ITS. System. in a number of cnvironments, indoor and outdoor, in a lie-of. (Quadrature Phase Shift Keyitg) modulation and

2006 6th Intermational Coference on ITS Teleconuiiic&ations Proceedings Palmnyre .: a MIMO reconfigurable transmission platform or ITS aplicatio...
2 downloads 0 Views 5MB Size
2006 6th Intermational Coference on ITS Teleconuiiic&ations Proceedings

Palmnyre

.: a MIMO reconfigurable transmission

platform

or

ITS

aplications

Yvon-Marie Le Roux, Patrick Lassudrie-Duchesne Jacky Menard, Daniel Bofrreau, Eimnanuel Daniel, Gabrielle Laidrac GET - ENST Bretagne, Techilopole Brest-Iroise CS 83818, 29238 BREST Cedex 3. France Email: Yvon.LeRoux denst-bretagne.fr AbstractIntelligent Transportation System telecommunications require the development of reconfigurable demonstration platforms. The PALMYRE platform is a federative project with the aim of coordinating and developing research activities and application programs in the fields of radio communications and lITMO systems within public and private laboratories in Brittany, France.

in a number of cnvironments, indoor and outdoor, in a lie-of sight configuation. The robust digital transmiission sclhme involves a QPSK (Quadrature Phase Shift Keyitg) modulation and synchronization algorithms. The digital parts are implemented on FPGA-based (Field Programtmale Gate Arrays ) prototyping boards. The radiofreqency circn1it performs the signal transposition from the low IF to the eission frequLency, cuiently in the 5.15 - 5.3 5 GHz band. - The second step consists in the development of a reconfigLirable MIMO platfonri. The simulation of realistic MIMO systems requires advanced propagation and system modelling be performed. In particular, crucial questions htve to be investigated conwerning the decorrelation of simultaneous MIMO propagation paths within an urban or indoor environment. Noise and interference issues are also to be considered. The influence of antenna coupling and the charactenrstics of RF sub-systems may also have an impact on MIMO system perfonrances. For the experimeital evaluation of propagation models, two MIMO channel sotnnders are used to complement the software tools and hardwvar testbeds pertamining to the PALMYRE project. These channel sounders allow the analysis of the main spatial and temporI propagation parameters of transmission chantnels. Furthermore, co siinulation between the digital and analog parts of a system is a way to investigate in detail the effect of the RF components on the overall perfonnances [1]. It enables the imodelling of a complete communication system using die most appropriate RF analog or behavioural models [2].

This project involves the design and implementation of a platform for radio system development and evaluation. This platform allowvs a modular desigzn of systems with reconfigurable, interchangeable and reprogrammable software and hardware components. The objective is to develop and to test radio communication techniques and electronic systems in realistic conditions. A robust serial transmission system, involving digital and radiofrequency parts, has been developed to be used as a test reference for the platform. In this paper, the NIIMO 4X4 compliant radiofrequency circuit and its implementation are described. Lastly, simulation results and validation measurements are presented.

I.

INTRODUCTION

The PALMYRE project is structured into sub-projects addressing specific researh studies such as development of aMlog radioofeqiency componWnts and antennas, digital architectures for signl processing, study of new emergent modulation techniques (Multi Carrier - Code Division Multiple Access), spatio temporal coding and chaatenrization and modeling of radio propagation channels. The PALMYRE platfoirm has been initiated by the regional etwork for information and communication technologies (R3TIC),l which is supported by the Brittany Regional Council in France. It is gathering public and pnvate laboratories in Brittany like GET/ENST Bretagne (Ecole Nationale Superienre des Telecommunications de Bretagne), IETR (Institut d'Electronique et de Teleconnnunications de Rennes) and UBS-LESTER (Laboratoire d'Electronique des Systemes Temps Redels) SUPELEC, ENSSAT, ENSIETA, SACET, THOMSON Rennes, COMSIS. This paper deals with sonme works which have been mailny camred out in ENST Bretagne. - The first step of the PALMYRE pro ect is to provide the platforn xvith a SISO (Single Input Single Output) senal transmiSsion system. This will be used as a test reference for the vanious evolutions of the platforn. It will also be utilized

0-7803-9586-7/06/$20.00 C2006 IEEE.

II. PROPAGATION MODELLING

A. Propagation Modes Propagation modelling issues related to the Palmyre platform

have been investigated with the cooperation of the University of Poitiers [3]. In most wireless communication simulations, a statistical behaviour of the Channel is considered. By contrast, the goal of the present xworks consists in modelling and clharctenrzing the channel corresponding to indoor or outdoor envirorm:ent and in evaluating its interest for wireless conumiications. These woiks rely on a 3D propagation simulation software previously developed by the University of Poitiers [4]. This software, based on an optimized 3D ry tracing techique,

907

computes the complex channel impulse response and the directions of arival of the multipaths for SISO radio links. Using this information, the SISO and MIMO channel charactenistic are computed, e.g. dispersion and coherence in the spatial, temporal and frequency domains. For MIMO systems, the software computes the impulse response matnnx and the correlation matnix of the channels. B. Proagaion simulation By using ts software, the chanmel charactenrstics for vxaious system configurations and propagation environments can be studied. For instance, a 2x2 MIMO configuration for a frequency of 5.18 GHz (H) perlan II) in the indoor enviromnent has been considered (Figure 1). The transmitter is located in the central corrdor and two receivers separated by 2 meters are placed in the room noted R. The results are illustated in Figure 2 which shows the evolution of the channel capacity according to the distance (until 3.5 X) between the two receive antemus for the tWo radio links. The distance betWeen the transmitter antemuas is constant and equal to K. The studied polarization is VV.

IV. RADIOFREQUENCY EQUIPMENT

C. General desron The radio-freqtency circuit performs the signal transposition from the low IF to the emission frequency, selected in the 5.15 - 5.35 GHz band. Other frequiencies located beyond 40 GHz should be plaaned for and this stage will then be used as the IF. The global characteristics are close to the 1EEE802. 11a system parameters (radiated power, receiver sensitivity and bandwidth) but they can easily be adapted to other needs ([5],). For this purpose, all parts are interchangeable. The chosen architecture is a double changing frequency ([6]). This structure failitAtes elimination of the image fequenlcies and provides flexibility for the choice of the low IF (2OMIz to 50MHz) and tlh channel bandwidth. The 2 local oscillators are synthesized using thed same reference frquency (1OMIHz). For Tx modules thed FT input power is defined at -16dBm. The R-F output power should be controlled over a 25dB range: form 5dBm to the saturation of the power amplifier. The output power compression point is about 25 dBm. For Rx modules, the FT gain control has more than 65dB variation (flatuess bandwidth over 10 to lOOMIHz frequeny range). The -85dBm -30dBm dynamic range have been obtained in respect of the IEEE802.1 la sensitivities. The FT output power (Rx) shotuld be remote from -6dBm to lOdBm with very good linearity (IMD greater than 25dB). Photograhs of developed hardware are given Fig. 4, Fig. 5 and Fig. 6.

III. DIGITAL TRANSMISSION SCHEME In this initial version, the transmission scheme involves a serial waveform with a QPSK modulation. Thie modulation and demodulation functions are perffried completely numencally. The system will operate in a line-of-sight configuration, so the defects of synchronizations are assumed to be limited to the drift of the oscillators and no eqalizer is implemented. To pemnnit futue evolutions, a modular concept was selected as illustmted in Fig 3. But although separated, algorithms present total coherence.

D. Antnmas

Dipoles and patch antennas as shown in Fig. 7 are used for our MIMO 4x4 solution. Many scenarios for transmission tests or channel characterization are under investigation. Flexibility is the property we wat most for ths platform: polarisation, diagram and aperture, linear or surface netxork, distance between antennas

Distributed between the transnitter and the receiver, the adapted filter is a raised cosine with a roll-off equal to 0 5. This value allows for a certain inaccuracy over symbol timing recovery and limits the number of coefficients of the fimite impulse response filter (FIR). At the transmitter, this filter includes 17 coefficients, on the basis of four coefficients to a symbol. It is located on each line behind the over-sampling fLnction and it ensures interpolation. At the receiveie the same filter operates with 33 coefficients on the basis of eight coefficients to a symbol and then provides the eight samples per symbol for the timing correction. In order to simplify frequency transposition, the sampling rate selected is equal to four times that of the digital carrier. Thus, we avoid the use of multiplications and sine tables. This sampling rate is also equal to four timnes that of the symbol rate. Consequently, a bit rate of 18 Mb/s is achieved with a symbol rate, a sampling mte and a digital carrier respectively equal to 36 Mbauds, 36 MHz and 9 MHIz. Finally, tianks to the intemal transposition device enclosed in the DAC (Digital to Analogue Converter), the low intermediate frequency obtained is 45 MIHz.

E Remote control The remote control of the radiofrequency equipment is ensured by using small boards built arotund a MC68HC908GZl6 nmcrocontroller. They receive their instnictions from the host system using the same 1 Mbit/s CAN (Controller Aera Network Iso 11898) type serial link The protected protocol of this multi-master type bus directed short message binngs substantial modulanrty to the scheme and allows communication between boards too. Thus, for a SISO version, two boards ensure remote control of the local oscillators in the transmitter a:nd the receiver. The third controls the gain and the phase in the transmitter and the fourth manages automatic gain control in the receiver.

V. SIMULATION AND RESULTS

A. Snbol timing recory andcarrierpphase reovery simulation The digital transmission scheme described in III. has been developed and simulated with CoCentic System Studio,

908

VI. CONCLUSION We have presented the robust digital transmission scheme developed for the PALMYRE platform. Implementation of digital parts and the radiofreqency circuits have been described and simulation results and validation measures have been shown. Further work will include an extension of the transmission scheme to a MIMO 4x4 system, in accordance with the PALMYRE objectives This evolution, currently in progress, is facilitated by the modularity of the implementation. Furthermore, the MIMO specifications have been taken into account sine the start of the project. ACKNOWLEDGMENT

edited by Synopsys. The first version has been impletmented in floating point. At this stage particular attention has been paid to synchromization algorithms. Subsequently, the fixed point version of the algorithls has been tested and compared to the floating point version. We assume that the channel generats a fixed frequency offset and of on phase, independently, timing Af / N 0.000015%, coespOndi to an offet of 150Hz on a l0MHz firquency. This frequency offset is applied to the over sampled sigal at 8 samples per symbol, provided by the emitter. In order to observe the system at working point, we choose a SNR of 1 IdB yielding a BER of 2.10-4 in a theoretical QPSK configuration. We select BlT 5.10-3 for the symbol timing recoveir loop and BlT10-3 for the carrer phase recovery loop. Because of the lack of precision in the timing correction, the estimated timing error evolves by visible steps [7]. Conversely, the estimated phase error steadily follows the evolution of the true phase error.

The authors wants to thank Yaniick Chartois and Frnois Le Pennec from GET/ENST Bretagne, Rodolphe Vauzelle from the University of Poitiers, Ronan Cosquer, Ghais El Zein, Hanna Farhat from INSA Rennes, Christian Brousseau from University of Rennes and the enterprise COMSYS for their collaboration within the Palmyre projet.

The optimization of the loop parameters, typically loop bandwidth, has been performed through several simulations.

On-board optimization will be held on the platform thanks to this parameters set.

REFERENCES [1]

F. Escarieu, Y. Pousset, R.

Vauzelle,

L. Aveneau: ( A 3D simnulation

software for outdoor ond indoor rodio mobile chonnel >>, pp213-216, EOCWT'2001, Sept 2001, London, ULK. [2] Y. Le Roux, Y. Chartois. P.Lassudrie-Duchesne, F. Le Pennec, R. Vauzelle, R. Cosquer, G. El Zein, H. Farhat, C. Brousseau,

B. Digital transmission results In order to be able to validate the vanrous inodules of the digital sysstem tlie low intennediate frequey sig nal generated by the transmitter is injected at the entry of the receiver. The intermediate results are memorized and can be displayed on control screens. Thus the eye and constellation diagram given in Fig. 8 a) were obtained with perfectly synchronized eiittter and receiver frequency, and with manual ajustinent The error rate Ineasured under these conditions for a 14 dB signal-to-noise ratio is approximnately equal to 10-6. Constellations presented in FPig. 8 b) and c) were obtained respectively after tiImnng and carrier recovery with an offset of

« Propagation and system modelling for the Palmyre MIMO platform >>, ECPS 2005, Brest, France, 15-18 march 2005. [3] Y. Chartois, ( Etude parametrique avancee de canaux SISO et MIMO en environnements complexes: application au systeme HiperLAN/2 >i Ph. D. Thesis, GET / ENST Bretagne, 08 Dec 2005. [4] Y. Huang. S. Saadaoui, Y. Chartois, F. Le Pennec, Ph. Rostaing. R. Vauzelle, Y. Pousset a Antenna coupling in ray tracitg based MIMCO channel models is, ECPS 2005, Brest, France, 15-18 march 2005. [5] Lavton, G. "Is MIMO the futture of Wireless communications ?" Computer, Volume: 37, Issuie: 7, July 2004 Pages:20 - 22 [6] Shigeru HIRRA, Masaaki ISHIDA, Takaya KITAHARA, Testsuiya Y AMAMOTO "RF Module LUsing MCM-L and BGA Technology for 5GHz WLAN Application" 33rd Europea Microw ave Conference Munich 2003: pp 895-898 [7] J. Menard, D. Bourreau, E. Daniiel, Y. Le Roux, "PALMYRE serial transmission system for radio channels, insvestigations for a MIMO system", ECPS 2005, Brest, France, 15-18 march 2005.

Af /RFo0.015o.

C. Radioequenc measurement results Using the R&S SMIQ geierator and a FSQ26 signal analyzer, the first Rx prototype ias been tested. As the serial transmission envisaged, we used a QPSK signal at 18Mb/s and a 16dB CNR at 45Mhz IF frequency. The experimental setup is slhown fig. 9 and the measured eye diagram is in close

agreement with the MATLAB simulation results [7]. Finally, fi. 10 shows the SISO and MIMO experimental setups with video transmission.

909

trnsn-itter E =

m

Figure 1. Studied indoor environment (from [2], software from University of Poitiers, SIC laboratory)

Figure 2: Capacity variations according to the distance between the two receiver antenmas (from [2])

Figure 3 : Structure of the digital transmission system

910

Figure 4 Local oscillators division for MIMO 4x4 (Rx or Tx Rak)

Fig. 5 MIMO 4x Tx platfonn

Fig. 8. a)

Sy)nchronized oscillators

Fig. 7 patch anitenna in the 5 GHz band

c) Camrrer recovery

b) Timing recovery

911

Figre 9: Eye diagranm measurement setup

1. iuKvicu v llt1LL1ai 1U1s witIi LtK JrAILA I JDId pliUonLL left: 2x2 MIMO experiment (COMSYS) - right: SISO h:igh data rate transmission

riguwi

912

Suggest Documents