of friction exerted by the tip on the sliding rail. Additionally, these oscillations cause a lateral movement of the tip, which over one period moves the tip in one direction when the force of friction is low and in the opposing direction when the force of friction is high. Due to the inertia of the sliding rail a resultant force is produced, which actuates the motor. Fig 1: Ultrasonic motor (adapted from ref. [1])
The same principle can be used for both linear and rotary motion. Ultrasonic motors are often used in precision actuation for robotics or measurement technology because they can achieve accuracy in the micrometer range while having an infinite actuating range and very small physical dimensions. In this project, an ultrasonic motor is used to position the micro-lens of an interferometric distance measurement system (“Optical Coherence Tomography”). Phase monitoring: There is a direct correspondence between the mechanical properties of the motor (speed/torque and energy efficiency/wear) and its electrical impedance. The phase angle of complex impedance in the electrical equivalent circuit of the piezoelectric crystal determines the ratio between effective and reactive power used in the crystal and characterizes the operating point of the motor. This phase can be measured by analyzing the sinusoidal waveforms of the driving voltage and the current drawn by the motor. In real life, local variations in the surface properties of the sliding rail (i.e. friction coefficient and non-even slope) lead to changes in the mechanical oscillation behavior of the motor while running and subsequently alter the phase angle of the electrical impedance on short time scales. It is presumed that in keeping the phase of the electrical impedance constant by changing the driving frequency, the uniformity of the motor speed will increase. Also, energy efficiency and wear properties should improve since the power consumption of the motor should adjust to the current local conditions on the sliding rail. For these reasons a phase control circuit was proposed and has been implemented in this project. Implementation of a phase control circuit: The phase control circuit consists of a feedback loop including a phase detection unit, a control algorithm unit and a driving signal generation unit. The control algorithm tries to control the phase to a given set-point by varying the frequency of the driving signal for the motor. In this project, the control algorithm and support functions were implemented digitally on a FPGA chip and a hardware phase detection circuit was designed. Results: Preliminary results indicate that the phase of the ultrasonic motor can indeed be kept constant by controlling the driving frequency and that this leads to the power consumption of the motor adjusting to the local conditions on the sliding rail. The presumed uniformity of speed still has to be proven. * Bachelor-Thesis of T. Kissinger (completed in Dec. 2008) Ref. [1]: O. Vyshnevskyy, S. Kovalev, W. Wischnewskiy, A Novel, Single-Mode Piezoceramic Plate Actuator for Ultrasonic Linear Motors, IEEE Trans. Ultrason., Ferroelect., Freq. Contr., vol. 52, no. 11, p. 2047-2053, 2005
Modern Methods of Signal Processing for Analysis of Structure-Borne Noise from Bearings Systems * A. Seybold 1, 2, A. Pecher 2, R. Job 1 1
Department of Mathematics and Computer Science, AG LS, University of Hagen, Germany 2 Schaeffler KG, Schweinfurt, Germany
Efficient working ball and roller bearings do not only make sure the integrity of operation, but they also can reduce the energy consumption. Modern diagnostic methods of structure borne noise could primarily contribute to detect potential failures already in the run-up of damages, which enables to cost saving counteractive measures early enough. Furthermore the energy demand of mechanical systems could be significantly reduced. The ambition of this work was to study the significance of acoustic signals from bearing systems by assistance of new signal processing techniques. Defects on bearing parts create characteristics structure borne noises. Such noise originates from over-rolling frequencies, which are carried by structure resonances. This noise can be converted into electric signals by sensors like piezoelectric accelerometers. The resulting analog signal is recorded with sufficient resolution. By assistance of MATLAB™ simulations the signals can be digitally processed by diverse mathematic algorithms. Currently those noises are investigated by envelope demodulation. By this process it is possible to locate defects by means of there peaks in the frequency spectrum. New methods of signal processing could offer more precise detection results, even in an early damage phase. They could possibly break new ground in the analysis of noise and are therefore in focus for a closer investigation. A typical representative of these methods is the Fourier spectra analysis. In these patterns the information about the characteristic signals are embedded.
Fig. 1: Damaged rolling element
Fig. 2: Characteristic damage frequencies
Figure 1 shows a typical proceeding defect of a rolling element from a cylinder roller bearing. This damage leads to characteristic over-rolling frequencies, which can be detected by signal processing methods. Figure 2 exhibits the significant peaks which appear in the analysis of structure borne noise by envelope demodulation.
* Master thesis of Alexander Seybold (completed in Jan. 2009)
Analyses of OTEC-Systems with a Self-Developed Simulation Tool * T. Tchoumi, R. Job Department of Mathematics and Computer Science, AG LS, University of Hagen, Germany In a first order approximation the vertical temperature distribution in the open ocean can be simply described as two layers separated by an interface. The upper layer is warmed up by the sun; and under such model assumptions the warm temperature is distributed down to a depth of about 100 m depending on the geographical location. The bottom layer, i. e. the deep ocean regions, consists of rather cold water (e. g. 4 °C). The interface is sometimes marked by an abrupt change in temperature, but more often the change is gradual. The temperature difference between the upper (warm) and bottom (cold) layers ranges from 10 °C to 25 °C, with the higher values found in equatorial waters. A practical application is found in a system of a thermal power cycle (heat engine) designed to transform the thermal energy into electricity; and it is based on a Claude cycle, which uses flash evaporation of seawater under a partial vacuum. The system is referred to as OTEC-system, where OTEC stands for Ocean Thermal Energy Conversion. A special benefit of OTEC-systems is, that unlike most renewable energies it is base-load, i. e. the thermal resource of the ocean ensures that this power source is available day and night, and concerning equatorial locations with only modest variations between summer and winter.
Fig. 1: Simulation of an OTEC-system (screenshot)
Within a Master Thesis a graphic user interface (Fig. 1) for an OTEC power plant is developed. Moreover, the operation of an OTEC-system – i. e. the complete OTEC power plant – is simulated under the assumption of real conditions. Based on the employed simulations the OTEC-system can be analyzed for instance by • recording of the electricity power by different pump settings • recording of the electricity power by various temperature differences • recording of the electric power by the quantity of water, which is used within the applied thermodynamic cycle * Master thesis of T. Tchoumi (to be completed in July 2009)
Systemic Analysis of Power Generation with Biomass * M. Wolfkühler, R. Job Department of Mathematics and Computer Science, AG LS, University of Hagen, Germany In frame of a systemic investigation the overall demand of resources for the generation of thermal and/or electrical power with biomass and the related input-oriented environmental impact is studied. The study is based on the MIPS-concept, where MIPS means material input per service unit. MIPS stands for a measure that was decisively developed at the Wuppertal Institute for Climate, Environment and Energy (http://www.wupperinst.org/de/home/), and it is an indicator for preventive environmental protection. The decision for the choice of a certain biofuel for the generation of thermal and/or electrical energy depends strongly on the base material; hence, the availability of the initial biomass, environmental aspects, costs, and even ethical aspects have to be taken into account. Taking the materials input as the main criterion instead of the output (waste, emissions, etc.), some serious advantages can be addressed. As an example we name the following two points: First of all, reducing the material input (including also the energy input) automatically reduces the output, i. e. waste, emissions, etc. A second very important point is that concerning the harmful properties of the output, its impact on the environment and on humans and animals is simply not known for the vast majority of materials set free by human activities. Only a very minor part of materials, substances and emissions is investigated in detail with this regard. Hence, if the knowledge of the output is strongly limited, negative impact and consequences can be leastwise significantly and automatically reduced by the reduction of the overall input. Energy Crop Plants
Crop Residues
Organic Byproducts
Organic Waste
Preparation
Refining
Transportation
Storage
Anaerobic Decay
Solid Fuel
Gaseous Fuel
Combustion
Thermal Energy / Electrical Energy
Fig. 1: Process chain for a power generation based on biomass
Complex process analyses, which consider the complete consumption of resources and evaluate all material demand coverage, require a systemic way of reflection. The starting point of the conducted analyses is the appointment of the biofuels, to be compared in the study. These biofuels are strongly related to the basic biomass input, which is used for the biofuel production. The base for the comparative study is the energy content of the used biomass quality. The material turn-over and the input of all resources – including soil, water and air – as well as the outputs are analyzed for all process steps, where outputs become – sometimes only partially – the inputs of the following process steps. The overall process chain has to be defined to model the complete material input from ‘cradle to product’. * Diploma thesis of M. Wolfkühler (to be completed in July 2009)
1st Yaoundé International College on Novel Materials and Technologies, and their Impact on Energy, Environment and Sustainable Development R. Job Department of Mathematics and Computer Science, AG LS, University of Hagen, Germany In July 2008 for the first time, an international conference and summer school on Materials Science – the ″1st Yaoundé College on Novel Material and Technologies, and their impact on Energy, Environment and Sustainable Development″ – was organized in Yaoundé the capital of Cameroon. The college was organized by Prof. Dr. Jean-Marie Njaka (University of Yaoundé 1, Cameroon) and Prof. Dr. Esidor Ntsoenzok (University of Orléans and CNRS, Orléans, France). The author had the pleasure to participate in the international scientific committee and in the organization committee of the College as well as an invited lecturer. The College brought international experts from Congo, France, Germany, Morocco, Nigeria and the United States together with local faculty members and students from the Universities of Yaoundé and Douala in Cameroon. During the conference fruitful discussions were established creating new research and educational collaborations between African researchers and the participating universities and laboratories abroad.
Fig. 1: Group photograph of the ″1st Yaoundé International College on Novel Materials and Technologies″
Fig. 2: R. Job presenting a tutorial lecture dressed with the official college T-shirt
During a very intense meeting, the audience received a large variety of information in various fields of materials research ranging from basic concepts to the latest, state-of-the art research on semiconductors. Lectures were also given on environmental impacts, nuclear waste management and new technological issues facing our planet, including aspects of global warming, fuel and energy efficiency research, coming along with interesting topics on medical related research, e. g. the use of plants in Africa for health and economy. Approximately 150 participants attended the College, eighteen oral tutorial and review talks were presented as well as a large variety of contributed talks and four poster sessions. The tutorial and review talks were presented by international experts in the field of Materials Science: S. Ashok (Penn State University, USA), K. Awitor (University Blaise Pascal of Clermont-Ferrand, France), C. Bonafos (CNRS, Toulouse, France), J. Chevalier (University of Versailles and CNRS, Meudon, France), F. Boujrhal (University of Rabat, Morocco), R. Elenga (University Marien Ngouabi, Congo), F. Ezema (University of Nigeria, Nigeria), P. Guèye (Hampton University, USA), R. Job (University of Hagen, Germany), E. Leoni (CNRS Orleans, France), B. Malonda-Boungou (University Marien Ngouabi, Congo), D. Mathiot (University Louis Pasteur of Strasbourg, France), C. Mbane Biouele (University of Yaoundé I, Cameroon), P. Moussounda (University Marien Ngouabi, Congo), B. M’Passi-Mabiala (University Marien Ngouabi, Congo), L. Ottaviani (University of Marseille, France), G. Regula (University of Marseille, France), and A. Thiaville (University Paris-Sud, France).
Demands and Challenges for a Sustainable Energy Supply Concept * R. Job Department of Mathematics and Computer Science, AG LS, University of Hagen, Germany The 21st century is facing an extraordinary challenge concerning the global energy supply concepts. With growing economies and especially with the uprise of the new global players China and India, with today’s global markets and logistics, there is a strong demand for rapidly growing energy supply. However, on the other hand – as a consequence thereof – global warming occurs and dramatically will affect the wellbeing of people all over the world. Based on this background, for the ″1st Yaoundé College on Novel Material and Technologies, and their impact on Energy, Environment and Sustainable Development″ – an international conference and summer school on Materials Science and related subjects, which was held in Yaoundé (Cameroon) in July 2008 – a tutorial lecture was developed. This tutorial addressed various technical key items, which play an essential role for sustainable power supply concepts for the next decades. Moreover the CO2 problem and global warming were briefly summarized and discussed. The decline of the mineral oil era, i. e. the peak oil dilemma, was discussed as well as new and alternative energy concepts for the future (solar, wind, geothermal, nuclear, etc.). Finally, also life cycle analyses were made a subject of discussion.
Demands and Challenges for a Sustainable Energy Supply Concept - Tutorial Reinhart Job, University of Hagen, Germany
1st Yaoundé International College on Novel Materials and Technologies and their Impact on Energy, Environment and Sustainable Development University of Yaoundé I, July, 7th – 11th, 2008, Cameroon
Climate change (the IPCC scenario)
The peak oil dilemma
The facts Þ Massive increase of greenhouse gas concentrations (CO2, CH4, N2O, …) caused by anthropogenic activities (during the past 650.000 years the CO2 concentration was not as high as today) Þ Increase of greenhouse gas concentration ⇒ intensification of the greenhouse effect and global warming Þ Global warming ⇒ rising of extreme weather conditions during the past decades worldwide Þ Global warming ⇒ reduction of snow and ice covering especially on the northern hemisphere Þ Global warming ⇒ warming of the oceans and increase of the mean sea level (up to now already 10 – 20 cm)
Estimation of the peak oil: • Africa: 2004 • Asia: 2002 • Europe: 2000 • Middle East: 2011 • North Amarica: 1985 • South Amerika: 2005
Folie 17
July 10th, 2008
Prof. Dr. rer. nat. Reinhart Job, Power Devices & Sensors
Faculty of Mathematics and Computer Science
Folie 53
July 10th, 2008
Carriers & Storage • Electric power grid • Micro-power materials • Electrical energy storage • Catalysts • etc. Use & Efficiency • Transportation • Buildings • Lightning • Industry • etc.
Prof. Dr. rer. nat. Reinhart Job, Power Devices & Sensors
• World peak:
2006
Folie 37
July 10th, 2008
500 m
Various systems: 1: Heat collector (heat pump) 2: Heat probe (heating) 3: Deep heat probe (heating, electricity) 4: Hydrothermal reservoir (electricity, heating) 5: Hot-Dry-Rock method (electricity, heating) Folie 72
July 10th, 2008
Prof. Dr. rer. nat. Reinhart Job, Power Devices & Sensors
Faculty of Mathematics and Computer Science
Are we on the right track?
Geothermal energy Advanced technology • Coal • CO2 storage • Oil • Gas • etc.
2003 2009
Faculty of Mathematics and Computer Science
- Lifecycle analysis and the concept of MIPS
0m
Resources • Solar energy • Wind energy • Bioenergy • Hydropower • Geothermal energy • Ocean energy • Methane hydrates • Hydrogen fuel • Nuclear power • etc.
• Non-OPEC: • OPEC:
! !!! s !! Up 2008 s It‘
R. C. Duncan, W. Youngquist, The World Petroleum Life-Cycle (1998) http://www.dieoff.com/page133.pdf
Prof. Dr. rer. nat. Reinhart Job, Power Devices & Sensors
Towards a sustainable energy concept
Faculty of Mathematics and Computer Science
Faculty of Mathematics and Computer Science
Faculty of Mathematics and Computer Science
1
2
1000 m 1500 m 2000 m
3 permeable sedimentary rock
4
2500 m 3000 m
non-permeable igneous rock / granite
3500 m 4000 m
5
4500 m
Ecological backpack of a T-shirt (weight of the T-shirt: 0.17 kg) *: • Abiotic raw materials: 2.0 kg • Biotic raw materials: 1.2 kg • Water: 1480.0 kg • Air: 12.5 kg • Erosion: 223.0 kg (included → 2.83 kWh electrical current) ________________________________ Σ 1718.7 kg
5000 m
⇒ Ecological backback of a T-shirt: 1718.7 kg – 0.17 kg = 1718.53 kg
5500 m
* M. Ritthoff, H. Rohn, C. Liedtke (in cooperation with T. Merten), Calculating MIPS, Wuppertal Institute for Climate, Environment and Energy (2002)
6000 m
Prof. Dr. rer. nat. Reinhart Job, Power Devices & Sensors
Folie 124 July 10th, 2008
Prof. Dr. rer. nat. Reinhart Job, Power Devices & Sensors
Fig. 1: Some examples of Power-Point pages giving an impression of the content of the tutorial
The general aim of the tutorial was focused on a systemic overview, where especially the interactions of various related topics were punctuated. Off course, the tutorial could not present a general solution for a sustainable energy supply. Certainly, based on the presented facts, some fundamental correlations were explained and discussed in detail. Especially, various arguments, which are presented again and again by politicians and managers from industry – for instance concerning oil resources or others – are critically evaluated. Based on fundamental physical and chemical knowledge and/or geological facts a rather large amount of such political statements can be quite often reduced to absurdity – at least partly. The private opinion of the author is that serious critical efforts and analyses, based on scientific facts and not on political and particular economic interests or the dogmatic political believe in globalized economy and free markets are urgently needed, to come to an acceptable solution for all people all over the world. * Invited Tutorial presented at the "1st Yaounde International College on Novel Materials and Technologies, and their Impact on Energy, Environment and Sustainable Development", July 7th – 11th, 2008, Yaoundé, Cameroon
Formation and Annihilation of Hydrogen-Related Donor States in Proton-Implanted and Subsequently Plasma-Hydrogenated N-Type Float-Zone Silicon * R. Job1, F.-J. Niedernostheide2, H.-J. Schulze2, H. Schulze3 1
Department of Mathematics and Computer Science, AG LS, University of Hagen, Germany, 2 Infineon Technologies AG, Germany, 3 Infineon Technologies Austria AG, Austria The formation and annihilation of hydrogen-related donor states were studied in protonimplanted and subsequently hydrogen plasma-treated Float-Zone (FZ) silicon wafers. H+implantations were carried out at energies of 1 or 3 MeV, applying fluences of 1⋅1014 cm-2. 13.56-MHz RF-plasma hydrogenations were done at a power of 150 W for 15 min or 1 hour, applying substrate temperatures between 350 °C and 500 °C. The samples were analyzed by depth-resolved two-point-probe spreading-resistance (SR) measurements. +
+
+
H -Implantation & 60 min H-Plasma
H -Implantation & 15 min H-Plasma 500 °C
3 MeV H -Implantation & H-Plasma (400 °C)
500 °C
6
10
10
8
10
7
10
6
10
5
15 min H-Plasma
5
10
5
5
10
400 °C 5
10
350 °C
450 °C
6
10
5
10
400 °C
6
10
5
10
350 °C
6
10
Spreading Resistance (Ohm)
450 °C
Spreading Resistance (Ohm)
Spreading Resistance (Ohm)
10
4
108 10 10
7
10
6
10
5
10
4
5
10
60 min H-Plasma
5
10 0
5
10
15
20
25
Depth (µm)
Fig. 1: SR profiles of H+-implanted (1 MeV) & H-plasma treated (15 min at 350 – 500 °C) FZ Si
0
5
10
15
20
25
Depth (µm)
Fig. 2: SR profiles of H+-implanted (1 MeV) & H-plasma treated (60 min at 350 – 500 °C) FZ Si
0
20
40
60
80
100
Depth (µm)
Fig. 3: SR profile after deep H+implantation (3 MeV) & 15- / 60min H-plasma exposure (400 °C)
H-plasma exposure for 15 min at elevated temperatures caused an enhanced donor concentration close to the projected ion range Rp (Fig. 1). The excessive donor states can be related to hydrogenated vacancy defects. Near the wafer surface, n-type doping is also enhanced, since the mobile vacancies diffuse towards the surface at elevated temperatures. However, applying higher temperatures during plasma exposure (≥ 450 °C), acceptor-like states overcompensate for the enhanced n-type doping close to the surface. After 1-hour H-plasma exposure, the measured SR values in wafer regions deeper than 2 µm are identical to the value of the untreated starting material; i. e. the excessive donor states in the wafers subsurface region down to Rp are completely annealed (Fig. 2). However, close to the surface, the n-type doping is still overcompensated for by acceptor-like defects that are tentatively attributed to hydrogenated vacancy defect clusters. Fig. 3 exhibits SR profile for samples treated by deep H+implantation and 15-min or 60-min H-plasma exposure at 400 °C. After 15-min H-plasma exposure the high SR profile between 30 and 90 µm depth reflects the implantation damage; surplus n-type doping occurs only closer to the surface, where in-diffusing hydrogen has penetrated the sample and causes the surplus n-type doping as discussed above. In wafer regions, where hydrogen has not shown up yet, the SR still follows the profile corresponding to the implantation damage. After 60-min H-plasma exposure the high resistive implantation damage zone completely disappeared; and very close to the wafer surface, the concentration of acceptor-like states is significantly higher compared to the 15-min plasma hydrogenation. Acknowledgement: Thanks a lot to Prof. Fahrner (LGBE) for providing the plasma setup. * R. Job, F. J. Niedernostheide, H. J. Schulze, H. Schulze, in: " High Purity Silicon X", Editors: C. L. Claeys, R. Falster, M. Watanabe, P. Stallhofer, ECS Transactions 16(6), 151, (2008)
Formation of Doping Profiles in Float-Zone Silicon by Helium Implantation and Plasma Hydrogenation * R. Job1, F.-J. Niedernostheide2, H.-J. Schulze2, H. Schulze3 1
Department of Mathematics and Computer Science, AG LS, University of Hagen, Germany 2 Infineon Technologies AG, Germany, 3 Infineon Technologies Austria AG, Austria
By means of two-point-probe Spreading Resistance (SR) analyses, the formation and evolution of hydrogen-related and vacancy-related donor and acceptor states were studied in helium implanted and hydrogen plasma-treated n-type Float-Zone (FZ) silicon wafers. (E = 3.75 or 11 MeV, F = 1⋅1014 cm-2). After 15-min post-implantation H-plasma exposures at temperatures between 350 and 500 °C, surplus doping profiles could be observed in the subsurface layers down to the projected ion range Rp of the treated FZ Si samples. Also acceptor-like states occurred, at least partially compensating for the initial n-type doping, so that even buried p-type layers could be created under appropriate process conditions. -2
14
Tplasma = 350 °C
Spreading Resistance (Ohm)
Spreading Resistance (Ohm)
8
10 7 10 6 10 5 10 4 10 8 10 7 10 6 10 5 10 4 10 8 10 7 10 6 10 5 10 4 10 8 10 7 10 6 10 5 10 4 10
-2
He-impl.: E = 3.75 MeV, F = 1 x 10 cm H-plasma: 60 min at various temperatures
Tplasma = 400 °C
Tplasma = 450 °C
Tplasma = 500 °C
10
7
10
6
10
5
Tplasma = 350 °C
He-impl.: E = 11 MeV, F = 1 x 10 H-plasma: 15 min, 400 °C
9
10
7
10
6
10
5
10
7
10
6
10
5
10
7
10
6
10
5
10
Tplasma = 400 °C
Spreading Resistance (Ohm)
14
He-impl.: E = 3.75 MeV, F = 1 x 10 cm H-plasma: 15 min at various temperatures
Tplasma = 450 °C
Tplasma = 500 °C
14
cm
-2
8
10
7
10
6
10
5
10
4
0
5
10
15
20
Depth (µm)
Fig. 1: SR profiles of He+-implanted and H-plasma treated (15 min at 350 – 500 °C) FZ Si
25
0
5
10
15
20
Depth (µm)
Fig. 2: SR profiles of He+-implanted and H-plasma treated (60 min at 350 – 500 °C) FZ Si
25
10
0
20
40
60
80
100
Depth (µm)
Fig. 3: SR profile after deep He+implantation (11 MeV) and 15-min H-plasma exposure at 400 °C
The results shown in Fig. 1 for 15-min H-plasma exposure can be explained as follows: Radiation damage around Rp causes a strong increase of the SR in the buried layer between a depth of 12 and 20 µm. After 15-min plasma exposure at 500 °C, two p-n junctions appeared (peaks at ∼15 and ∼18 µm depth); i. e. the region between the two peaks exhibits a p-type behavior. Hence, during 15-min post-implantation plasma hydrogenation at 500 °C, acceptorlike defect states overcompensate for the n-type doping, which might be associated with a deep acceptor level of multi-vacancy-oxygen complexes, Vn-Om, or maybe hydrogenated divacancy complexes, V2-H2; however, up to now it is not clear, if V2-H2 is stable up to 500 °C. Fig. 2 shows the SR profiles for FZ Si, where the H-plasma was applied for 60 min under otherwise the same process conditions. Weak surplus n-type doping can be observed in the local region somewhat shallower than Rp, where the implantation damage is maximal. The surplus n-type doping can be attributed to vacancy-hydrogen defect complexes acting as donor states. Fig. 3 exhibits the SR profile for a sample treated by deep He+-implantation and 15-min Hplasma exposure at 400 °C. The high resistance profile between 20 and 90 µm depth reflects the implantation damage; surplus n-type doping occurs only closer to the surface. Within 15min H-plasma exposure at 400 °C in-diffusing hydrogen penetrates the samples down to 20 µm depth and causes the surplus n-type profile as discussed above. In wafer regions deeper than 20 µm depth, where hydrogen has not shown up yet, the SR still follows the profile corresponding to the initial non-modified implantation damage. Acknowledgement: Thanks a lot to Prof. Fahrner (LGBE) for providing the plasma setup. * R. Job, F.-J. Niedernostheide, H.-J. Schulze, H. Schulze, in: "Performance and Reliability of Semiconductor Devices", Editors: M. Mastro, J. LaRoche, F. Ren, J-I. Chyi, J. Kim, MRS Symp. Proc. Series, Vol. 1108, 1108A12-03 (2009)
Chair of ELECTRONICS (Lehrgebiet Elektronische Schaltungen) Universitätsstraße 27 58084 Hagen Tel.: Fax: Email: WWW:
+49 2331 987 1163 +49 2331 987 355
[email protected] www.fernuni-hagen.de/LGES/
Staff: Phone: Email: Dr.-Ing. F. Heinrichmeyer 1166
[email protected] Dr.-Ing. Josef Eulenbrok 1701
[email protected]
On the Next Generation of Satellite Systems Dirk Schmitt and F. Heinrichmeyer In the next generation of satellite systems the code rate of the error correction codes cannot be further increased by keeping at the same Eb/No, as in the DVB-S2 standard the error correction codes are almost at the Shannon limit. Therefore the next step will be to shrink the transmission bandwidth by keeping the same symbol rate of the transmitted symbols to improve the channel capacity, given by the Shannon-Hartley theorem. To further decrease the channel bandwidth two methods are investigated. The first method is to decrease the roll-off factor of the pulse shaping filter for the transmitted symbols (so called root-raised-cosine filter). This step needs more efficient timing recovery as the known timing error detectors are very sensitive to this roll-off factor. The effect of shrinking the roll-off factor is directly coupled with a higher timing jitter in an eye plot. Investigations are made using eye plots to measure the timing jitter, depending on the roll-off factor. Also if the roll-off factor is decreased the so called peak-toaverage-power ratio (PAPR) is increasing, as due to the larger filter impulse response the superimposed pulse peaks becomes higher. It can also be seen in an eye plot where more energy is wasted in the parts which are lying beside the non-optimum sampling point. This higher PAPR forces to transmit with a smaller transmit power to get a higher outputback-off (OBO) in the receiver. If the transmitter keeps the same power for a very small roll-off factor as for a high roll-off factor the peak power pulses are driving the amplifier in the satellite (TWTA, traveling wave tube amplifier) into the non-linear region and causes nonlinear effects, like intermodulation products (IM products) or nonlinear phase distortion. This distortion is not only related to the PAPR it is also related to the frequency of these so called worst case pulses. Therefore the measurements are done on both, the PAPR and the probability of such worst –case pulses. The second method is the idea to transmit the symbols with a higher frequency than the nyquist rate, so called FTN (faster than nyquist). Faster than nyquist transmission, was first described by Mazo and Landau [1] and was further investigated under Rusek and Anderson [2]. FTN transmission leads to a higher distortion, even for Nyquist II pulses, in inter-symbol-interference (ISI) manner. The timing recovery has to be adapted to that problem and also solutions have to be found for the carrier recovery. Solutions were found for the timing recovery working at very low roll-off factors, the first method to improve the channel capacity. Also the known solutions of carrier frequency and phase recovery are investigated and improved, especially for the low-SNR cases. A further improvement focus were also put on the non-data aided carrier recovery, which works only with the data samples and does not need any preambles or pilot fields, like in the DVB-S2 standard. This fact will also improve the transmission rate due to the absence of pilot symbols. A new idea of Lottici and Luise (“Iterative Carrier Phase Synchronization for Coherent Detection of Turbo-Coded Modulations”), to implement a kind of turbo carrier recovery is currently adapted and improved to our problem. [1] J.E. Mazo, “Faster-than-Nyquist Signaling,” Bell Syst. Tech. J., vol. 54, pp. 14511462, Oct. 1975 [2] F. Rusek and J.B. Anderson. “Non binary and precoded faster-than-Nyquist Signaling,”, IEEE Transactions on Communications, May, 2008.
Chair of COMPUTER ENGINEERING Prof. Dr. Dr. Wolfgang A. Halang 58084 Hagen Tel.: +49 2331 987-372 Fax: +49 2331 987-375 E-Mail:
[email protected] http://www.fernuni-hagen.de/IT
Staff:
Ext.
E-Mail:
Dipl.-Ing. Jutta D¨ uring 4529
[email protected] Lars Gollub, MSc 4527
[email protected] Dr. Yi Zhao 4526
[email protected] Students: Shourong Lu, MSc Gloria S. Robleto, MSc Panchalee Sukjit, MSc
1731
[email protected] —
[email protected] 1153
[email protected]
Secretary: Renate Zielinski
372
[email protected]
Visitors: F.-L. Cui, China University of Mining and Technology (Jan.–May 2008) X.-Q. He, China University of Mining and Technology (Jan.–Nov. 2008) Dr. X.-M. Xiong, China Agricultural University (Jan.–Jul. 2008)
Distributed Embedded Systems R. Gumzej, M. Colnariˇc and W.A. Halang To facilitate the development of distributed embedded real-time applications in the Unified Modeling Language (UML) with consistent and temporally predictable re-configuration support, the concepts of Specification PEARL were incorporated [1] into the UML-RT profile and pattern. It enables to set up hardware/software configurations, and to specify conditions and methods for dynamic re-configuration. With the help of this Specification PEARL profile and re-configuration management pattern, diverse architectures can be designed, and many different target platforms specified on which real-time application programs, designed with UML-RT, can run predictably and with re-configuration support. Since the re-configuration management pattern also defines uniform interfaces and protocols for inter-collection/station communication, predefined port/interface definitions can easily be adapted for different communication infrastructures by extension. Since the Rational Rose RT development environment and C++ programming language were used to design the pattern and program the patterns’ native code, only an appropriate C/C++ compiler is necessary to deploy a design project for a specific platform. The portability and flexibility to the extent made possible by the mentioned design environment are inherent properties of the profile and the pattern. The latter can easily be extended with mechanisms providing an application with dependability and predictability features, by options for fault tolerance and graceful degradation, or by support for real-time operating systems in order to allow for higher degrees of flexibility and heterogeneity. HW specification
SW specification
CM Module Specification PEARL
Coherency check
Virtual machine
HW/SW architecture + Task representation Specification PEARL model Deployment
Co-simulation
Return information Final system model
Fig. 1: Elements of the Specification-PEARL methodology
The above mentioned methodology Specification PEARL is mainly a specification language derived from the constructs of the programming language PEARL for Distributed Systems according to the standard DIN 66253 Part 3 which is, in turn, an extension of the “Process and Experiment Automation Realtime Language” re-defined as PEARL90 in DIN 66253-2. In [2], its strengths, expressive power, and suitability for teaching were detailed, and the objectives for its further development, in particular inherent safety, were pointed out. Current applications developed with PEARL are documented in [3]. [1] R. Gumzej, M. Colnariˇc and W.A. Halang: A Reconfiguration Pattern for Distributed Embedded Systems, Software and Systems Modeling 8, 1, 145 – 161, 2009, ISSN 1619-1366 [2] W.A. Halang: PEARL in eingebetteten Systemen, DIN-Mitteilungen 87, 12, 25 – 27, 2008, ISSN 0722-2912 [3] W.A. Halang and P. Holleczek (Eds.): Aktuelle Anwendungen in Technik und Wirtschaft – Echtzeit 2008, Reihe “Informatik aktuell”, Berlin-Heidelberg-New York: Springer-Verlag 2009, ISBN 978-3-540-85323-7, ISSN 1431-472-X
Safety Shells for Real-time Projects R. Gumzej, M. Colnariˇc and W.A. Halang A safety shell pattern for real-time applications to be developed with UML-RT was worked out [1,2]. It was defined based on a re-configuration management pattern, inspired by the architectural specifications in the S-PEARL methodology. As main features the safety shell comprises timing and state guards as well as input/output protection and exception handling mechanisms. The pattern is parameterised by defining the properties of its components as well as by defining how software is to be mapped to hardware architectures. Initial and alternative scenarios as well as the method for switching between them were established. The goal pursued with the safety shell is to obtain clearly specified operation scenarios with well-defined transitions among them. In order to achieve safe and timely operation, the pattern provides safety shell mechanisms for application designed, i.e., enabling their predictable deterministic and temporally predictable operation.
Fig. 1: Structure of the safety shell
With the help of the safety shell features of the S-PEARL configuration management pattern introduced, distributed real-time application programs designed with UML-RT can run with safe, predictable behaviour and re-configuration support. Besides the structure of the applications, the configuration management pattern also defines uniform interfaces and protocols for intra- and inter-component/-node communication using pre-defined port/interface definitions. In (hard) real-time systems, it shall provide the necessary support for deterministic and dependable dynamic system re-configuration. The safety shell features are enabled by the pattern, but the selection and usage of the mentioned mechanisms remain the responsibility of a real-time application’s designer, since no two safety critical real-time applications are equal. [1] R. Gumzej, M. Colnariˇc and W.A. Halang: Safety Shell for Specification-PEARL oriented UML Real-time Projects, Computer Languages, Systems & Structures 2008, ISSN 1477-8424 [2] R. Gumzej and W.A. Halang: A Safety Shell for UML-RT Projects, Proc. Intl. Multiconference on Computer Science and Information Technology, pp. 629–632, Warsaw: Polish Information Processing Society 2008, ISBN 978-83-60810-14-9, ISSN 1896-7094
Inherent Safety in Computer Architectures W.A. Halang, L. Gollub and H. Stieger Real-time computers must execute instructions in a predictable way, since results arriving too late are useless or even dangerous. Owing to their features direct memory access, asynchronous interrupt service, pipelining, and caching current processors do not operate predictably. Therefore, a novel architecture [2] renounces these features, and executes in any clock cycle exactly one instruction independent of its kind and addressing mode. The instructions manipulating data have three addresses, all of which refer directly to data memory as there are no data registers. The width of the instruction words accommodates for operation code, all operand addresses and their modes, and for additional control data. The input/output is memory-mapped. Thus, the processor can be implemented by a combinational circuit. Program and data memory are separate. There are several independent paths for reading access to data memory, and exactly one for writing access. It is this multiple bus system which renders possible to implement absolute, immediate and indirect operand addressing as well as immediate and indirect result addressing without sequential processing. For each task there are separate program counters status registers enabling context switches from one instruction to the next.
Fig. 1: Simple example with two jump destination instructions
Instructions are stored sequentially at ascending addresses and usually executed one after the other. Using jumps and branches, program execution can be continued at arbitrary addresses. Conventional processors are unable, however, to detect if a jump’s target is indeed a valid jump destination as planned during program design. Such a processor just presumes that the data found at a jump destination form a correct instruction sequence, and executes it. If any design error or malfunction causes a jump to a destination not anticipated, it is impossible to detect and handle this problem adequately — a situation not acceptable for systems with high integrity requirements. This problem is solved with special instructions [1] allowing to designate valid jump targets. Jump destination instructions represent the complement of conditional and unconditional jumps including calls to and returns from subroutines. In the instruction set of a processor implementing them, each jump instruction is paired with a matching jump destination instruction. In a correct program, only jumps to addresses with matching jump destination instructions are allowed. The execution of a jump to a target address lacking a matching jump destination instruction, which indicates a deviation from the particular program’s planned control flow, raises an error condition that can be handled appropriately. ¨ [1] L. Gollub and W.A. Halang: Prozessor mit Ansprungbefehlen zur Uberwachung des Kontrollflusses, German patent application No. 10 2008 029 231.1, 2008 [2] H. Stieger: Echtzeitrechnerarchitektur mit exakt vorhersehbarer Befehlsverarbeitung, in Aktuelle Anwendungen in Technik und Wirtschaft – Echtzeit 2008, W.A. Halang and P. Holleczek (Eds.), pp. 121–130, Reihe “Informatik aktuell”, Berlin-Heidelberg-New York: Springer-Verlag 2009, ISBN 978-3-540-85323-7, ISSN 1431-472-X
Computer Architectures for Safety-Related Control ´ zek W.A. Halang, M. Skambraks and M. Snie˙ Society increasingly depends on computerised systems for automation functions in safetyrelated environments. The programmable electronic systems currently employed for safetycritical applications follow either a strictly periodical or a task-based operating policy. Combining the advantages of these approaches with regard to safety aspects, programming style, temporal behaviour, and system architecture, a novel real-time execution concept was introduced [1]. What distinguishes it from other ones is that design for simplicity was followed as leading idea. The simplicity achieved eases verification and, more importantly, lowers the cost for safety-licensing. Its main characteristics are quantisation of time into discrete execution intervals, and partitioning of tasks into execution blocks matching these intervals. Further features are implementation of a real-time operating system kernel in form of a combinational logic circuit, support of the international time standard UTC, and hardware-supported provision of error detection, forward recovery, and non-intrusive monitoring in a unified manner. The operating principle allows for task-based software execution without utilising asynchronous interrupts, eases integration into a holistic safety concept, and complies well with the safety standard IEC 61508. TDAU Task List One set of data for each task Load ACS input registers
Read ACS output registers
3-phase process for each task
ACS UTC Asynchronous event signals [] ID of task’s next execution block
RTOS algorithms implemented as combinational logic circuit Operations related to one task executed in parallel; most urgent tasks determined sequentially
[]
ID of execution block to be processed
[] : Latch for temporary storage within execution interval
Fig. 1: Function principle of the real-time operating system in hardware
For fault tolerance reasons controllers as the one mentioned above often work in redundant systems. Responsible for their safe operation are then units comparing intermediate results and actuation data. As this function cannot be implemented with redundancy, such comparators need to exhibit fail-safe behaviour. This requirement is met by a patented fast electronic comparator of two binary words [2] featuring a dual-channel, digital preprocessing unit working in parallel, which subjects its single-bit results via gates to a fail-safe galvanically separated analogue comparator. Here impulses are generated, which occur simultaneously in case of identical inputs. This causes oscillations in a resonator, which are amplified and forwarded via a transducer to a rectifier, whose output signals the comparison result. If the analogue circuit detects any differences in its digital input signals or own malfunctions, by closing the gates it enters a safe state, in which it persists. [1] W.A. Halang and M. Skambraks: Architectural Concepts for Embedded Systems in Safety-critical Applications, Keynote address, Proc. IEEE Intl. Conf. on Automation, Quality and Testing, Robotics, L. Miclea and I. Stoian (Eds.), pp. 60–65, Piscataway: IEEE 2008, ISBN 978-1-4244-2576-1 ´ zek: Elektronischer Vergleicher zweier Bin¨arworte mit ausfallsicherheits[2] W.A. Halang and M. Snie˙ gerichtetem Ausgabeverhalten, German patent No. 198 61 281, granted 2 October 2008
Concept Similarity and Ontology Mapping Y. Zhao, X. Wang and W.A. Halang To enable interoperability in the semantic web, ontologies are developed by different organisations at a large scale, also in overlapping areas. Therefore, ontology mapping has come into forth to achieve knowledge sharing, semantic integration, information retrieval, query mediation, or web service mapping in an environment where knowledge and information are represented by different underlying ontologies. The ontology mapping problem can be defined as acquiring the relationships holding between the entities of two ontologies. The method [1] to map concepts and properties between ontologies first applies syntactic analysis based on token strings, and then executes semantic analysis according to WordNet and tree-like graphs representing ontological structures. Experimental results indicate that this algorithm finds mappings with high precision. A weighted rough ontology mapping method [2] based on rough set theory and formal concept analysis first transform two source ontologies into formal contexts with linguistic processing techniques, then the two formal contexts are merged to obtain a complete concept lattice and, finally, a rough similarity measure is introduced to produce the ontology mapping results. 1 Ontology 1
Formal Context 1
0.8 Merged Concept Lattice
Ontology 2
Formal Context 2
Weighted Rough Similarity Measure Mapping result
0.6
Precision
0.4
Recall F-Measure
0.2 0 0.8
Fig. 1: Principle of the method proposed
0.85
0.95
Fig. 2: Mapping results with ontologies “Russia”
Determining concept similarity in heterogeneous ontologies is a vital problem for the semantic web. Current approaches only consider single hierarchial concept relations failing to express rich and implied information, and concept similarity under multiple relations as required in many application scenarios of semantic web services has not been investigated, yet. Basing on a method to merge heterogenous ontologies into an application ontology, in [3] a representation model for application ontologies in form of a semantic net with multiple weighted concept relations was introduced. This approach extends multiple concept relations to measure ontological concept similarity, measures concept similarity by employing concept knowledge from ontology structure and concept information content, and solves the problem of calculating similarity of compound concepts. [1] Y. Zhao and W.A. Halang: Mapping Ontologies by Utilising Their Semantic Structure, in Encyclopedia of Artificial Intelligence, J.R. Rabu˜ nal Dopico, J. Dorado de la Calle and A. Pazos Sierra (Eds.), pp. 1049 – 1055, Hershey-New York: Information Science Reference 2008, ISBN 978-1-59904-849-9 [2] Y. Zhao, X. Wang and W.A. Halang: A Rough Similarity Measure for Ontology Mapping, Proc. 3rd IEEE Intl. Conf. on Internet and Web Applications and Services, A. Mellouk, J. Bi, G. Ortiz, D.K.W. Chu and M. Popescu (Eds.), pp. 136 – 141, Los Alamitos: IEEE Computer Society Press 2008, ISBN 978-0-7695-3163-2 [3] X. Wang, Y. Zhao and W.A. Halang: Concept Similarity in Multirelational Application Ontologies, in Computational Intelligence in Decision and Control, D. Ruan, J. Montero, J. Lu, L. Mart´ınez, P. D’hondt and E.E. Kerre (Eds.), pp. 169 – 174, Singapore: World Scientific 2008, 978-981-279-946-3
Classification and Detection of Biological Warfare Agents G. Sartorius, S. Talbot and W.A. Halang The evaluation of hazard potential in threat scenarios with biological substances requires prompt and reliable recognition by emergency responders on site through the use of highly dependable transportable detection systems. The Raman technology is suitable to measure characteristic properties of bacterial samples, since it provides good prospects for the design of mobile measuring units due to its potential for size reduction. Aiming for ease of use, an adaptive system dividable into a training unit and multiple operational units was envisaged. To achieve this, principally unaltered input data need to be reduced to their essential content. This is possible by employing a wavelet transformation in combination with dimension reduction. For the latter a special smoothing method is used to refine input data sets for the application of certain dimension-reducing techniques in a way numerically stable even for discontiguous data sets, e.g., in classification tasks. The Multivariate Adaptive Embedding (MAE) method introduced acts as a universal approximator in the adaptation phase, to a very large extent without iterations and parameter adjustments. The method deduces redundancy-free models of training data sets with outstanding generalisation properties. Untrained input data, in terms of an application, can be processed in the working phase so that multivariate measurement data can efficiently, precisely, and reliably be transformed into information-conveying displays and signals. The training unit can centrally be administrated, and transmits training data to the various mobile operational units via a data link. This structure renders optimal operational efficiency and suitability for real-time use in the field.
Fig. 1: Raman spectrometer (right), displayed spectra (left), and MAE-process (PC)
The MAE method [1] enables to detect biological warfare agents, toxic agents, and other compounds employing transportable field-proven appliances, which have permanent or temporary connection with a central unit. The method is carried out in the following steps: (1) measuring characteristic features and the coumpounds’ concentration with a measuring unit, and provision of the results measured in form of parameterised curves such as, e.g., spectra, (2) training of the central unit with the available measuring data representing the different compounds to be detected, (3) establishing a model of these compounds, (4) transmission of the model into the transportable devices, and (5) acquisition and processing of new measuring data in the transportable devices in such a way that these new data can visually and gradually be associated with the classes trained. [1] G. Sartorius, S. Talbot and W.A. Halang: Verfahren und Vorrichtung zur Detektion von Kampfstoffen, German patent application No. 10 2008 054 345.4, 2008
Fault Diagnosis of Fully Variable Valve Actuators I. Sara¸c Fully Variable Valve Actuation (FVVA) systems enable to employ a wide range of combustion strategies by providing the actuation of a gas exchange valve at an arbitrary point in time, with variable lift and adjustable ramps for opening and closing. Making such a system ready for the market requires appropriate fault-diagnostic functionality. Diagnosis of fully variable valve actuators can be achieved by introducing position sensors for each actuator. This approach increases, however, the cost of the system undesirably. Thus, it is necessary to explore alternative methods by re-defining the system. Model-based diagnosis requires a known input and an observable output. Different system definitions are presented in the figure: the one according to the innermost dark grey box requires position sensors for an observable output, whereas the one relating to the outermost blue box requires the combustion process to be modeled, and the faults of the fuel actuator to be analysed. Therefore, the definition expressed by the dotted box is found to be the most suitable one for diagnostic purposes of FVVA systems [1]. m fuel
throttle
Full Variable Intake Valve Actuator n,M
Engine Demand Map
IV
IV*
EV
Gas Exchange EV* Process
RG
mair
Combustion Process
y
Full Variable Exhaust Valve Actuator
Fig. 1: Alternative system definitions for solving the diagnostics problem without position sensors are presented by differently coloured boxes; M : torque, n: engine revolution speed, IV : reference intake valve parameters, EV : reference exhaust valve parameters, IV ∗ : actual intake valve parameters, EV ∗ : actual exhaust valve parameters, χRG : residual gas fraction, mair : mass of air, mf uel : mass of fuel, θthrottle : throttle valve angle
The possible faults were generated, and their influences on critical parameters – such as emissions – defined by current on-board diagnosis standards measured. A 4-cylinder test bench engine was operated at the early intake opening strategy under different loads, and at medium-range rotational speeds on steady-state conditions. Four of the faults were found to be emission-critical and likely to occur on such systems. To distinguish between critical fault cases, the dynamic time information of the Manifold Absolute Pressure (MAP) sensor placed in the intake manifold was found to be useful. To capture this behaviour, a signal model of the MAP sensor in the time domain using Fourier coefficients for periodic signals was used. This method was able to identify actuator-specific critical fault cases of individual cylinders when they fail to open. The other two critical fault cases, viz., the late opening of the intake and the late closing of the exhaust actuator were also found to be detectable and distinguishable from one another by applying adaptive thresholds on the residual. [1] I. Sara¸c and K. Mischker: Fault Diagnosis of Fully Variable Valve Actuators on a Four Cylinder Camless Engine, in Variable Valve Optimization, SAE International 2008, ISBN 978-0-7680-2016-8
Analysis of the Peer-to-Peer Paradigm for Supply Chains P. Sukjit, C. Akinalp and H. Unger Research in various scientific fields increasingly focuses on decentralised systems of any kind – and not only file-sharing and peer-to-peer ones as before. This growing interest is caused by two effects of economic and technical globalisation, viz., (1) the information processing structures become more and more distributed all over the world, and (2) the processes of different contributors are interconnected in integrated global processes in many different and complex ways. Since it is still almost impossible to monitor, survey, or control such systems from a central authority, the peer-to-peer paradigm became a viable alternative in devising system architectures. Nevertheless, the anarchically growing neighbourhood relations in such systems motivate to look for several improvements with significant impact on system performance and resources to be allocated, e.g., network bandwidth. The main question to be answered in this context are (1) which structures may serve as foundations for improved message-routing algorithms, to manage the computational resources and to search for information, (2) which of these structures may be generated and maintained with comparably low overhead by so-called local algorithms, i.e., procedures executed locally on each participating peer with local data, only, and (3) how the respective algorithms look alike and which performance parameters they have.
Fig. 1: Structure of a typical complex supply chain
The first outcome of the research outlined above is an approach to improve the routing of messages obtained in the course of handling data read from Radio Frequency Identification (RFID) tags, with which already in 2010 80% of all merchandise is expected to be labeled. Therefore, an algorithm was proposed, which prevents the use of any centralised Electronic Product Code server, and which may use for routing purposes mesh-like structures generated on top of a fully decentralised server network. In [1] it could be shown that corresponding structures can rapidly be built, and that they allow for fault-tolerant work without any centralised instances, bottlenecks, or single points of failure. Their main advantage is that the traffic is well distributed over the whole networks. [1] P. Sukjit, C. Akinalp and H. Unger: Analysis of the P2P Paradigm in an Application of Workflow Management Systems in Business Processes for an e-Manufactory, Proc. 11th World Conf. on Integrated Design and Process Technology, pp. 406–412, Society for Design and Process Science 2008, ISSN 1090-9389
Junior Professorship of EMBEDDED SYSTEMS Prof. Dr. habil. Zhong Li 58084 Hagen Tel.: +49 2331 987-2383 Fax: +49 2331 987-375 E-Mail:
[email protected] http://www.fernuni-hagen.de/IT
Students:
Ext.
E-Mail:
Hong Li, MSc 1732
[email protected] Lin Pan, MSc 1731
[email protected] Sunantha Sodsee, MSc 1160
[email protected] Visitor: Dr. Shujun Li, Hong Kong Polytechnic University (Jan.–Jun. 2008)
A Multi-objective Genetic Algorithm based on Sexual Selection S. Sodsee, Z. Li and W.A. Halang Many real-world problems have a number of solutions, i.e., there is a set of alternatives instead of only one single solution. A single-objective optimisation problem consists of searching for the best solution at a time. In order to search for another optimal solution, objective weights need to be re-selected, and the algorithm to be re-run. By specifying different weights of objectives, another optimal solution can be reached. This can be achieved by using a Multi-Objective Genetic Algorithm (MOGA). In [2], a genetic algorithm is presented focusing on sexual selection employing the Pareto-based approach for solving multi-objective optimisation problems. It uses the concept of sexual selection with different types of sex and mutation rates based on the sex to produce off-spring.
Fig. 1: Structure of a genetic algorithm with sexual selection
The figure above shows the structure of the proposed algorithm. The population of solutions consists of the same number of male and female individuals. The temporary solutions are temporarily non-dominated ones, which can be found in the every generation of the population. Solutions are the Pareto-optimal solutions selected from the temporary solutions. The individuals in the population are chosen to reproduce a new population by using roulette wheel selection. In the new population, the numbers of male and female individuals may not be equal. For the crossover operator, the parents are chosen from the new population by using sexual selection [1]. The crossover probability is used to measure which ones should be parents. The latter consist of male and female individuals. If there is only one sex of individuals, then there is no off-spring. When generating off-spring, the parents and a position of the crossover point are chosen randomly. For the mutation operation, the mutated individuals are chosen from the new population by using the different mutation rates of male and female individuals. The population of the next generation consists of the off-spring resulting from crossover and mutation. They are sorted according to the values of an objective function (fitness values). If individuals have lower fitness values, they are rejected. Then the number of the next generation’s population remains unchanged. [1] Z. Li, H. Mo, J.B. Park, Y.H. Joo and X. Li: On the Fitness of High Order Schema of a LinearWeighted Coded Genetic Algorithm, in Computational Intelligence in Decision and Control, D. Ruan, J. Montero, J. Lu, L. Mart´ınez, P. D’hondt and E.E. Kerre (Eds.), pp. 713– 718, Singapore: World Scientific 2008, ISBN 978-981-279-946-3. [2] S. Sodsee, P. Meesad, Z. Li and W.A. Halang: A Networking Requirement Application of MultiObjective Genetic Algorithms with Sexual Selection, Proc. IEEE Intl. Conf. on Intelligent System and Knowledge Engineering, Xiamen, 2008.
EMI-Reduction with a Chaotic Peak Current-mode Boost Converter H. Li, Z.Li and W.A. Halang In recent decades, the applications of chaos in DC-DC converters mainly focus on reducing electromagnetic interference (EMI), and most of them are based on the peak current-mode controlled boost converter. The first design can be dated back to 1996 when Deane and Hamill suggested to utilise chaos to improve the electromagnetic compatibility (EMC) of power supplies. The idea was later re-formulated and improved. The literature shows that the current control technique can introduce chaos in power converters and effectively improve EMC. Despite of the success of using chaos for EMI suppression, two prominent problems remained unsolved. One is that the ripple of the output waveforms is much bigger than that with periodically running DC-DC converters, implying a degradation of performance; the other is that the background spectra may be increased, although the peak values of the power spectrum have been reduced, resulting in a larger power consumption. These two disadvantages not only exist in the peak current-mode controlled boost converter, but also in other chaotic power converters, which have seriously impeded the popularity of chaos-based DC-DC converters. Therefore, the questions of how to improve the control method for such converters so that low EMI and small output ripples can be achieved, and to verify the relationship between ripple and background spectrum constitutes were addressed.
(a) i(t) for Ilow = 0A
(b) Spect. f. Ilow = 0A
(c) i(t) for Ilow = 3.2A
(d) Spect., Ilow = 3.2A
Fig. 1: Current waveforms and corresponding spectra in chaotic modes for three different cases
In [1, 2], a novel peak current-mode boost converter was proposed. With a simple circuitry design, the proposed current control can easily adjust the magnitude of ripple and, at the same time, reduce EMI. A chaotic current mapping function corresponding to this boost converter was also derived, reporting the complex bifurcating and chaotic phenomena. The design was verified both by simulations and experiments. It turned out that not only the EMI can be suppressed, but also the ripple of the output is reduced when the boost converter works in chaotic mode. The figure shows the cases when the Boost converter operates in chaotic modes. By comparing the results shown in subfigures (a)–(d), an improvement of EMI suppression is clearly demonstrated with positive Ilow , while a large reduction of ripples is also achieved at the same time. It is also noticeable that there is a shift of the dominant frequencies in the power spectrum when Ilow is increased. Further studies are to derive the relationship between them, and to identify the factors influencing the energy distribution in the proposed chaotic power converter. [1] H. Li, Z. Li, W.K.S. Tang and W.A. Halang: A Chaotic Peak Current-mode Boost Converter for EMI Reduction and Ripple Suppression, IEEE Trans. on Circuits and Systems II, 55, 8, 763–767, 2008, ISSN 1549-7747 [2] H. Li, Z. Li, J.G. Barajas Ram´ırez and W.A. Halang: Controlling EMI with a Chaotic Peak Currentmode Boost Converter, Proc. First Intl. Workshop on Nonlinear Dynamics and Synchronization, K. Kyamakya (Ed.), pp. 118–124, Aachen: Shaker Verlag 2008, ISBN 978-3-8322-7225-8, ISSN 18667791
A Chaotic Soft-switching PWM Boost Converter H. Li, Z. Li and W.A. Halang
40
40
40
20
20
20
0
0
-20 -40
0 Amplitude
Amplitude
Amplitude
The electromagnetic interference (EMI), resulting from the high rates of changes of voltage and current, has long become the major design criterion in power electronics due to extensive applications of various electronic devices in industry and daily life. The question of how to reduce the annoying and harmful EMI has attracted the interest of many researchers. Normally, it is handled by properly-tuned filters to reduce the conducted interference in specific frequency bands. Its design is, however, not easy under the stringent requirements of the EMI standard, and is also dependent on the operating conditions expected of the converters. Besides, the additional circuits of the filter not only increases the cost, but also implies an increase of size and weight, making the product lack of portability. More recently, the EMI problem is tackled employing spectral spreading by chaotic modulation. Chaotic modulation cannot, however, directly reduce the rapid change rate of voltage and current. Therefore, another more popular and practical technique, i.e., soft switching, has been proposed, which is to switch off or on at zero current or zero voltage to alleviate the high rates of changes of voltage and current so that EMI can be reduced. There, the switching loss is reduced, which implies that the energy loss is also reduced, resulting in an increase of efficiency.
-20 -40
-20 -40
-60
-60
-80
-80
-80
-100
-100
-100
0 1 Frame: 13
2
3
4 5 6 Frequency (MHz)
7
8
9
10
(a) PSD of inductor current with hard switching PWM
-60
0 1 Frame: 13
2
3
4 5 6 Frequency (MHz)
7
8
9
10
(b) PSD of inductor current with soft switching PWM
0 1 Frame: 13
2
3
4 5 6 Frequency (MHz)
7
8
9
10
(c) PSD of inductor current with chaotic soft switching PWM at β=0.2
Fig. 1: PSD of inductor currents based on different control methods
The research [1] is concerned with combining chaotic modulation with soft switching in order not to spread the energy distribution over the whole frequency band (thus reducing the spectrum’s peaks), only, but also to reduce switching loss or energy loss, such that EMI can be highly reduced and the efficiency can be improved. Therefore, a novel approach combining soft switching with chaotic modulation was proposed for EMI reduction in boost converters. In order to provide the required chaotic signals, a practical method for the generation of chaotic carriers through feedback control was also proposed. This chaotic soft-switching pulse-width modulation (PWM) is applied in a boost converter. The simulation results clearly demonstrate that EMI and efficiency can be improved by this method as compared with hard-switching PWM and conventional soft-switching PWM control as shown in the figure. Furthermore, the new control method can easily be implemented and is suitable for different kinds of DC-DC converters. [1] H. Li, Z. Li, W.K.S. Tang and W.A. Halang: A Chaotic Soft Switching PWM Boost Converter for EMI Reduction, Proc. IEEE Intl. Symp. on Industrial Electronics, pp. 341–346, Piscataway: IEEE 2008, ISBN 978-1-4244-1666-0
Dynamics of a Chaotic Attractor L. Pan, W. Zhou, Y. Xu and H. Lu Since Lorenz found the first chaotic attractor in a smooth three-dimensional autonomous system, there was considerable research work addressing the search for other chaotic attractors. In 1976, R¨osler found a three-dimensional autonomous and smooth chaotic system. Later, more and more attractors were found, up to the recently named systems of Chen, L¨ u, and Liu, the generalised Lorenz system family, and the hyperbolic type of the generalised Lorenz canonical form.
Fig. 1: Butterfly-shaped attractor of a new chaotic system
In [1] a new chaotic system is discussed, whose process is similar to the one of the L¨ u attractor. It is a three-dimensional autonomous system of the form, x˙ = a(y − x)
y˙ = bx − xz − y − u , z˙ = xy + cz
where a, b, c are constants, currently not in the range of chaos, and u is a linear feedback controller in the form u = l1 x + l2 y + l3 z, in which l1 , l2 , l3 are constant gains to be determined. Letting u = y results in the following new system, x˙ = a(y − x)
y˙ = bx − xz z˙ = xy + cz
,
which is chaotic when a = 10, b = 16, and c = 0.1. According to detailed numerical simulation as well as theoretical analysis, the chaotic attractor obtained from this new system is also butterfly-shaped as shown in the figure. Furthermore, some basic properties of the system were also investigated, such as symmetry and invariance, dissipativity, the attractor’s existence, Lyapunov exponent, as well as its forming mechanism. It was found that the butterfly attractor has a compound structure obtained by merging two simple attractors after performing a mirror operation. Although this new attractor is similar to the well-known Lorenz chaotic attractor, topologically it is not equivalent to the latter, since the eigenvalues of the corresponding Jacobians differ for the new system from those of the Lorenz system. [1] L. Pan, W. Zhou, Y. Xu and H. Lu: On Dynamics Analysis of a New Chaotic Attractor, Physics Letters A, 372, 36, 5773–5777, 2008
Security of Chaos-based Image Encryption Schemes ´ S. Li, D. Arroyo, C. Li, R. Rhouma, G. Alvarez, V. Fernandez, G. Chen and K.-T. Lo In this project, the security of two chaos-based image encryption schemes was re-evaluated. One of the schemes is based on the use of a chaotic map lattice. The other one is called RCES or RSES, and implemented by combining pixel value masking and pixel swapping under the control of a chaos-based pseudo-random bit sequence. In both schemes, the chaotic map employed with the control parameter µ is the logistic map xn+1 = µxn (1 − xn ). For the image encryption scheme based on a chaotic map lattice, it was found [1] that the cryptosystem is unable to recover the plain images with an acceptable quality due to the wobbling precision problem of floating-point arithmetic. Two attacks were proposed to break part of the secret key, which include the control parameter of the logistic map, the number of iterations, and the number of encryption cycles. The latter two subkeys are broken with a timing attack, since they are linearly related with the encryption speed. Some potential enhancements to the original design were also given.
a) A recovered plain image of size b) Another recovered plain image 256 × 256 of size 768 × 768 Fig. 1: Two breaking results of the combined known plain-text attack
For the RCES/RSES image encryption scheme, it was found [2] that it scheme is very weak against known/chosen plain-text attacks. Only one or two known/chosen plain images are enough to obtain very good breaking results. Two different known plain-text attacks were proposed. The first attack can only reconstruct an equivalent M × N masking image for decrypting the first M N pixels of any cipher images. The second attack can successfully recover the control parameter and a chaotic state xk of the chaotic logistic map, which can be used as an equivalence of the key to exactly decrypt all pixels except for the first k ones. Combining the two known plain-text attacks can yield a complete breaking of the RCES/RSES image encryption scheme. Two breaking results of the combined attack are shown in Fig. 1. Furthermore, all proposed known plain-text attacks can be modified to be chosen plain-text ones with better performance. Some lessons and design principles have also been drawn from this cryptanalytic work. ´ [1] S. Li, D. Arroyo, R. Rhouma, G. Alvarez and V. Fernandez: On the Security of a New Image Encryption Scheme based on Chaotic Map Lattices, Chaos, 18, 3, art. no. 033112, 2008 [2] S. Li, C. Li, G. Chen and K.-T. Lo: Cryptanalysis of RCES/RSES Image Encryption Scheme, Journal of Systems and Software, 81, 7, 1130–1143, 2008
Cryptanalysis of Digital Chaotic Ciphers ´ S. Li, C. Li, D. Arroyo, G. Alvarez, G. Chen, K.-T. Lo and J. Nu˜ nez In this project, the security of two digital chaotic ciphers was studied. One of them is based on the two-dimensional H´enon map, and the plain text modulates the map’s control parameters for the generation of the cipher text. The second one is based on four different one-dimensional chaotic maps and two dynamic tables, which generate a pseudo-random number sequence to mask the plain text. The research [1] showed that neither of the two digital chaotic ciphers is sufficiently secure. For the first chaotic cipher, some problems with the implementation were also distinguished. It was found that the H´enon map does not behave chaotically with many plain texts, and there exist decryption errors due to the finite precision computations involved in the decryption process. In addition, two parts of the secret key are related to each other, which leads to a partial key recovery attack that can reconstruct a secret function and the initial condition y0 of the H´enon map (both of which are part of the key). An example of the attack is shown in Fig. 1. A claim about a certain data compression effect of the digital chaotic cipher was also found not to be true. 1.4
1.33 Real function
Real function
1.32
Guessed function
1.35
Guessed function
µ1
µ
1
1.31 1.3
1.3 1.29
1.25 1.28 1.2 −1
−0.5
0 y
0.5
1.27
1
0
0.1
0.2
0.3
0.4
0.5
0.4
0.5
y
1.4
1.33 Real function
Real function 1.32
Guessed function
Guessed function
1.35
µ1
µ
1
1.31 1.3
1.3 1.29
1.25 1.28 1.2 −1
−0.5
0 y
0.5
1
1.27
0
0.1
0.2
0.3 y
Fig. 1: Breaking a secret function µ1 (y)
For the second chaotic cipher, some weak keys were found and a known plain-text attack was proposed [2]. The computational complexity of the known plain-text attack is not greater than O(221 ), and only 120 plain bytes are needed for a successful attack. Furthermore, there exists one more problem with the chaotic cipher: the second dynamic table DT2 is generated by a linear congruential pseudo-random number generator, which turned out, however, to be incapable of generating good pseudo-random numbers for the purpose of encryption. It was also noticed that all these security problems exist for an improved edition of this digital chaotic cipher as well. ´ [1] S. Li, D. Arroyo, G. Alvarez, C. Li and J. Nu˜ nez: Cryptanalysis of a Discrete-time Synchronous Chaotic Encryption System, Physics Letters A, 372, 7, 1034–1039, 2008 ´ [2] S. Li, C. Li, G. Alvarez, G. Chen and K.-T. Lo: Cryptanalysis of a Chaotic Block Cipher with External Key and its Improved Version, Chaos, Solitons & Fractals, 37, 1, 299–307, 2008
Cryptanalysis of Two-channel Chaos-based Secure Communication Schemes ´ S. Li, A.B. Orue, V. Fernandez, G. Alvarez, G. Pastor, M. Romera and F. Montoya In this project, the problem of how to estimate the parameters of the Lorenz system was studied by exploring the geometrical properties of its chaotic attractor. It was found [1] that the co-ordinates of the two equilibriumqpoints xC± = ± β(ρ − 1) can approximately be obtained by observing the waveform of x(t). This leads to a reduced space in which the two parameters β and ρ are searched for. The above result was applied to the cryptanalysis of the two-channel chaos-based secure communication scheme proposed by Jiang in 2002. Jiang’s system uses one channel to transmit a driving signal x(t) for the synchronisation between the master and the slave chaotic systems, and makes use of a second channel to send the cipher text to the receiver. Since x(t) is observable, the parameter estimation method can be used to obtain the values of β and ρ. The cipher text signal is filtered to estimate the jamming noise added to the plain text signal. The average power of the noise signal serves as an indicator to obtain a more precise estimation of the values of β and ρ. Experiments showed that the two parameters can be adjusted separately and, thus, a very fast iterative process can be designed to search for β and ρ. Fig. 1 shows an experimental result on the relationship between the noise power and the parameters to be estimated.
−10 xC ± − 4%
−20
xC ± + 2%
Error power (db)
−30
@ I @ @
−40 xC ± + 0.5%
−50
@ x ± − 1% C
@ I @ @ x ± − 0.25% C
xC ± + 0.12%
−60 −70 x∗C ± = xC ±
−80 22
24
26
28
30
32
ρ∗
Fig. 1: Logarithmic representation of the recovered noise power’s mean ε2 for several values of x∗C ±
With the recovered parameters of the Lorenz system one can easily reconstruct the dynamics of the master system, and then decrypt the cipher text signal to obtain the plain text. Experiments verified that such an attack works well for Jiang’s two-channel chaos-based secure communications scheme. ´ [1] S. Li, A.B. Orue, V. Fernandez, G. Alvarez, G. Pastor, M. Romera and F. Montoya: Determination of the Parameters for a Lorenz System and Application to Break the Security of Two-channel Chaotic Cryptosystems, Physics Letters A, 372, 34, 5588–5592, 2008
Cryptanalysis of Multimedia Encryption Schemes S. Li, C. Li, Z. Li, W.A. Halang, G. Chen, N.G. Bourbakis and K.-T. Lo In this project, a general quantitative cryptanalysis of permutation-only multimedia ciphers and an cryptanalysis of two ad hoc multimedia encryption schemes was worked out. In addition, a brief survey on multimedia encryption was contributed to an encyclopedia [1], pointing out some challenges and new research directions. The general cryptanalysis of permutation-only multimedia ciphers [2] was carried out on digital images, and the results can easily be generalised to other types of multimedia data. The main findings of this cryptanalysis are: (1) all permutation-only multimedia ciphers are practically insecure against known/chosen plain-text attacks in the sense that only O(logL (M N )) known/chosen plain-texts are sufficient to recover – on average – not less than half of the plain text’s elements, and (2) the computational complexity of the known/chosen plaintext attack is only O(n · (M N )2 ), where n is the number of known/chosen plain texts used. It was also found that the redundancies existing in natural images help to enhance the performance of the cryptanalysis.
Fig. 1: A result of the differential cipher-text-only attack to BSS-based multimedia encryption
The two ad hoc multimedia encryption schemes studied [3,4] are based on blind source separation (BSS) and two-dimensional discrete prolate spheroidal sequences (DPSS), respectively. Both schemes use a matrix as part of the secret key, which is used to transform the plain text into the cipher text. It was found that both schemes are not sufficiently secure against all kinds of common cryptographic attacks, which include cipher-text-only attack, known/chosen plain-text attack, and chosen cipher-text attack. For the BSS-based scheme, a specific differential cipher-text-only attack is able to obtain a mixed view of the two plain-texts involved as shown in Fig. 1. For the DPSS-based scheme, an interesting feature called “lossy perceptual decryption” was noticed, which is related to the fact that the encryption/decryption processes are not sensitive enough to mismatches of the input. [1] S. Li, Z. Li and W.A. Halang: Multimedia Encryption: A Brief Survey, in Encyclopedia of Multimedia Technology and Networking, 2nd ed., M. Pagani (Ed.), Vol. II, 972–977, Hershey-New York: Information Science Reference 2008, ISBN 978-1-60566-014-1 [2] S. Li, C. Li, G. Chen, N.G. Bourbakis and K.-T. Lo: A General Quantitative Cryptanalysis of Permutation-only Multimedia Ciphers Against Plaintext Attacks, Signal Processing: Image Communication, 23, 3, 212–223, 2008 [3] S. Li, C. Li, K.-T. Lo and G. Chen: Cryptanalysis of an Encryption Scheme Based on Blind Source Separation, IEEE Trans. on Circuits and Systems I: Regular Papers, 55, 4, 1055–1063, 2008 [4] S. Li, C. Li, K.-T. Lo and G. Chen: Cryptanalysis of an Image Scrambling Scheme without Bandwidth Expansion, IEEE Trans. on Circuits and Systems for Video Technology, 18, 3, 338–349, 2008
Chair of Communication Networks (Kommunikationsnetze) Prof. Dr.-Ing. Herwig Unger Universit¨atsstrasse 27 58084 Hagen Tel.: +49 2331 987-1155 Fax: +49 2331 987-353 E-Mail:
[email protected] http://www.fernuni-hagen.de/kn/
Academic Staff:
Ext.
Dipl.-Inf. Daniel Berg Dipl.-Inf. Hauke Coltzau Lada-On Lertsuwanakul, M.Sc. Sunantha Sodsee, M.Sc.
1154
[email protected] 1142
[email protected] 1123
[email protected] 1160
[email protected]
Technical staff:
Ext.
Dipl.-Ing.(FH) Werner Schubert 1149
E-Mail:
E-Mail:
[email protected]
Secretary: Jessica Gabski
1141
[email protected]
Crisis Response on Infectious Diseases - Computational Epidemiology Hauke Coltzau In 2008, the chair of communication networks has continued its work on computational epidemiology. The focus has been laid on the question, in how far current and new modelscan be integrated into crisis-response-management facilities. As a first step, the current individual-based simulation model for the spreading of influenza-like diseases in case of eithera local epidemic or even a global pandemic, which was developed at the chair of communication networks, was implemented using the P2PNetSim distributed simulation tool. It could clearly be pointed out that individual-based simulations considering the underlying contact networks can and should be used for estimating the impact of an epidemic. As a second step, the chair of communication networks has participated in a crisismanagement exercise, which has been run by the administrative district ‘M¨arkischer Kreis’ in L¨ udenscheid. The background of the exercise was to find out, how the district’s crisisresponse staff was able to manage the formerly unknown situation of an influenza pandemic. Using the individual-based epidemic simulation, we were able to provide estimations on the influence of the staff’s decisions on the future course of the epidemic. Additionally, we were able to estimate the amount of influenza-cases in critical parts of the population, especially medical personnel, firemen and Red-Cross personnel. The exercise clearly showed that there is an immense need for a tool that allows a crisismanagement staff to estimate the impact of their decisions on the population and so be able to balance different alternatives. In 2009, the chair of communication networks will continue its work into that direction.
P2PNetSim Daniel Berg P2PNetSim is a distributed simulation tool for modelling and analysing large networks. Its development started in 2007 in the Chair of Communication Networks. P2PNetSim is the only tool available that can handle up to 1 million peers. Its design is extremly flexible and allows to simulate not only any kind of computer networks but also social networks, rfid-processing, and even spreadings of infectual diseases. A simple Java-based API makes it possible to implement aribtrary individual or collective peer behaviour including interaction with the user or with real existing networks. An open hierarchical message passing and -processing infrastructure allow a distributed analysis process, that can include whatever is programmable in Java.
In 2008 research topics like Computational Epidemiology, and Ant-based routing algorithms went on, and new topics like Pattern-Formation came up. All of them still get essential benefits from P2PNetSim’s power. The intensive work with this tool also led to more experiences and new ideas related to the tool itself. So we also worked on a lot of improvements for P2PNetSim. A new user interface, a small-world generator, more stability, yet easier use, a detailed user manual and a lot more features are the results.
Since 2008 P2PNetSim is also in use at King Mongkut’s University in North Bangkok. In the occasion of the partnership between our faculties, a team member visited King Mongkut’s University in June/July to give a P2PNetSim workshop.
Multi-criteria Routing Lada-On Lertsuwanakul Communication over the Internet becomes more and more complex and chaotic. Missing or insufficient Quality-of-Service (QoS) facilities in those networks lead to many problems such as packet loss and high latency due to overloaded routers, high overhead caused by adaptive routing data exchange and by big routing tables. Routing performance is one of the most important criteria to evaluate network schemes. The analysis of routing performance is based in a set of metrics which include packet delay, thourghput, available bandwidth, and packet loss, but resource utilization would provide information for more efficient In this research we consider criteria, like shortest path, available bandwidth and buffer-fill level. Existing routing algorithms use several techniques like flooding, ant- or DHT-bases approaches, or genetic algorithms. In this article, we bring these algorithms together, to find optimal routes. The Colored Ant is an algorithm that proposes self-balanced and self-adaptive paths using ant pheromone in different colors [3]. Routing decisions are made from pheromone color, and pheromone concentration. The Thermal Field is an approach derived from the physical analysis of development of heat, and adapted to search-problems within p2p-communities [2]. The method aims to improve the information management and organization in p2p-communities. In this research, the temperature corresponds to the intensity of the activities of nodes in the network. The Euclidean Space is used for finding the shortest distance between a sourceand a destination-node with a coordinated network.
The last part describes the process that derives a routing-decision, based on the information, given by the different routing-approaches, mentioned above. The Neural Networks and Fuzzy rules are used to determine optimal decision and to implement learning processes that react to network-changes in real-time. [1] E.K. Lua, J. Crowcroft, M. Pias, R. Sharma, and S. Lim, ‘A Survey and Comparison of Peer-to-Peer Overlay Network Schemes’, in IEEE Communications Survey and Tutorial, March 2004. [2] H. Unger and M. Wulff, ‘Search in Communities: An Approach Derived from the Physic Analogue of Thermal Fields’, Proceedings der ISSADS 2004, Guadalajara, Mexico, Lecture Notes in Computer Science (LNCS) 3061,Guadalajara, Mexico, 2004. [3] M. A. Rojas, H. Unger, and H. Coltzau, ‘Self-Balanced and Self-Adaptive Routes in Unstructured P2P Networks’, Second International Conference on Systems (ICONS’07), IEEE Computer Society, Martinique, France, 2007.
A Multi-objective Genetic Algorithm based on Sexual Selection S. Sodsee, Z. Li and W.A. Halang Many real-world problems have a number of solutions, i.e., there is a set of alternatives instead of only one single solution. A single-objective optimisation problem consists of searching for the best solution at a time. In order to search for another optimal solution, objective weights need to be re-selected, and the algorithm to be re-run. By specifying different weights of objectives, another optimal solution can be reached. This can be achieved by using a Multi-Objective Genetic Algorithm (MOGA). In [2], a genetic algorithm is presented focusing on sexual selection employing the Pareto-based approach for solving multi-objective optimisation problems. It uses the concept of sexual selection with different types of sex and mutation rates based on the sex to produce off-spring.
The figure above shows the structure of the proposed algorithm. The population of solutions consists of the same number of male and female individuals. The temporary solutions are temporarily non-dominated ones, which can be found in the every generation of the population. Solutions are the Pareto-optimal solutions selected from the temporary solutions. The individuals in the population are chosen to reproduce a new population by using roulette wheel selection. In the new population, the numbers of male and female individuals may not be equal. For the crossover operator, the parents are chosen from the new population by using sexual selection [1]. The crossover probability is used to measure which ones should be parents. The latter consist of male and female individuals. If there is only one sex of individuals, then there is no off-spring. When generating off-spring, the parents and a position of the crossover point are chosen randomly. For the mutation operation, the mutated individuals are chosen from the new population by using the different mutation rates of male and female individuals. The population of the next generation consists of the off-spring resulting from crossover and mutation. They are sorted according to the values of an objective function (fitness values). If individuals have lower fitness values, they are rejected. Then the number of the next generation’s population remains unchanged. [1] Z. Li, H. Mo, J.B. Park, Y.H. Joo and X. Li: On the Fitness of High Order Schema of a LinearWeighted Coded Genetic Algorithm, in Computational Intelligence in Decision and Control, D. Ruan, J. Montero, J. Lu, L. Martnez, P. D’hondt and E.E. Kerre (Eds.), pp. 713– 718, Singapore: World Scientific 2008, ISBN 978-981-279-946-3. [2] S. Sodsee, P. Meesad, Z. Li and W.A. Halang: A Networking Requirement Application of MultiObjective Genetic Algorithms with Sexual Selection, Proc. IEEE Intl. Conf. on Intelligent System and Knowledge Engineering, Xiamen, 2008.
FOCUS GROUP on PHOTONICS Universit¨atsstr. 27/PRG, D-58084 Hagen, Germany Fundamentals and Theory of Electromagnetics (Allgemeine und Theoretische Elektrotechnik) Prof. Dr.-Ing. Reinhold Pregla -
[email protected] Phone: +49-2331-987-1140, FAX: -353 Staff Wolfgang Benner PD Dr.-Ing. Stefan Helfert (until 04/2008)
Phone 1149 1144
Email
[email protected] [email protected]
Optical Information Technology (Lehrgebiet Optische Nachrichtentechnik) Prof. Dr. J¨ urgen Jahns -
[email protected] Phone: +49-2331-987-340, FAX: -352, Email:
[email protected] Staff Michael Bohling (until 03/2008) Dr. Qing Cao (until 07/2008) Tina Heldt PD Dr.-Ing. Stefan Helfert (since 04/2008) Hans-W. Knuppertz Thomas Seiler Students Xenia Br¨ uhn Ulrich Lohmann Secretary Veronika Kral
Phone 1195 1133 1121 1144
Email
[email protected] [email protected] [email protected] [email protected]
1128 1122
[email protected] [email protected]
341
[email protected] [email protected]
1120
[email protected]
Optical Microsystems (Juniorprofessur Optische Mikrosysteme) Jun.-Prof. Dr. Matthias Gruber -
[email protected] Phone: +49-2331-987-1131, FAX: -352 Staff Michael Bohling (since 04/2008) Students Christian Bonerz Richard Heming Udo Vieth ∗
Phone 1195
Email
[email protected]
Prof. Dr.-Ing. Wilfrid Pascher, formerly Electromagnetic Field Theory, is now Professor f¨ ur Theoretische Elektrotechnik, Universit¨ at der Bundeswehr M¨ unchen.
Analysis of complex periodic structures Reinhold Pregla Periodic structures play an important rule in designing microwave and optical waveguide circuits. Efficient algorithms based on the Method of Lines were developed in the past for normal periodic structures like fibre gratings [1]-[3]. The Floquet’s modes were obtained in a special way from half of the period by calculating the short and open circuit impedance matrices. In this report will be shown how complex periodic structures can be analyzed. Examples are ring resonators that are periodic and coupled with each other or with straight waveguides (see Figures 1 and 2). A single periodic ring resonator (propagation in φ-direction) can be analyzed analog to straight periodic structures. In case of the complex structures the structures must be divided in homogeneous and inhomogeneous parts (see especially Figure 1). The MoL will be combined with finite differences in conjunction with impedance/admittance transformation[4]. Results for the field distribution in the ring part II and the scattering parameters |Sik |2 for the structure in Figure 1 – analyzed with the algorithms described – are shown in Figure 3a,b [5]. S2 ABC
R3 z x
φ
r
III
y VI Δφ φ r
R4 Ro
S1
φA
z
II φB
φ
r
z x
x
S2
I
S1
P1
P3
P2
P4
S2
WR
S1
R2
W
Fig. 1: Ring resonator coupled by straight
R
Wg
S2
(a)
S1
φA R1
waveguides with segmentation into four parts I to IV. Fig. 2: Coupled ring resonators. Coupled region with ports P1 to P4
(b)
Fig. 3: (a) Field distribution in the homogeneous part II of the resonator in Fig. 1
(b) Scattering parameters between the ports as function of the normalized wavelength. [1] R. Pregla, “Modeling of Optical Waveguide Structures with General Anisotropy in Arbitrary Orthog[2] [3] [4] [5]
onal Coordinate Systems”, IEEE J. of Sel. Topics in Quantum Electronics, Vol. 8, No. 6. ¨ R. Pregla, “Efficient Modeling of Periodic Structures”, Int. J. of Electronics and Comm. (AEU), vol. 57 (2003) No. 3, pp. 185–189. R. Pregla, “Analysis of gratings with symmetrical and unsymmetrical periods”, 6th Int. Conf. on Transparent Optical Networks, Wrozlaw, Poland, July 2004. R. Pregla, “Modeling of optical waveguides and devices by combination of the method of lines and finite differences of second order accuracy”, Optical and Quantum Electronics, Vol. 38, pp. 3-17. R. Pregla, “Analysis of complex periodic structures”, 6th Int. Conf. on Transparent Optical Networks, Athens, Greece, June 2008.
Comparison of the eigenmodes in dielectric and metallic waveguides S. Helfert, J. Jahns Due to the need of miniaturizing photonic waveguides metallic structures will play an important role in the future. Therefore, we are interested in plasmon waves that propagate in these structures. In dielectric waveguides the fields are concentrated inside a dielectric due to the refractive index contrast. In contrast, the plasmon waves are guided at the surface between the dielectric and metallic media. Here we examined the eigenmodes of these different materials to see the differences and common behavior. Figure 1 shows the metallic and dielectric structures that we used for comparison. The magnetic field distribution of the fundamental mode is shown in figure 2. The above mentioned guidance mechanism can be clearly observed, where we have the maximum of the metallic waveguide at the interface, whereas the field is concentrated inside the higher refractive index media on the right. On the other hand, the curves in figure 3 look very similar. Here we present the fields in planes where they are maximum (i.e. on the surface for the metallic structure and in the middle of the of the film on the right). The latter behavior shall be utilized to design circuits, analog to those that were designed for dielectric waveguides.
w
εm
εd1 t
εS
w εF
εd2
h
Fig. 1: Examined waveguide structures; metallic (right) dielectric (left)
0.5
0.6
0.4 0.3
0.4
0.2
0.2 y [µ m]
y [µ m]
0.1 0
0
−0.1
−0.2 −0.2
−0.4
−0.3 −0.4
−0.6 −0.5 −1.5
−1
−0.5
0
0.5
1
1.5
2
−1.5
x [µ m]
−1
−0.5
0 x [µ m]
0.5
1
1.5
2
1
1
0.9
0.9
0.8
0.8
0.7
0.7
0.6
0.6
|Hx| a.u
|Hx| a.u
Fig. 2: Magnetic field distribution of the even–mode, metallic–structure (left), dielectric–waveguide (right)
0.5
0.5
0.4
0.4
0.3
0.3
0.2
0.2
0.1 0 −2
0.1 −1.5
−1
−0.5
0 x [µ m]
0.5
1
1.5
2
0 −2
−1.5
−1
−0.5
0 x [µ m]
0.5
1
1.5
2
Fig. 3: Magnetic field distribution of various eigenmodes; metallic–structure (left), dielectric–waveguide (right)
[1] P. Berini, Plasmon–polariton waves guided by thin lossy metal films of finite width: Bound modes of symmetric structures, Phys. Rev. E, 61 (2000) 10484–10503.
Examining the electromagnetic fields in the the corners of dielectric waveguides S. Helfert It is well known that sharp corners in dielectric waveguides or metal structures can give rise to poles of the electromagnetic fields. These poles may lead to inaccuracies in numerical algorithms that are used to model such devices. The behavior of the fields in the the static case is well known, and analytic expressions can be given, (see e.g. [1], [2]). The situation is completely different in waveguiding structures, i.e. if there is a time varying field. Studies for the non-static case were e.g. done in [3], [4]. However, an exact solution has not been derived so far, which was the motivation for our research. In particular, we determined the eigenmodes of a rib waveguide (see figure 1a – also for the coordinate system) and studied the fields in planes through the corners. The goal was twofold: a) find ways to describe the fields in the vicinity of the corner as precisely as possible, b) introduce this description into numerical algorithms to improve their performance. Figures 1b and 2 show some of the computed field components in a plane through the corners of the waveguide (i.e. in a plane through the points 1 and 2 in figures 1a). Some interesting features can be observed. The amplitude of the electric field is much larger in point 2 than in 1 (see Fig. 2a) and indicates the appearance of a pole here. This is in clear contradiction to the situation in the static case, where Ex would have a pole in point 1, but was zero in point 2. On the other hand, the curves indicate a maximum value (pole) only in the air side in point 1, whereas the field seems to remain finite in the rip-side. Also the magnetic field shows some unexpected behavior, where we observe sharp maxima in the corners. These findings show that the static approach (i.e. assume a similar field distribution in the corners as in the static case), does not lead to suitable results for waveguide problems. 0.07
x
0.06
w
t
H
nF
=
zt
2
z
=
zb
0.04
x
z
z
|H |
0.05
1
0.03 0.02
nS
0.01
a)
b)
0 −3
−2
−1
z
0
1
2
Fig. 1: a) Rib waveguide and b) magnetic field distribution in a plane through the corners 0.09 0.01
0.08 0.07
0.008 0.05
|Ez|
|Ez|
0.06
0.04
0.006
0.004
0.03 0.02
0.002
0.01
a)
0 0
0.2
0.4
z
0.6
0.8
1
b)
0 0.8
0.85
0.9 z
0.95
1
Fig. 2: a) Electric field distribution in a plane through the corner of the rib waveguide figure 1a, b) the field distribution through point 1 is enlarged [1] R.E. Collin, Field theory of guided waves, series of electromagnetic waves, IEEE press, 2nd edition (1991) 26. [2] J. Meixner, The behavior of electromagnetic fields at edges, IEEE Trans. Antennas. Propagation, AP-20 (1972) 442–446. [3] J. Bach Andersen and V.V. Solodukhov, Field behavior near a dielectric wedge, IEEE Trans. Antennas. Propagation, AP-26 (1978) 598–602. [4] G.R. Hadley, High-accuracy finite-difference equations for dielectric waveguide analysis II: dielectric corners, J. Lightwave Technol., 20 (2002) 1219–1231.
Three-dimensional analysis of hexagonal structured photonic crystals using oblique coordinates A. Edelmann, S. Helfert The elementary cells of hexagonal structured photonic crystals are described by parallelograms. Figure 1 shows such type of photonic crystal and the elementary cells to calculate the various bands. Here the calculations were done with the Method of Lines (MoL) where oblique coordinates were introduced to describe the elementary cells. The theoretical background and results for the two-dimensional case can found in [1] and [2].
Fig. 1: Hexagonal structured photonic crystal with the elementary cells for the bandstructure calculations
For real-word problems three-dimensional vectorial calculations have to be performed. Results for this case are shown in figure 2. The underlying photonic crystal is designed as array of hexagonal holes in a substrate and can found in [3]. In figure 2 we can see a) the odd- and b) the even-Modes of the photonic crystal. The calculations show a good agreement up to certain frequencies. For higher frequencies the approximations must be done more precisly.
Fig. 2: Calculated Γ-X Band of a) odd-Modes and b) even-Modes (MoL) and the literature values from [3] (Villeneuve) as reference
[1] S.F. Helfert, Introducing Oblique Coordinates in Numerical Methods, Applied to the Computation of Band Structures, ICTON Conf., 2, 80–83, Rome, Italy, 2007. [2] S. Helfert, Applying oblique coordinates to the method of lines, PIERS, 61 (2006) 271–278. [3] P.R. Villeneuve, S. Fan, S.G. Johnson, J.D. Joannopoulos, Three-dimensional photon confinement in photonic chrystals of low dimensional periodicity, IEE Proc.-Optoelectron., 145 (1998) 384–390.
Three-dimensional analysis of waveguide structures using oblique coordinates A. Edelmann, S. Helfert Oblique coordinates are interesting for analyzing tilted structures. The structure can immediately be described with such oblique coordinates, in contrast to cartesian coordinates which require e.g. a staircase approximation. An example is given in [1] where a tilted connection between two straights waveguides was studied. Further examples using the oblique coordinate system can found in [2] and [3]. Figure 1a shows a three-dimensional waveguide structure. In our examination, this waveguide was analyzed with oblique coordinates.
Fig. 1: a) Three-dimensional waveguide structure under study and b) results for the proagation constant β of the dominant TE-Mode as function of the angle θ
Figure 1b shows the calculated proagation constant β of the dominant TE-Mode as function of the angle θ. We can see a decrease of βT E by increasing the angle θ. The subplot in figure 1b shows the definition of the angle θ in the oblique coordinate system. As illustrated, the width of the structure dθ gets smaller by increasing the angle θ. This explains the behavior of the propagation constant β. In figure 2 we can see the widening to the field distribution of the dominant TE-component with increasing angles.
Fig. 2: Dominant TE-component Ex for increasing angles. The structure in figure 1a was analyzed by 90 × 90 discretisation points
[1] S. Helfert, Applying oblique coordinates to the method of lines, PIERS 61 (2006) 271–278. [2] J. Yamauchins and J. Shibayama and H. Nakano, Propagating beam analysis based on the implicit finite-difference method using the oblique coordinate system, OSA Int. Photon. Res., San Francisco, USA, 1994. [3] T.M. Benson and P. Sewell and S. Sujecki and P.C. Kendall, Structure related beam propagation, Opt. Quantum Electron., 31 (1999) 698–703.
Studies of the self–imaging effect in multimode waveguides with longitudinal periodicity S. Helfert, B. Huneke, J. Jahns In this study we examine the behavior of multimode waveguide structures where we introduce periodic perturbations. We are interested in the influence on the self–imaging effect for this case. Self–imaging means that an input field distribution repeats itself in a certain distance. The Talbot effect [1] appears in (transverse) periodic structures: the transverse periodic fields are periodic in longitudinal direction as well. Therefore, discrete values of the wave vector occur. In multimode waveguides, e.g. , couplers were built utilizing this effect [2]. For exploring further applications of this effect we studied the basic structure shown in Fig. 1a. Fig. 1b shows the propagation constants of the undisturbed multimode waveguide. We compared them to those in homogeneous media where a grating is illuminated by a plane wave, thus leading to a periodic field pattern. We see a very good agreement of the lower order modes, which propagate quasi paraxial. For the higher modes an increasing discrepancy is recognizable. After reconfirming the similarities between waveguides and homogeneous media we introduced periodic perturbations into the waveguide structure [3] (see Fig. 1a). We have a large parameter space for these perturbations, e.g. we can vary thickness, shape of the teeth, refractive index, position etc. Therefore, we started with the easiest ones (with respect to the implementation in our numerical algorithm); we used Dirac-like (i.e. very thin) rectangular teeth, and varied their position and depth. The fundamental mode of a monomode waveguide was injected. 1.53 1.52 1.51
n=1.45
t zT
n=1.52
kz / k0
1.5
20 µm
0.4 µm
zP
1.49 multimode waveguide homogeneous media
1.48 1.47 1.46
a)
periodic perturbation
b)
1.45 0
2
4
6 N,M
8
10
12
Fig. 1: a) Multimode waveguide, b) kz of homogeneous media and multimode waveguides
Fig. 2: Electric field distribution in the multimode waveguide (see Fig. 1a), undisturbed (left), t = 4µm (right)
Figure 2 shows the influence of the perturbation on the field distribution. In the original waveguide (left) we see the periodic repetition of the fields as predicted by the Talbot effect. In the figure on the right we see how the field distribution changes, e.g. we obtain two foci in a short distance. [1] H.F. Talbot, Facts relating to optical science, No. IV, Phil. Mag., 9 (1836) 401–407. [2] L.B. Soldano and E.C.M. Pennings, Optical multi-mode interference devices based on self-imaging: principles and applications, J. Lightwave Technol., 13 (1995) 615–627. [3] S.F. Helfert, B. Huneke, and J. Jahns, Studies of the self–imaging effect in multimode waveguides”, ICTON Conf., 1, 255–258, Athens, Greece, 2008.
Optical wavefields with lateral and longitudinal periodicity J. Jahns, A.W. Lohmann∗ ∗ Affiliation:
Lehrstuhl f¨ ur Multimediakommunikation, Univ. Erlangen-N¨ urnberg
Wavefields with lateral or longitudinal periodicity have been widely discussed. There is a direct connection between the symmetries of a wavefield and the self-imaging phenomenon. Wavefields with a lateral periodicity px exhibit the well-known Talbot effect [1]. Wavefields with a longitudinal periodicity pz are related to the more general case of Montgomery selfimaging [2]. Montgomery self-imaging includes the Talbot effect as a special case. Here, we consider the propagation of wavefields that are periodic in both, lateral and longitudinal direction. The situation is exemplified in figure 1a which shows the case of a Fabry-P´erot resonator with periodically structured mirrors and the outcoupled wavefield u(x, y, z). The situation also occurs for multimode waveguide devices with added longitudinal periodicity (Fig. 1b) [3]. G1
reflectivity: R
G2
1
0.8
0.6
u(x,z)=? 0.4
0.2
x
a)
z
L=pz/2
px
b)
- π 10
π 10
kx
Fig. 1: a) Setup showing a Fabry-P´erot resonator with periodically structured mirrors. The mirror reflectivity is denoted as R. b) Transfer function for the angular spectrum for R = 0.5.
Since the problem is independent of the y-coordinate, it is sufficient to use a 1D analysis. The influence of the device on the wavefield can be expressed in terms of its transfer ˜ x ) for the angular spectrum where kx is the x-component of the k-vector. It function, h(k ˜ x) = h ˜ T (kx )h ˜ M (kx ). Here, h ˜ T (kx ) describes the is given as the product of two terms: h(k ˜ M (kx ) the z-periodic structure. The latter can be influence of the x-periodic structure, h ˜ T (kx ) has peaks at kx = ±n(2π/px ) expressed in terms of kx since kx2 + kz2 = (2π/λ)2 . h M ˜ (kx ) for kx = ±(2π/pz )(2mpz /λ − m2 )1/2 , m = 0, 1, 2, . . .. with n = 0, 1, 2, . . . and h Since px and pz can be varied independently, situations can occur where the total transfer function exhibits “bands” where some peaks in the transfer function are suppressed more or less. How strongly they are suppressed, depends on the mirror reflectivity R. Figure 1b ˜ x) shows an example for R = 0.5. The black curve shows the total transfer function h(k M ˜ (kx ) shown as the grey curve. with every other peak diminished by h [1] H.F. Talbot, Facts relating to optical science, No. IV, Phil. Mag. 9 (1836) 401–407. [2] W.D. Montgomery, Self-imaging objects of infinite aperture, J. Opt. Soc. Am. 57 (1967) 772–778. [3] S. Helfert, B. Huneke, J. Jahns, Studies of the self-imaging effect in multimode waveguides with longitudinal periodicity, this Annual Report
Characterization and simulation of an integrated optical pulse shaper H. Knuppertz, M.Bohling, J. Jahns The optical performance of a integrated microoptical pulse shaper was simulated and verified experimentally [1]–[3]. Figure 1 illustrates the PIFSO-based (planar-integrated free-space optics) design, which consists of a folded 8f-system. Due to its symmetry, the design is self-compensating for certain aberrations.
Fig. 1: Microoptical pulse shaper in PIFSO-technology using an 8f setup with four reflective lenses.
These are, in part, caused by fabrication errors. In our case, the part with the four lenses was fabricated by using ultraprecision micromachining. The lenses were characterized by optical interferometry (Fig. 2). The interferograms indicate centro-symmetric aberrations.
Fig. 2: Profile of the sag deviation of the four parabolic mirrors. The overall deviation is +1 µm to –0.5 µm.
The data from the measurement of the lens profiles were transferred to standard Zernike coefficients (m, n) that here are significant for astigmatism (2/2, −2, 2, 3, 3) and coma (1, 3). Then a ray-tracing simulation model calculated the effect of the fabrication errors. In the computer simulation, we used the separation distance d of the two substrates (Fig. 1) as a parameter.
Fig. 3: Simulation results: Input beam with constant phase profile (left). After adjustment (right) the phase deviation is less than π.
By varying d , it was possible to find an optimized configuration where the influence of the fabrication errors could be essentially compensated (Fig. 3). Tests with the real system confirmed this result. [1] H. Knuppertz, M. Bohling, J. Jahns, Integrierter mikrooptischer Pulsformer f¨ ur fs-Anwendungen, 109. Jahrestagung der Deutschen Gesellschaft f¨ ur Angewandte Optik, Esslingen, 13.–17.05.2008. [2] H. Knuppertz, M. Bohling, J. Jahns, M. Bock, R. Grunwald, Ultraprecisely machined microoptics for fs-pulse shaping and replication, Ultrafast Phenomena 2008, Stresa (Italy), 09.–13.06.2008. [3] H. Knuppertz, M. Bohling, J. Jahns, Integrated microoptical pulse shaper for femtosecond applications, MOC 2008, Brussels, 24.–27.09.2008.
Multichannel microoptical rotary joint X. Br¨ uhn, J. Jahns Optical interconnects are usually used under static conditions. Due to the rapid development in the fields like spectrometry, life sciences, robotics, and data communications, it is sometimes desirable to implement interfaces between movable systems. One can distinguish between different types of movements: “slide”, “flip”, and “rotate”. Our goal here is to demonstrate a new approach to a multichannel microoptical rotary joint. In an earlier publication, Johansson and H˚ ard [1] showed the implementation of a rotary joint using diffractive optics. As other authors, too, they assumed that the used light is monochromatic and collimated. Collimation requires additional optics like microlenses in front of each source. Taking advantage of recent advances in microfabrication techniques [2] which allow one, for example, the refractive or reflective implementation of complicated optical elements, we propose a solution that does not require collimated nor monochromatic light. In figure 1 the general structure of such a rotary joint for one channel is shown. The phase function of the optical component is given as: h
φ(x, y) = −(π/λ∆z) (x − xs )2 + (x − xe )2 + (y − ys )2 + (y − ye )2
i
where λ denotes the wavelength of the incoming light, ∆z is the distance between the plane of the rotating light source and the plane of the optical element. The element is designed in such a way, that light from any position xs = R cos(α) and ys = R sin(α) on a circle with radius R in the input plane is focused to the fixed position (xe , ye ) of the receiver in the output plane. In figure 2 results of a ray-tracing simulation are depicted for eight positions of a light source on a circular path. xe x xs α ye
R y ys circular path
ring-shaped lens aperture
Fig. 1: Setup with light source on circular path (left) and ring-shaped “lens” aperture showing the variable phase profile.
Fig. 2: Raytracing result for eight positions of emitter (at the bottom) and detector position at the top.
[1] M. Johansson, S. H˚ ard, Design, fabrication, and evaluation of a multichannel diffractive optic rotary joint, Appl. Opt. 38 (1999) 1302–1310. [2] J. Jahns, Q. Cao, S. Sinzinger, Micro- and nanooptics - an overview, Laser & Photonics Reviews 2 (2008) 249–263.
Automatical adjustment of optimal VCSEL sources bias points U. Lohmann, H. Knuppertz, J. Jahns Vertical Cavity Surface Emitting Laser are important sources for optical data communication, because of the good optical characteristics and a the low threshold current of less then 1 mA. The relation between bias-, modulation current and threshold point of the VCSEL has a strong bearing on the optical performance of the laser modul. For example the lower deviation of the VCSEL threshold point creates signal degenerations as with the “turn on delay” effect (Fig. 1). Therefore an automatically adjust method for the bias point of VCSEL-Sources is needful to manufacture those transmitter modules in an efficient process.
Fig. 1: Eye diagram at 1, 25 Gb/sec data rate, with and without “turn on delay” effect
The developed method of bias point adjustment from VCSEL sources base on a rapid measuring of bias current and optical power quotient of the VCSEL source to fit the optimal working point / extinction ratio of the laser (Fig. 2). We developed a software interface for rapid measuring of the current values, an integrated fitting-algorithm and adjusting the resistor values by controller register for the laser driver.
Fig. 2: Scheme of the system for automatically adjustment of VCSEL source bias points
Optical correlation of ultra-wideband signals using time-space-conversion C. Seifarth, H. Knuppertz, J. Jahns Optical signal processing is a very promising and efficient alternative to signal processing in the electrical analog or digital domain of signals with high data rates and wide bandwidths. It offers signal processing, especially spectral filtering and correlation operations, with the speed of light and can help to reduce the speed and memory requirements in correlation receivers for RF signals significantly. Planar integration of free-space optical (PIFSO) components and systems such as optical correlators even enhances the use of processing electrical signals in the optical domain by shrinking the size of the optical setup to the size of conventional monolithic microwave integrated circuits (MMIC). Optical signals can be converted from time to spectral (spatial) domain using dispersive elements such as prisms, diffraction gratings or arrayed waveguide gratings by refraction or diffraction of a monochromatic intensity modulated laser beam. Acousto-optic deflectors are also used to modulate a laser beam by surface or bulk acoustic waves and deflect it at different angles depending on the temporal spectral components of the input signal. Ultra-wideband (UWB) devices as defined by the European Commission spread their RF energy across a bandwidth of at least 50 MHz with a low power spectral density of, for example, −41.3 dBm/MHz in the upper UWB frequency band of 6.0 GHz to 8.5 GHz. An optical correlation receiver could substitute the high data rate energy detection receiver for time-critical UWB applications presented in [1] to enhance receiver performance and robustness against noise and jammers.
Fig. 1: Spatial integrating correlator with time-space-conversion using a diffraction grating
Figure 1 shows a schematic setup of a spatial integrating correlator with a temporal laser input signal g(t) incident on a diffraction grating in plane P1 and with a spatial reference image h(x) designed as a matched filter in plane P2. The setup shown is a conventional optical 4f -setup realizable as a free-space optical setup as well as a PIFSO setup. A first feasibility study of the implementation of such an optical correlator in correlation receivers for UWB applications showed good and promising theoretical results on system level [2, 3]. Future work will include more detailed investigations on component level, due to the near-wavelength spatial resolution of the matched filter especially on wave-optical effects. [1] C. Seifarth, T. Jurenz, G. Scholl, Sub-microsecond ultra-wideband transceiver for time-critical wireless sensor networks, FREQUENZ, 62 (2008) 191–194. [2] C. Seifarth, Konzeptstudie zur Implementierung eines optischen Korrelators in Korrelationsempf¨anger f¨ ur Ultra-Wideband-Anwendungen, Master Thesis, FernUniversit¨at in Hagen, 2008 [3] C. Seifarth, G. Scholl, J. Jahns, On an optical correlator for ultra-wideband microwave signals using time-space-conversion, Proc. German Microw. Conf. 2009, Munich, 16.–18.03.2009
Volume holographic data storage in the photopolymer PQ:PMMA M. S¨ollner, Y.-N. Hsiao∗ , S.-H. Lin∗ , K.-Y. Hsu∗ , M. Gruber Volume holographic data storage offers both, high storage density and high storage capacity and it inherently supports page-oriented addressing schemes. For many years it was the lack of suitable storage materials that prevented a broad commercial use of this interesting technology. Meanwhile the situation has changed, in particular newly developed photopolymers may lead to a breakthrough. We investigated the bahaviour of one such material, phenanthrene-quinone-doped polymethylmethacrylate (PQ:PMMA) [1] which we plan to use in combination with a planar-integrated holographic r/w-head [2]. PQ has a chemical structure as shown in Fig. 1a and is added to PMMA at a typical ratio of 0.7 %. It gives the storage material a yellowish color (cf. Fig. 1b) and makes it photosensitive in the blue and green spectral regions. For preparatory experiments we could therefore use a DPSS laser with λ = 532 nm in the setup of Fig. 2a.
O
a)
O
b)
Fig. 1: a) Chemical structure of PQ and b) a material sample of PQ:PMMA.
For the recording of a test hologram a two-dimensional black and white image (containing the logo of the FernUniversit¨at) was printed out on transparent foil and positioned in the back focal plane of a Fourier lens. The spectrum was then superimposed with a collimated reference beam to record a phase-only hologram in a PQ:PMMA sample that is positioned in the front focal plane. The hologram can be reconstructed instantaneously by blocking the signal beam and illuminating the material sample only with the reference beam. Fig. 2b shows the reconstructed image as it appears on a screen.
reference beam path
b/w image
signal beam path PQ:PMMA
a)
b)
Fig. 2: a) Experimental setup for testing holographic storage in PQ:PMMA; b) screen projection of a reconstructed hologram showing the logo of the FernUniversit¨ at in Hagen.
Taking into account that the FernUniversit¨at in Hagen is an institution that exclusively engages in distance education, it is noteworthy that the above experiments were carried out in a student’s home laboratory (located about 500 km away from Hagen) that was originally designed for making large display holograms. [1] S.H. Lin, Y.N. Hsiao, M. Gruber, K.Y. Hsu, Ch. 16 in A. Friberg, R. D¨ andliker (eds.), Advances in Information Optics and Photonics, SPIE Press, 2008, pp. 317-341. [2] M. Gruber et al., Proc. DGaO 2007, http://www.dgao-proceedings.de/download/108/108 p47.pdf. ∗
Institute of Electro-Optical Engineering, National Chiao Tung University, Taiwan
Design considerations for a planar-integrated holographic r/w-head U. Vieth, M. Gruber The considerable progress that has been made in the development of photosensitive polymer materials in recent years has moved holographic data storage techniques back into the focus of scientific research [1]. Will there be something like a holographic DVD in the foreseeable future? We investigate how a suitable holographic DVD player, and in particular a r/w-head can fundamentally be constructed and propose a technological approach based on planar-integrated free-space optics (PIFSO). R*
R
R holographic storage medium
S a)
M
S* S*
S b)
R*
c)
d)
R'
Fig. 1: Holographic data storage: Conventional recording (a) and reconstruction (b), reconstruction with the phase conjugate beam R (c), single-sided r/w access by use of address beam R and mirror M .
Holographic data storage basically works as shown in Fig. 1: A signal beam S and a reference beam R are superimposed to record a hologram (a). Conventionally, S is regenerated by illuminating this hologram (only) with R, however, in this case S has to be picked up at the opposite side of the medium (b) which is not desirable for a holographic DVD player. Single-sided r/w access can be achieved by means of phase-conjugate beams: S which is counterpropagating to S can be generated by R (c), and R can be obtained through the reflection of an address beam R at a plane mirror M (d) provided that plane waves are used as reference beams. L5
L4
L3
L2
L1
LCD microdisplay collimator lens
CMOS sensor signal beam S
sm fibers = 532 nm
PBS
address beam R'
FT lens
fused silica substrate reference beam R
disk
a)
reference beam path
PIFSO system PQ:PMMA
reflection coatings
b)
Fig. 2: a) PIFSO-based design of a holographic r/w head, b) experimental demonstrator. The blow-up schematically shows the zigzag-type relay of beam R by 5 diffractive lenses L1 . . . L5 at the top surface.
Fig. 2a shows a schematic systems design that implements the above holographic r/w approach in PIFSO. Based on this design an experimental demonstrator was constructed and fabricated on a fused silica substrate by means of lithographic structuring and surface etching techniques (Fig. 2b). It will be tested with the novel photopolymer PQ:PMMA [2] as storage material. [1] D. Sarid, B.H. Schechtman, Optics & Photonics News 18 (2007) 34-37. [2] M. S¨ ollner et al., “Volume holographic data storage . . .” in this Annual Report.
Integration of microscopic darkfield illumination into a lab-on-a-chip P. Deiter, C. Bonerz, M. Gruber The life sciences have become an increasingly important field of applications for microsystems integration in recent years, the term “lab-on-a-chip” (LOC) is often used in this context. Because the manipulation of liquid substances is a standard task of a LOC it is essential that microfluidic functionalities can be integrated easily. Recently, we proposed a suitable micro-integration approach [1] that evolves from planar-integrated free-space optics (PIFSO) and makes use of planar microfluidic actuators on the basis of surface acoustic waves (SAW) [2]. Here we present a special implementation of this integration approach, a LOC demonstrator that contains a ring-shaped closed flow channel with integrated microscopic darkfield illumination (Fig. 1). absorbing glue
flow channel
cover slip
reflective coating
LiNbO3 wafer
folded light paths
slit
PCB dark-field a) mode
bright-field mode
PIFSO components
LEDs
b)
Fig. 1: a) System architecture, b) CAD model of the roof-shaped optical layer for darkfield illumination.
The particular technical innovation lies in the roof-shaped planar substrate on top, 45degree mirrors at the edges deflect the light from ultra-bright LEDs on the printed circuit board (PCB) at the bottom in such a way that it enters the flow channel in nearly horizontal direction thus implementing darkfield illumination if the observation takes place from the top. Slit apertures near the edges confine the beam divergence adequately.
a)
b)
c)
Fig. 2: a) Roof-shaped optical layer, b) assembly of the experimental demonstrator, c) microscopic observation of micro-beads (typical diameter 40 µm) in water under darkfield illumination.
The experimental demonstrator (Fig. 2) is made of PMMA and was manufactured on a turning lathe. After polishing the “roof” face relflective Al coatings were thermally evaporated and the system was assembled (Fig. 2b shows an intermediate stage in which the illumination was tested). For a “wet” test fused silica micro-beads were immersed in water and injected into the flow channel. With a microscope images like Fig. 2c could be observed. [1] P. Deiter, C. Bonerz, M. Gruber, IP 2008 Tech. Digest, Japan (2008) 3-P2, 144-145. [2] A. Wixforth, Superlattices and Microstructures 33 (2004) 389-396.
Closed-loop heating system for microfluidic channels on a lab-on-a-chip C. Bonerz, M. Gruber For the proof-of-principle demonstrator of a lab-on-a-chip on the basis of planar-integrated free-space optics as introduced on the previous page [1] a closed-loop heating system was designed and integrated into the PCB that controls the operation of the whole device [2]. With this equipment and with the ring-shaped fluidic channel that allows one to keep a liquid sample in constant circular flow the demonstrator could serve as an artificial blood vessel for biological experiments in vitro. LED Rh
NTC
a)
b)
c)
Fig. 1: a) Block diagram and b) photograph of the PCB with the control circuit. The two highlighted circles indicate the designated position of the flow channel. In c) the fluidic layer is mounted for a test of the brightfield illumination mode.
Figs. 1a and 1b show a block diagram and a photograph of the control circuit, respectively. A micro-controller IC (ATtiny26 from ATMEL) controls both illumination and heating of the microfluidic channel. For these two functionalities eight ultra-bright LEDs and eight heating resistors Rh are used and positioned around and below the circular channel, respectively, as indicated in Fig. 1b. The actual temperature is measured by means of a NTC-type resistor and the detected voltage UT is fed back into the micro-controller to determine the momentarily required heating power. A pulse width modulation (PWM) scheme is used to drive the resistors Rh appropriately.
UT
Uh
a)
b)
c)
Fig. 2: a,b) PWM heating displayed on an oscilloscope: actual temperature (blue) and heating current (yellow) for two different temperatures, c) exemplary heat image of the PCB.
This control mechanism is illustrated in the two oscilloscope images of Fig. 2, the blue and yellow lines indicate the voltage UT (→ temperature signal) and the voltage Uh over Rh (→ heating power), respectively. In the case of the lower temperature (Fig. 2b) the width of the heating pulses is larger to deliver more power. Fig. 2c is a heat image of the PCB in which one can readily recognize the eight heating resistors, in this test run they are about 25 degrees warmer than the environment. [1] P. Deiter et al., “Integration of microscopic darkfield illumination . . .” in this Annual Report. [2] C. Bonerz, MSc thesis, FernUniversit¨ at in Hagen, 2008.
CAD model and optical simulation of a PIFSO-type optical collector module M. Bohling, M. Gruber Recently, an integrated microsystems approach for chemical gas analysis by means of atomic emission spectrometry (AES) was introduced [1, 2]. It is based on sensor modules that contain a micro-hollow cathode (MHC) structure for plasma generation and a planarintegrated collector system for the efficient free-space optical coupling of the plasmaemitted light to a fiber-based spectrometer. Here, we present a CAD model of such an optical collector that serves as a basis for computer simulations of the optical coupling efficiency. top plate micro hole
inlay
collector
Cartesian oval shape
inlay
y
bottom plate
fiber
x
collector
p1
fiber mounts
Fig. 1: Exploded view of the CAD model of the optical collector system.
p2
Fig. 2: Optical imaging by means of a Cartesian ovalshaped profile in the x-y cross section.
In Fig. 1 an exploded view of the optical system which was designed in hybrid technique can be seen. The microplasma is generated in the tiny central hole of the transparent inlay from where the light spreads in all directions and enters the neighboring planar component which is denoted collector. Due to reflective coatings on the top and the bottom plate the optical system acts as a planar waveguide in x-z direction. In the orthogonal x-y cross section the curved surface performs an optical imaging operation onto the end of the fiber to the spectrometer (Fig. 2). With a Cartesian oval-shaped profile a nearly stigmatic imaging can be achieved.
a)
b)
Fig. 3: a) Calculated irradiance in the image plane p1 at the fiber end and b) across the fiber core (plane p2 ) after the in-coupling. In this example an efficiency value of η = 0.12 is obtained. R The CAD model of Fig. 1 was imported into ZEMAX to carry out raytracing simulations of the optical coupling efficiency η. Fig. 3 shows calculated irradiance patterns in the two observation planes p1 and p2 as depicted in Fig. 2. Here, a substrate thickness of 0.5 mm was assumed and η was numerically approximated by the relative number of in-coupled rays.
[1] M. Gruber, R. Heming, J. Franzke, Annual Report 2006, p. 90. [2] R. Heming et al., Proc. DGaO 2007, http://www.dgao-proceedings.de/download/108/108 p1.pdf. This project is funded by the Deutsche Forschungsgemeinschaft (DFG) of Germany (grant GR 3343/2-1)
Calculated spectral behaviour of a PIFSO-type optical collector module M. Bohling, M. Gruber On the previous page of this Annual Report [1] we reported about a CAD model for an optical collector system that is part of an integrated microsystems approach for chemical gas analysis by means of atomic emission spectroscopy (AES). Raytracing simulations showed that the plasma-to-fiber coupling efficiency can be the higher the better the fiber core diameter and the thickness of the collector substrate match [2]. What remains to be investigated is the wavelength dependence of this coupling operation. top / bottom plate fiber point source
y
x Cartesian oval profile
z
p1
p2
Fig. 1: Simulated segment that is representative for the whole optical collector system. R For this purpose additional ZEMAX simulations were implemented. Due to the angular periodicity of the system architecture it is suffcient to restrain the considerations to a single architectural segment as shown in Fig. 1. The raytracing runs were carried out in the non-sequential mode for different wavelengths and substrate thicknesses. As before, the number of rays that cross the two target areas p1 and p2 (cf. also Fig. 2 in Ref. [2]), respectively, was determined and their ratio used as an approximation of the coupling efficiency η. Fig. 2 shows a plot of η as a function of the wavelength for four different thickness settings t = 0.125 mm, 0.25 mm, 0.5 mm, 1 mm.
η
λ [nm]
Fig. 2: Plasma-to-fiber coupling efficiency η as a function of the wavelength.
As expected, η is the highest for the design wavelength λ = 532 nm. For other wavelengths chromatic focus shifts cause degradations of the coupling efficency. These shifts become a signficant problem if the sensor device is to be applicable in the visible and the near infrared spectral region. In this case it may be necessary to devise a purely reflective optical coupling system. [1] M. Bohling, M. Gruber, “CAD model and optical simulation . . .” in this Annual Report. [2] M. Bohling, R. Heming, J. Franzke, M. Gruber, Proc. EOSAM, Paris (2008) This project is funded by the Deutsche Forschungsgemeinschaft (DFG) of Germany (grant GR 3343/2-1)
Optical CCD lock-in device for Raman difference spectroscopy R. Heming, H. Herzog∗ , V. Deckert∗ Raman difference spectroscopy (RDS) is a proven technique in photo or Raman chemistry to compare different spectra of a sample probe against a reference. A common method is to use a rotating cell filled with two or more different samples. A new multi-channel lockin technique for the detection of periodically varying Raman spectra was developed [1]. As a performance check a light-activated and a dark Bacteriorhodopsin (BR) sample were measured with the two-cell disk. This biological Bacteriorhodopsin probe as a difference spectra demonstrate the capabilities of the scheme which incorporates sub-pixel precision spectra while avoiding environmentally induced drift. The modular VHDL-design enables chip-level compatibility with modern high-end liquid-cooled CCD camera devices.
Fig. 1: Schematic design of the instrument.
Fig. 1 gives an overview of the instrument design. The main task of the FPGA-board is to provide a proper periodic timing to shift the CCD charges from the different Raman spectra up and down synFig. 2: Test with Bacteriorhodopsin. chronized with the different probes. Depending on the position of the rotating samples the Raman laser is chopped on and off phased by the FPGA-board. In order to improve signal-to-noise ratio one Raman measurement can contain up to 10000 cycles of optically accumulated signals from the same sample position. In order to gain maximum flexibility many important parameters like the number of accumulations and the shifting distance can be chosen by the user. Fig. 2 shows the two simultaneously detected BR forms (below) and the (light-dark) difference spectrum (above). Note the change in terms of absolute intensity. The results demonstrate the capabilities of the technique [2]. In particular when only small intensity and frequency shifts occur. The precision of the method is much higher than comparable setups, because temperature drift and laser intensity changes between measurements are avoided. Furthermore, sensitivity variations on the detection chip are cancelled, as always the same region is used for the acquisition. Last but not least, the general scheme of the method can be easily applied to the detection of all periodically varying signals. [1] R. Heming et al., Proc. DGaO 2008, http://www.dgao-proceedings.de/download/109/109 p33.pdf. [2] V. Deckert et al., “Raman detection of oscillating signals using a phase synchronized CCD readout”, ICORS 2008. ∗
ISAS – Institute for Analytical Sciences, Dortmund
Control Systems Engineering Prof. Dr.-Ing. Helmut Hoyer
Mechatronics Prof. Dr.-Ing. Michael Gerke Universit¨atsstraße 27 58084 Hagen Phone: +49-(0)2331-987-1100 Fax: +49-(0)2331-987-354 E-Mail:
[email protected]
Staff:
Phone:
M.Sc. Naef Al-Rashedi Dr.-Ing. Andreas Bischoff Dr.-Ing. Ulrich Borgolte Wolfgang G¨ulland Dipl.-Ing. Heiko Heinze Dipl.-Ing. Frantiˇsek Jelenˇciak Gabriele Kappenstein Dr.-Ing. Ivan Mas´ar Dipl.-Ing. Peter Seibold Josef Tschuden Dipl.-Inf. Bernd Winnige
-1702 -1105 -1106 -1112 -1118 -1119 -1101 -1102 -1702 -1103 -1108
Forest fire prevention and extinction (iWBB) U. Borgolte, M. Gerke The Control System Engineering group and Mechatronics group are engaged in the project International forest fire combat, which is supported by the Ministry of Economic Affairs and Energy of the State of Northrhine-Westphalia [1]. The project consortium consists of 5 companies, 2 universities, and 1 research institute. It started at the beginning of 2008 and has a duration of 24 months. The project is motivated by the fact that every year several thousand square kilometres of forest are destroyed by fire. For the mediterranean region alone, WWF estimates 8000 square kilometres every year. Within the iWWB project, a technically advanced and more general system approach will be implemented and tested. This approach covers fire detection, fire fighting, rescue of persons, and fire aftercare, and it will be worked out by our research group and other project partners. Here our blimp-type airship is intended to be used as a carrier of sensoric systems (heat/smoke/ gas detection, pollution monitoring) and as relay station for wireless communication. Its advantages are the possibility to stay above ground for long periods of time and to be able to lift large payloads. Together with micro-drones, it will be responsible for the following mission tasks: • immediate inspection of suspicious areas, reconnaissance of passable access routes; • establishment of a flying communication relais (airborne router); • long-term monitoring of areas, where the fire is supposed to be extinguished.
Figure 1: Our Research Centre in ZSK-Hemer [ 1 ] Verbundvorhaben iWBB: Projektf¨orderung durch das Ministerium f¨ur Wirtschaft, Mittelstand und Energie des Landes Nordrhein-Westfalen F¨orderkennzeichen 213-81-01/01-08
Mine detection and removal (iMR) U. Borgolte, M. Gerke The Control System Engineering group and Mechatronics group are engaged in the project International mine detection and removal, which is supported by the Ministry of Economic Affairs and Energy of the State of Northrhine-Westphalia [1]. The project consortium consists of 7 companies, 2 Fraunhofer institutes, and 1 university partner. It started at the beginning of 2008 and has a duration of 24 months. The project is motivated by the fact that large areas in the world are almost unlivable due to landmines. Within the iMR project, some new technical approaches will be implemented and tested by our research group and other project partners. These approaches cover both mine detection and mine removal. The blimp-type airship of our research group is intended to be used as a carrier of the sensoric systems. Its advantages are the possibility to stay above ground for long periods of time, to be able to lift larger payloads than mini-drones, and to exert no force to the ground while hovering in low altitude. • Thus the first task of the blimp is to locate mines by appropriate sensorics. • The second task is the inspection of the mine after its destruction to ensure that the mine is totally destroyed.
Figure 1: Airship prototype The control systems engineering group and the mechatronics group at FernUniversit¨at are currently concentrating on the projects iWBB and iMR. The following 7 pages are describing different aspects of the project work. [1]
Verbundvorhaben iMR / LASER: Projektf¨orderung durch das Ministerium f¨ur Wirtschaft, Mittelstand und Energie des Landes Nordrhein-Westfalen F¨orderkennzeichen 213-82-01/01-08
Dynamical Modeling of a Robotic Airship O. G¨oller, I. Mas´ar and M. Gerke A basic flight control for a robotic airship can be set up with a Fuzzy rulebase for simple types of motion (e.g. stationary hovering with disturbances). However, the derivation of a complete mathematical model of the blimp dynamics is mandatory for the control of complex flight missions and for path planning. In general, the non-linear dynamic model of an airship, its geometric shape, its aerostatical and aerodynamical behavior, and its actuation system, have to be considered: F
dx = C(x) + A(x) + G(λ, W , B) + P dt
A dynamical model of our Indoor-airship has been derived ([1],[2]) in a state-space representation regarding all physical effects occuring during flight. To provide a powerful simulation environment, the full dynamical model has been integrated into MATLAB/Simulink and it has been visualized by means of its Virtual Reality-Toolbox for controller development.
Figure 1: Dynamical model of the airship in a Virtual Reality environment This work was presented on the 5th International Conference on Electrical and Power Engineering (EPE’2008), in Iasi, Romania, where Dr. Mas´ar gave the Plenary Talk about our activities in flight robotics. [1]
Stefan Schnabel: (in German), Herleitung eines dynamischen Modells f¨ur ein Luftschiff unter Verwendung von MATLAB/Simulink Diploma Thesis 2006, FernUniversit¨at in Hagen
[2]
Oswald G¨oller: (in German) Erstellung eines MATLAB/Simulink-basierten Virtual Reality-Modells zur Simulation des Flugverhaltens eines Luftschiffes M.Sc. Thesis 2008, FernUniversit¨at in Hagen
Marker based self-localization with a digital signal processor H. Heinze
We utilize an autonomous airborne blimp as a sensor platform to support wildfire and land mine fighting. To assure easy manoeuvrability and handling, flight sensors and a tracking system for path-planning and trajectory validation are used. To support the staff and to protect the blimp against accidental collisions during take-off and landing operations, a self localization system is necessary. To assure accurate positioning, besides conventional systems such as GPS and acceleration sensors, an image processing system for self-localization was tested. Encoded visual markers with known positions allow referencing with respect to the environment, therefore the position of the airship can be determined for all six degrees of freedom. By this way, a landing field can be flagged by such a marker, which can serve as a landing beacon. To implement such a marker tracking ability, the MATLAB augmented reality Toolbox was used as the basis of research. Future applications will have to manage the memory in a way, that high resolutions of video data are supported. With increasing distance, accuracy and detection quality decrease, therefore the use at greater distances and heights is limited. For such situations, other systems on board the airship have to take the lead. Markers for image recognition are nowadays used in many forms. But there is no light-weight and energy saving system on the market. This can be achieved by use of a digital signal processor (DSP), particularly with fixed point arithmetics, that lower the cost and gain speed while offering low weight and low energy consumption. Focus of the research lies in the implementation of known algorithms for position detection, which are computable in real-time.
Figure 1: Self-localization and autonomous landing of a blimp
[1] Heinze, H.: Selbstlokalisation eines autonomen Roboters basierend auf einem Fiducal Marker System. Diploma thesis, FH S¨udwestfalen, 2008 (in German). [2] Kato, H.; Billinghurst, M.: ARToolKit Documentation. http://www.hitl.washington.edu/artoolkit/
Path planning and mapping with autonomous and semi-autonomous aerial systems U. Borgolte Within the two ongoing projects iWBB and iMR, the PRT-blimp acts either as carrier for application-specific sensor systems or as relay station for communication between the control station and other mobile systems. Therefore, it is necessary to drive the blimp to a defined position in airspace and to be able to move it on a precalculated path in 3D-space with a given velocity. There are two reasons to take orientation into account. First, application sensors need to be aligned according to given values. Second, due to aerodynamical and actuation constraints, a proper direction towards the goal pose needs to be achieved. Thus, controllers for six degrees of freedom are to be developed. Both stationary pose data and dynamic paths are defined with respect to the global earth coordinate system. Commands for movements are usually given with values taken from conventional geodetic maps (e.g. topographical maps). The 6D-feedback from the blimp about its current pose in space should be in a corresponding format. This is not only needed for control of the position/orientation and trajectory of the blimp, but also for referencing of the data from the application sensors. Up to now, work is concentrated on identification of common data formats and appropriate software to access and manipulate map data. It is expected that commercial software will not cover all demands, therefore open source software is preferred. Position data needs to be exchanged with the control station, so it is necessary to use standardized map formats. Printed maps are not useful if automated processes are employed in the control station. Thus, digital maps are needed. There are two kinds of digital maps: digital raster graphics based and vector graphics based (fig. 1). The usage of a specific format is mainly determined by the availability of maps with a specified resolution.
AA A
@ @ A
Figure 1: Raster graphics vs. Vector graphics
In principle, maps open to the public as Google Maps can be used, but these are often out-dated and not sufficient with respect to resolution. Maps maintained by public bodies such as land surveying offices or military maps are better suited in most cases. To make the best use of maps, access software needs to fulfill the following requirements: possibility to adapt to specific demands (open source), low cost, updating with own data, and integration into the control software of the blimp. The land surveying office of Hagen (Vermessungsamt Hagen) gives support for the decision on maps and software to be used. [1] N. Bartelme: Geoinformatik. Springer 2005. [2] T. Mitchell: Web Mapping. O’Reilly 2008.
Path Planning for Autonomous Flight Systems Naef Al-Rashedi Path planning for a robot flight system is very important within the Airship project, since the flight system should move only some meters above the ground. The main goals of path planning are: 1- Find a minimum path length for the airship to scan a specific area. 2- Avoid any unexpected objects and reach the destination without getting lost or crashing into any object. To achieve these goals the path planner works into two phases: 1- Off-Line path planning: In this phase the path planner reads information from digital map for a specific area and uses genetic algorithms to find the minimum path length to scan this area. This problem corresponds to finding the shortest Hamiltonian cycle in a complete graph G = (V,E) of an n nodes. Thus the path planner consists of finding a permutation of the set {wp1,wp2,wp3,…,wpN} that minimize the quantity:
Minimize
N −1 ∑ d ( wp i , wp i +1 ) + d ( wp N , wp 1 ) i =1
where d(wpi, wpj) denotes the distance between waypoint wpi and waypoint wpj. 2- On-Line path planning: In this phase the path planner reads information from on-board sensor systems and uses the genetic algorithm to avoid any detected obstacles. A genetic algorithm for both Off-line and On-Line phases will soon be designed, implemented and tested using MATLAB® and Genetic Algorithm TB and Direct Search TB from the MathWork Inc.
Figure 1: Path between two points avoiding the highest points in a terrain. This work is supported by a joint Ph.D. grant of the German DAAD and the country of Yemen, starting in October 2008. (1) Anargyros N. Kostaras, Ioannis K. Nikolos, Nikos C. Tsourveloudis, and Kimon P. Valavanis, Evolutionary Algorithm Based Off-Line/On-Line Path Planner for UAV Navigation. Proceedings of the 10th Mediterranean Conference on Control and Automation – MED 2002 Lisbon, Portugal, July 9-12, 2002. (2) Isil Hasircioglu, Haluk Rahmi Topcuoglu, Murat Ermis: 3-D Path Planning for the Navigation of Unmanned Aerial Vehicles by Using Evolutionary Algorithms. Proceedings of the 10th Annual Conference on Genetic and Evolutionary Computation. Atlanta, GA, USA. 2008 (3) Zixing, CAI., Lingli, YU., Chang, XIAO., Lijue, LIU.: Path Planning for Mobile Robots in Irregular Environment Using Immune Evolutionary Algorithm. Proceedings of the 17th World Congress The International Federation of Automatic Control Seoul, Korea, July 6-11, 2008.
Navigation with camera pictures - Simulation with Virtual Reality P. Seibold Our airship is partly controlled by means of image processing with respect to position, orientation and speed. In order to receive 2D information, a camera is mounted underneath the airship. Every 20ms a picture is taken and compared to the preceding picture. The comparison delivers a measure of the translation and rotation of the two pictures in relation to each other. With this, the new position and orientation can be determined with respect to the preceding position. The height above ground is assumed as given. With a determined starting point, the absolute position is calculated by dead reckoning. The calculation method used is the “Optical Flow”, introduced by Lucas and Kanade [1, 2]. In an ideal case, for every picture dot the disparity in pixel is given. The following pictures show a simulation with the help of a virtual world (VRML) environment [3].
Figure 2: First picture
Figure 1: Test picture for dead reckoning
Figure 3: Second picture
In the left picture the given path is marked by big squares. Small squares depict the positions in relation to the preceding picture, while small circles indicate the result of the dead reckoning. Both pictures on the right contain the flow vectors of 2 consecutive pictures. Each arrow starts at the old pixel position and points to the new pixel position. [1] B. D. Lucas, T. Kanade: An Iterative Image Registration Technique with an Application to Stereo Vision. International Joint Conference on Artificial Intelligence, pages 674-679, 1981. [2] B. D. Lucas: Generalized image matching by the method of differences. PhD thesis, July 1984. [3] VRML, Virtual Reality Modeling Language, e.g.: http://de.wikipedia.org/wiki/VRML
Navigation with camera pictures - Experiments with Gantry Crane P. Seibold The final goal of the work described here is to control the position, orientation and speed of an airship by image processing. Two cameras are mounted on the airship. By comparing the pictures of both cameras, taken at the same time, the distance above ground can be calculated. By comparing two pictures from one camera taken at different times the distance and the rotation of the airship movement can be estimated and with the time difference the speed vector can be calculated. So far simulations were done in our lab with the use of virtual reality (VRML) [1]. The results were very promising. But the artificial pictures were not close enough to reality, because the two pictures compared were exactly the same in the matching section, the illumination remained constant and there was no variation in height above ground.
Figure 1: Gantry crane testbed
With the help of a gantry crane, real photos can be taken with an experimental setup. The dimensions of the table are 2.2m∗1.2m∗3.1m [X∗Y∗Z]. The crane covers an area of approx. 2m∗3m. The maximum speed is 10m/s. The position of the crane load (camera) in this experiment has an accuracy of 0.1 mm. The rotation of the camera is ± 170°, with 0,075° deviation. By changing the optical zoom different base altitudes can be simulated. Angle, zoom
The crane system is controlled by a host-PC which sends the desired position to a target PC. The position is set by a real time controller in the target PC. The cycle time is 10ms. When the correct goal position is attained, the target PC sends a message to the host PC. The camera rotates to the desired angular orientation and the picture data are captured. By taking several pictures on a given path, a flight of the airship is simulated. Taking pictures on a parallel path gives the stereo pictures for the elevation calculation.
RS232
Host
camera
ready Video data XY
TCP/IP
Position reached
XY
Target
XY-table
(XY control) position
Figure 2: Control layout of the testbed
Processing of the pictures will result in a description of the “flown” path. Processing is done by using methods of the Optical Flow [2]. [1] VRML, Virtual Reality Modeling Language, e.g. http://www.w3.org/MarkUp/VRML/ [2] B. D. Lucas, Generalized image matching by the method of differences, PhD thesis, 1985.
Communication scenario airship - base station Andreas Bischoff http://prt.fernuni-hagen.de/˜bischoff/ The communication link between the PRT autonomous airship (blimp, UAV) [1] and ground control station is realized redundantly over three communication channels in different frequency ranges. The intended communication channels consists of a WLAN connection following the IEEE 811.2h standard in the 5 GHz band, a serial connection within the 800 MHz band and a GPRS/UMTS broadband connection in 1,9 to 2.1 GHz band. Apart from the redundancy (connecting security, reliability) this structure offers Internet connection with the advantage to access external maps, GIS data (geographic Figure 1: Communication structure information systems) and Web services. Furthermore a simple interoperability can be realized to external services of the project partners. An extremely flexibly user interface can be integrated in such a way. Each notebook with WLAN or GPRT/UMTS Internet access or small devices like PDAs or smartphones can be used for ground control. In order to ensure real redundancy, advanced routing and load balancing techniques are used to realize traffic shaping and traffic control (quality of service) over a VPN (virtual private network). This approach allows the usage of the full bundled communication bandwidth. It realizes a safe and stable data link secured over VPN. The mobile phone providers in Germany have reached a nearly complete network coverage today. Even in case UMTS and EDGE coverage is not available GPRS connections with up to 58,6 kBit/s can be used country wide. This bandwidth is sufficient for transmission of highly compressed MPEG4 video streams, sensor and command data. It can be expected that in short time full UMTS coverage with bandwidths of 384 kBit/s up to 7.2 MBbit/s will be available. UMTS uses frequency ranges from 1.9 GHz to 2,1 GHz (Germany). The related communication hardware is very inexpensive. Very low power consumption and very low weight are further advantages of this approach. However communication latency is much higher (GPRS up to 2 seconds, UMTS ≤ 1 sec.) than with other proposed channels. Two mobile communication contracts are necessary for the implementation. However appropriate inexpensive flat fees are available. The coverage of this communication approach substantially reduces the risk of a loss of the blimp. The selected components are integrated at present and tested in an embedded PC system. Range and throughput measurements are still to be carried out. A Web-based user interface is currently under development. [1] http://prt.fernuni-hagen.de/pro/blimp/en/index.html [2] http://www.t-mobile.de/funkversorgung/inland
Constrained pole assignment control for a nonlinear plant F. Jelenčiak, M. Huba, I. Masár and M. Gerke Temperature control represents one of the basic tasks in many areas of control applications. At the same time, by its specific dynamics with the fast and slow mode, by its nonlinear character, available sensors, easy construction and maintenance and a simple physical interpretation of running processes, control of scaled laboratory thermal plant models is appropriate also for demonstrating and comparing different methods and tools of identification and control. The available thermo-optical plant laboratory model (Fig.1) consists of a halogen bulb 12V DC/20W (elements 1-6), of a plastic pipe wall (element 7), of its internal air column (element 8) containing the temperature sensor PT100, and of a fan 12V DC/0,6W (element 9 that can be used for producing disturbances, but also for control). Thermal plant represents one channel of the plant that can be simply used in university laboratories for remote experiments. In total it offers measurement of 8 process variables: controlled temperature, its filtered value, ambient temperature, controlled light intensity, its derivative and filtered value, the fan speed of rotation and current. The temperature and the light intensity control channels are interconnected by 3 manipulated voltage variables influencing the bulb (heat & light source), the light-diode (the light source) and the fan (the system cooling). Besides these, it is possible to adjust two parameters of the light intensity derivator. Within Matlab/Simulink or Scilab/Scicos schemes the plant is represented as a single block and therefore avoiding the need of costly and complicated software packages for real time control. The (supported) external converter cards are necessary just for sampling periods below 50ms.
Figure 1 - The thermo-optical plant and scheme of its thermal channel The plant is used for developing a broader library of tasks to be solved by students and, at the same time, also for demonstrating research results aimed at identification and control of linear and nonlinear systems. One paper presented at the EPE conference [1] was devoted to plant identification using a recursive method of consecutive integrals and by application of the identified model for tuning of constrained pole assignment controller. Experiments proved an excellent control quality that can be shown to be close to the fastest possible monotonic step responses. For the next future it is planned to use the plant for demonstrating possibilities of advanced nonlinear identification and control. [1] Jelenčiak, F.; Huba, M.; Masár, M.; Gerke, M.: Constrained Pole Assignment Control for Nonlinear Plant. In: Proceedings of the 5th International Conference on Electrical Power Engineering : Iasi, Romania, 3.-5.10.2008. - Iasi : Technical University, 2008. - ISBN 12238139. - S. 547-554
Online robot simulation contest Ivan Masár As mobile robots and various robot contests became very popular in the past years, we developed the concept of a novel event for students of our master module „Mechatronics and robotics“ – a virtual robot contest [1]. The main idea was to allow students to participate in a robot contest and to design control strategies for various robot tasks without having a real robot. Instead, they can use a simulation and a virtual model of the mobile robot F.A.A.K., which was designed at our department. Typical objectives of the competition are line following or path finding in a labyrinth. The virtual robot contest consists of an asynchronous and a synchronous phase and is organized as follows. At the beginning of the contest, the participants get the simulation model of the robot and development tools. The robot is programmed and simulated using a comprehensive integrated environment based on the MATLAB/Simulink software package. These developing tools include simulation models of all robot subsystems (drives, sensors, visual system, etc.) and also a model of its environment. Thereby, complex simulations of the robot interacting with its environment can be performed and visualized by a 3D robot model. In the asynchronous phase of the contest, the students work alone or in small teams on their solution of the given problem. During a term, they get an introduction from the lecturer and can exchange their experiences. Moreover, they can discuss various problems which arose while programming the robot during online sessions. These sessions take place in regular intervals (every two or three weeks) during this phase. In the final stage, the robot contest is organized as a synchronous event. The students use a VNC-client to watch the simulation of the competing mobile robot and use Teamspeak for voice communication during the contest. On the contest day, the students get the final map for the particular task and upload their control algorithms. Thereafter, the robots are simulated on the server in Hagen and the participants can watch the simulation. After the first round of the competition, the students get some time to change/adapt their algorithms to achieve better results. Afterwards, the final round decides about the winner. The online simulation robot contest was organized for the first time in the winter term of 2006. Because of the positive response from the students, this event takes place regularly in both winter and summer term. The main improvement in the future will be the option to program the real mobile robot with the best algorithms after finishing the contest. A video of this event will be available on our web-site and will demonstrate the robot’s performance in a real environment.
Figure 1: „Path following” contest
Figure 2: „Labyrinth crossing” contest
(1) http://prt.fernuni-hagen.de/pro/faak/roboterwettbewerb
Pediaphon presentation at the Learntec 2008 fair Andreas Bischoff http://prt.fernuni-hagen.de/˜bischoff/ At the Learntec 2008 [3] fair in Karlsruhe the ’Pediaphon’ project [5] was presented at the university joint booth. Pediaphon is a web based service which generates audio representations of Wikipedia articles dynamically. The tool is usable on- and off-line, as a web based service to listen to the articles directly in the web browser as well as to download MP3 files for later use on mobile devices like MP3-players and mobile phones. The mobile phone branch of the service was enhanced by a new touch tone input approach [1]. Like the input of a SMS the numerical keys must be pressed more than once to access the corresponding letter. Numerical keypads are usually already labeled in that way. The chosen letter will be repeated after each input. A voice menu assists the user during input. After successful processing of the search keyword the playback of the requested article starts. During audio playback Figure 1: ZDF camera team at PRT exhibit the users are able to navigate in the phone announcement by pressing predefined keypads. At any time users can stop the playback and redefine their search. The project was introduced in the 3SAT magazine ’neues’ [4] as an example of mobile learning [2]. [1] Andreas Bischoff. Location based cell phone access to the wikipedia encyclopedia for mlearning. In Dirk Ifenthaler Pedro Isaias, Miguel Babtista Nunes, editor, Proceedings of the IADIS International Conference WWW/INTERNET 2008, pages 366-370, Freiburg, Germany, October 2008. IADIS, IADIS Press. [2] Andreas Bischoff. M-learning, mobile experimentation and telepresence with cell phones and pdas. In Reinhard Langmann Michael E. Auer, Editor, International Conference on Remote Engineering and Virtual Instrumentation, REV2008, Duesseldorf, Germany, June 2008. Kassel University Press. [3] http://www.learntec.de [4] http://www.3sat.de/mediathek/mediathek.php?obj=7411&mode=play [5] http://pediaphon.fernuni-hagen.de
Augmented sensor data for mobile remote experimentation Andreas Bischoff http://prt.fernuni-hagen.de/˜bischoff/ To establish mobile learning environments we have adapted Webbased remote laboratories to mobile devices like PDAs and smart phones. Our on-line lab experiment is a remotely controlled Pioneer 3 AT mobile robot [1]. A platform independent approach to realize web-based user interfaces on mobile devices is the use of asynchronous Javascript. With asynchronous Javascript an application behavior like Java applets or interactive Flash movies can be implemented. In combination with the exchange of XML based messages between client and server, these techniques are well known under the term ’AJAX’. Windows Mobile does not support an AJAX framework directly, but asynchronous HTTP requests are possible in Javascript. The new asynchronous web client Figure 1: Windows Mobile 5 augmented asynchronous works on each Windows Mobile deJavascript client 240x320 pixel vice without any additional software. Opera Mobile on Symbian OS, the new Google smart phone OS Android [2] and Apples iPhone are supported as well, only the video streaming solution have to be adapted for these platforms. To overcome restrictions dealing with limited screen sizes, sensor data like position information and a moving map have been augmented in real time into the video stream to save limited place (320x240 pixels) for the GUI. [1] Andreas Bischoff. M-learning, mobile experimentation and telepresence with cell phones and pdas. In Reinhard Langmann Michael E. Auer, Editor, International Conference on Remote Engineering and Virtual Instrumentation, REV2008, Duesseldorf, Germany, June 2008. Kassel University Press. [2] http://www.android.com
Robosapien RS Media robots Andreas Bischoff http://prt.fernuni-hagen.de/˜bischoff/ The Wowwee Robosapien RS Media robot is an interesting device for remote experimentation, lab practice, e-learning and teaching ’robotics’. Since the robot was produced as a toy it is available on the market for a low price. The robot is equipped with a camera module, infrared sensors, a microphone and bumpers. The on-board ARM processor runs a customized embedded Linux OS. All robot joints can be controlled by shell scripts or by a J2ME API. The robots behavior can be programmed graphically via a simple Windows based software. The robot can playback prerecorded MP3 files depending on sensor events. The embedded Linux kernel does
Figure 1: PRT Robosapien robot fleet, bluetooth serial console modification not support TCP/IP networking and kernel configuration for the device is not available. To access the robots command prompt the ARM processor board serial console is connected to a bluetooth module [1]. We have realized a robot Web interface for PDAs and mobile phones. Due to the lack of TCP/IP support, a Web-based interface to the robot requires an external web server. This web server sends client movement requests via CGI script to the serial console of the robot. The user is able to request remotely from the robots integrated camera to take a photo, and to initiate the upload via zmodem to the web server.
Figure 2: Web interface Windows Mobile, Google Android simulator
[1] http://www.evosapien.com/robosapien-hack/nocturnal2/node/447
DAAD project „Autonomous airships“ M. Gerke and I. Masár The DAAD [1] PPP project „Autonomous Airships“ was initiated in 2006 and its main aim was to strengthen the collaboration between FernUniversität in Hagen and Slovak University of Technology in Bratislava, Slovakia, by the joint development of autonomous flying robots. At present, both participating research groups are working on similar mobile systems: - Control System Engineering Group in Hagen (Prof. Dr.-Ing. Michael Gerke) on an autonomous airship and - Institute of Control and Industrial Informatics in Bratislava (Prof. Dr. Mikuláš Huba) [2] on an autonomous helicopter. Within this project, several students and researcher exchanges took place between both universities. During the first year of the project (2007), the concept of an autonomous indoor airship and some of its subsystems have been developed: • Inertial navigation system, based on acceleration and angular rate MEMS sensors with data processing algorithm based on Kalman filtering; • Image processing unit for airship navigation using landmarks; • Remote control for teleoperation of the airship. The activities of the second year of the project (2008) have been concentrated mostly on identification and visualization of autonomous airships. Presentation of research results achieved during the project and academic activities were given to members of both participating universities. Moreover, information and experiences about eLearning methods in teaching of control and automation have been exchanged with special respect to experiments in mobile and flight robotics.
Figure 1: Presentation of Dr.-Ing. A. Bischoff on the 8th International Conference „Virtual University 2007“ in Bratislava
Figure 2: Slovak students and researchers during their visit at the FernUniversität in Hagen in summer 2008
After finishing the project it can be stated, that the exchange of scientific know-how and cooperation by solving of various problems has rapidly accelerated the development of both types of flying robots (airship and micro-helicopter) and also the relation between both groups has been tighted. Future cooperations and common projects are under consideration. (1) http://www.daad.de (2) http://www.urpi.elf.stuba.sk
Analysis of the ATF DINGO 2 with passive and active chassis Marco Kl¨ockner and Michael Gerke The development of a simulation model for the full-protection transport vehicle (ATF) DINGO 2 with respect to the analysis of its stationary and non-stationary dynamics during motion and its vertical dynamics as a reaction to steering and lateral runway profiles were subject of this M.Sc. thesis. A specific multi-body model of the vehicle was deduced from all substantial subsystems based on manufacturers data or on-site measurements, thus offering access to modal analysis of the passive SIMPACK model. This model was validated during experiments on the WTD-41 test tracks. This model was extended through active elements of the AGIL-R wheel suspension systems [2], leading to a coupling of MKS-SIMPACK and MATLAB/Simulink. This provides a first approach to controller implementation and gives insights to the performance of an active suspension system even before a mechatronical prototype exists.
Figure 1: DINGO 2 - Test system The M.Sc. thesis of Mr. Marco Kl¨ockner was finished as number 10 in the specialisation ‘Mechatronics’ of our M.Sc. programme in Electrical Engineering. [1]
M. Kl¨ockner: Modellbildung und Analyse des Allschutztransportfahrzeuges DINGO 2 mit passiver und aktiver Fahrwerksauslegung M.Sc. Thesis 2008, FernUniversit¨at in Hagen (in German)
[2]
E. Sch¨afer, A. Wielenberg: Realisierung einer aktiven Federung f¨ur ein gel¨andeg¨angiges, allradgetriebenes Radfahrzeug der 8-12 Tonnen Klasse (AGIL-R) Abschlussbereicht 2006, Universit¨at Paderborn
Learning environment for behaviour based robot programming Th. Jourdan, U. Borgolte
In the 80s of the last century, developments were made to enable autonomous systems with some kind of ‘intelligence’, which needs no internal representation of the environment. This was inspired both by biological research on insect behaviour and by new theses on modelling of intelligent behaviour. First promising results have been achieved with small mobile robots. From the historical point of view, behaviour based programming is in contradiction to classical methods of artificial intelligence. Nowadays, both approaches are developed further on under the generic term embodied intelligence. The master thesis describes basic principles and elements of behaviour based programming and analyses existing solutions. The focus is on Brooks’ subsumption architecture and on Braitenberg vehicles. Based on constructivist didactics, a method is introduced to make students familiar with these concepts. Special emphasis is placed on the subject-specific dimension of behaviour based programming. It is intended that the learner is directly interacting with the robot. A prototypical learning environment for behaviour based programming is introduced (fig. 1). It is composed of an integrated software development environment, a domain specific programming language, and firmware for the embedded robot control. The very limited hardware resources of the target platform are taken into account. The implementation was done with open source software components and tools. Main elements of the GUI of the integrated development environment are: 1 Program editor, 2 actors, sensors, diagnostic LEDs, 3 process data, and 4 an interactive tutorial. The environment has been tested with a small mobile robot, the c’tBot. The computing architecture of the robot is based on an ATmega32 CPU, 32 KByte Flash, 2 KByte RAM, and 1 KByte EEPROM. The sensor instrumentation is very comFigure 1: Development environment prehensive for this kind of robot. It comprises two infrared distance sensors, two light sensitive resistors, two wheel encoders, 4 light barriers, and an optical mouse. [1] Th. Jourdan: Konzeption und Realisierung einer Lernumgebung f¨ur Behavior Based Mobile Robot Programmierung auf Basis g¨unstiger Systemkomponenten. Master thesis, FernUniversit¨at in Hagen, 2008 (in German). [2] R. C. Arkin: Behavior-based Robotics. MIT Press, Cambridge, MA, USA, 1998. [3] R. A. Brooks: A robust layerd control system for a mobile robot. http://people.csail.mit.edu/brooks/papers/AIM-864.pdf.
(AIM-864), 1985.
Dies Academicus 2008 Michael Gerke Dr.-Ing. Mas´ar, scientist of our research group, has received a prize for his outstanding Dissertation thesis in the field of robotics on the Dies Academicus 2008 event of FernUniversit¨at in Hagen. The topic of his thesis is on ’Construction, Modeling, Control and Path-Planning for a quasi Omnidirectional Mobile Robot’. The dissertation thesis and the related scientific work is focussed on the construction and development of a new type of omnidirectional robots with high manoeuvrability in restricted environments and led to the real robot F.A.A.K., which participated in several robotic contests. The mobile robot is equipped with Digital Signal Processors for control purpose and an image processing system for obstacle recognition and path following. It possesses 4 steered wheel, two of them are actuated. In his ’laudatio’, Prof. Hoyer (Rector of FernUniversit¨at) stressed the highly demanding scientific character of this Dissertation thesis with respect to its multidisciplinary system approach in the field of ’robotics and control’.
Figure 1: Laureates of the Dies Academicus 2008 (Dr. Mas´ar: rear row, third from left) Our research group congratulates Dr.-Ing. Mas´ar for receiving this honorable prize, which is donated by the Sparkasse Hagen. [1]
I. Mas´ar: Konstruktion, Modellierung, Regelung und Bahnplanung eines quasi-omnidirektionalen mobilen Roboters Dissertation Thesis 2007, FernUniversit¨at in Hagen (in German)
