Chinese Journal of Electronics Vol.25, No.2, Mar. 2016

Efficient Image Transmission Schemes over Zigbee-Based Image Sensor Networks∗ TAO Dan1,2,3, YANG Guangwei1 , CHEN Houjin1 , WU Hao3 and LIU Pingping2,4 (1. School of Electronic and Information Engineering, Beijing Jiaotong University, Beijing 100044, China) (2. Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education, Jilin University, Changchun 130012, China) (3. State Key Laboratory of Rail Traffic Control and Safety, Beijing Jiaotong University, Beijing 100044, China) (4. College of Computer Science and Technology, Jilin University, Changchun 130012, China) Abstract — Image transmission is one of the biggest challenges in wireless sensor networks because of the limited resource on sensor nodes. We proposed two image transmission schemes driven by reliability and real time considerations in order to transfer JPEG images over Zigbee-based sensor networks. By adding two bytes counter in the header of data packet, we can easily solve the repeated data reception problem caused by retransmission mechanism in traditional Zigbees network layer. We proposed an efficient retransmission and acknowledgment mechanism in Zigbees application layer. By classifying different data reception response events, we can provide data packets with differential responses and ensure that image packets can be transferred quickly even with large maximum number of retransmission. Practical results show the effectiveness of our solutions to make image transmission over Zigbee-based sensor networks efficient. Key words — Image sensor networks, Image transmission, Retransmission and acknowledgment mechanism, Zigbee.

I. Introduction With the recent advances in Micro electro-mechanical, systems (MEMS), CMOS cameras, which may ubiquitously capture multimedia content from the environment, wireless image sensor networks have been proposed and raised much attention of industry and academic communities[1−3] . Image communication has become more imperative due to the ubiquitous proliferation

of multimedia applications over wireless sensor networks, such as video surveillance, environmental monitoring, target tracking, traffic control and smart home[4] . Achieving reliable transmission with quick speed in wireless image sensor networks is still a challenging task, for two reasons. First, the wireless links are unreliable which will cause packets loss in transmission in an unpredictable manner. Second, the memory of sensor nodes is limited so that it is very easy to overflow and cause more packets loss[5] . ZigBee is a low-cost, low-power, wireless mesh network standard. The low cost allows the technology to be widely deployed in wireless control and monitoring applications. Low power usage allows longer life with smaller batteries. Zigbee has become the best choice of wireless image sensor networks[6]. Hence, it is necessary to investigate efficient image transmission solutions over Zigbee-based network. The existing Zigbee-based data transmission theories and algorithms have focused on energy consumption, data reliability, real-time communication, routing and QoS. Currently, several solutions have been proposed to ensure efficient data transmission in wireless sensor networks, such as priority queue, routing optimization, error protection, retransmission and acknowledgment mechanism, multipath transmission, multiple code transmission, and so on[7] . Retransmission and acknowledgment mechanism has played an important role in solving reliable data

∗ Manuscript Received Mar. 13, 2014; Accepted Apr. 3, 2015. This work is supported by the National Natural Science Foundation of China (No.61271305, No.61202431), Beijing Higher Education Young Elite Teacher Project (No.YETP0535), the State Key Laboratory of Rail Traffic Control and Safety (No.RCS2012K008), Beijing Jiaotong University, China Postdoctoral Science Foundation (No.2015M571363), Jilin Provincial Natural Science Funds for Young Scholar (No.20150520063JH), the Open Project Program of Key Laboratory of Symbolic Computation and Knowledge Engineering of Ministry of Education, Jilin University, the Open Projects of Beijing Key Laboratory of Intelligent Telecommunications Software and Multimedia, Beijing University of Posts and Telecommunications, and the Scientific Research Foundation for the Returned Overseas Chinese Scholars, State Education Ministry. c 2016 Chinese Institute of Electronics. DOI:10.1049/cje.2016.03.014 

Efficient Image Transmission Schemes over Zigbee-Based Image Sensor Networks

transmission[8]. Usually, when transmitting image packets over error-prone wireless links, packets can be corrupted, and require retransmission. And, correct reception is acknowledged. In this paper, we present a new retransmission and acknowledgment mechanism in order to make reliable and quick JPEG images transmission in Zigbeebased sensor networks.

II. Related Work Previous researches have been reported in the literatures regarding efficient image transmission by using various solutions in the context of wireless image sensor networks. Authors of Ref.[9] tested JPEG and JPEG2000 images over ZigBee network and shown that JPEG2000 images encoded into multiple quality layers are more error-resilient, while maintaining a high Peak signal-tonoise ratio (PSNR). Unlike traditional approaches where data is sent over the paths regardless of its importance at the destination, authors of Ref.[10] proposed a networkadaptive transmission mechanism that decomposes the source bit stream into segments of unequal importance and reserves the most reliable paths to transmit the segments with the highest importance. Aghdasi et. al.[11] proposed a DCT block based image compression using adaptive priority to categorize network packets. To increase the transmission rate and decrease the energy consumed in retransmission tries, the UDP protocol was used in sending image data packets. As discussed earlier, blind dropping of packets in UDP containing highly compressed image data may adversely affect the quality of transmission[1] . Authors of Ref.[12] proposed adaptive unequal error protection scheme for JPEG2000 image transmission through the IEEE 802.15.4. This method assign adaptive code rate

285

resulting from packet error rate. Wang et. al.[13] proposed a simple best effort protocol to transmit JPEG images over a ZigBee network. Considering that image quality can be degraded by image distortion during transmission due to packet losses and poor link quality, a Canny edge detector algorithm was used to recognize a distorted image, and then the system retransmitted the image automatically. Zhang et. al.[14] proposed a dynamic retransmission algorithm, which set the maximum number of retransmission for each task according to different reliability requirements. However, this method designed for network layer, it didnt consider the time consumption caused by retransmission and might result in the confusion of data transmission.

III. Efficient Image Transmission Scheme In the process of image transmission, micro controller module of image sensor nodes sends data packet with the size of 45 bytes (includes 10-byte packet information and 35-byte image data) at a specified time interval, wireless module stores this data packet and then sends to sink node. Assume that the dynamic storage method with L2 cache provided by Zigbee protocol is adopted. When the size of data packet received by wireless module is large (e.g. 45 bytes), it can be dynamically stored by Zigbee protocol stack. These data packets will be sent with unfixed size. That is, the amount of data in L1 cache and L2 cache are random variables. If data is lost, it is hard to identify which packet the data in L2 cache belongs to, let alone the ownership of data. Fig.1(a) shows image data packets sent by dynamic memory obtained by Packet Sniffer.

Fig. 1. Image data packets sent by two different memories

286

Chinese Journal of Electronics

Given this, in this paper, wireless module of image sensor nodes adopts static storage method, as shown in Fig.1(b). The storage space is big enough to store 5-6 data packets, in order to ensure each data packet can be sent in real time. 1. Image transmission scheme in network layer Zigbees network layer itself has data transmission reliability protection mechanism; that is retransmission mechanism. Experiments have shown that the retransmission mechanism should be enforced, and the default value of maximum number of retransmission is 8. Considering the complexity of wireless channel, the following situation often occurs. After receiving data packet from image sensor node, sink node will reply an ACK. If there is a channel or other interference, ACK will be lost. In this case, image sensor node will continue to retransmit the previous data packet, and which results in that sink node receive redundant data. This is not allowed. Particularly, for JPEG image, its format is very sensitive with packet head, quantization table, Huffman table, and has a critical order requirement[15] . The loss, repetition and mis-order of data packet will lead to the wrong image restoration. Motivated by this, we consider to add two bytes counter in the header of data packet. This counter, that is the ID of data packet, can be used to represent the sequence of data packet in the transmission process. The format of data packet can be shown in Table 1. Table 1. The format of image data packet Packet ID Packet size Image data amount ID code 2 bytes 2 bytes Packet size-6 bytes 2bytes

After receiving a data packet, sink node firstly extract Packet ID in the header of current packet and then judge whether this data packet has been received or not. If yes, the data packet will be discarded directly. In this way, we can solve the problem of repeated data reception caused by retransmission mechanism in traditional Zigbees network layer. The detailed process can be illustrated in Fig.2. 2. Image transmission scheme in application layer Retransmission mechanism in Zigbees network layer has its vulnerability. Specifically, as the time interval of data packet transmission is less than the maximum total retransmission time, data aliasing will occur. Data aliasing cannot be allowed for image or video, which has a critical requirement for data reception order. For example, if the time interval of data packet transmission is 0.5s and the time interval of 12 times retransmission is 0.69s, then there exists data aliasing in the process of data transmission. As shown in Fig.3, the 12th packet is retransmitted before the 11th packet, which may cause irreversible change on image restoration. Given the fragility

2016

of retransmission mechanism of traditional Zigbees network layer, we propose a feasible retransmission and acknowledgement mechanism in application layer to achieve reliable and quick data transmission.

Fig. 2. Eliminate repeated data reception caused by retransmission mechanism

Retransmission and acknowledgement mechanism proposed is designed for reliable image transmission, which is sensitive with data packet order and doesnt allow data duplication. In our scheme, we classify three kinds of different data reception response events: ACK, Busy and Duplication, and provide data packets differential responses in the stage of data transmission. In this way, image data can be transmitted quickly even with large maximum number of retransmission, and thus we can guarantee efficient image transmission in wireless channel.

Fig. 3. Data aliasing in Zigbees network layer

The implementation includes two parts: Zigbees application layer design of image sensor node and sink node. i) Zigbees application layer design of image sensor node

Efficient Image Transmission Schemes over Zigbee-Based Image Sensor Networks

Step 1: Firstly, two kinds of task events need be added, which are data retransmission event and ACK retransmission event. Step 2: Data can be retransmitted by triggering data retransmission event. The number of retransmission can be set by corresponding demand. Step 3: Data can be marked using ID. In this way, the repeated data cannot be received by sink node. Step 4: ACK mechanism can be added on image sensor node. ii) Zigbees application layer design of sink node Step 1: ACK mechanism can be added on sink node. Step 2: Since repeated data may be received due to retransmission mechanism, data filtering mechanism need be set. Step 3: Once having received image data packet or is printing serial data without receiving image data, sink node will send the corresponding response instruction to image sensor node. If the response instruction cannot be sent successfully, then the data reception response event should be triggered, and response instruction will be retransmitted. The detailed process of data transmission in Zigbees application layer can be discussed as follows. After Zigbee protocol monitors data transmission event by polling, the data transmission task will be executed. Then, data packet should be judged whether it has been received. If no, image sensor node sends ACK; otherwise, silently discard and send Duplication response instruction. When image sensor node sends data to sink node, if sink node is operating via serial, then it cannot receive data and send Busy response instruction. If sink node cannot send ACK successfully, then data reception response event will be triggered to send back ACK until it succeeds. Once image sensor node receives ACK from sink node or duplication data response instruction, next data packet will be sent. Particularly, when image sensor node receives Busy from sink node or doesnt receive any response instruction for a certain period, it will retransmit data packet until receiving ACK from sink node or approaching the pre-set maximum number of retransmission. The workflow chart of data transmission in Zigbees application layer is shown in Fig.4.

IV. Results and Discussion We have designed and implemented hardware testbed[7] to evaluate the performance of two image transmission schemes proposed. Image sensor node consists of four components: STM32 micro controller, image sensor module integrated with 0V7640 and 0V528, wireless transmission module CC2430, and 5V battery. The DEMO of image sensor node can be shown in Fig.5. In

287

our experiment, the size of image captured and transmitted is 160×128, and the size of image packet transmitted for each time is 45 bytes.

Fig. 4. The workflow chart of data transmission in Zigbees application layer

Fig. 5. DEMO of wireless image sensor node

Sink node is made up of CC2430 kernel components. The communication between image sensor node and sink node adopts Zigbee protocol. A simple application runs on an image sensor node and issues a request for an image transfer to sink node in the network. The composition of testbed can be depicted in Fig.6. 1. Effect on total retransmission time As packet transmission consumes time, we firstly evaluate the relationship between the total retransmission time and packet retransmission for different maximum numbers of retransmission (4, 8, 12, 16, 20) using these two solutions to verify real-time performance. In order to eliminate accidental error, for each maximum number of retransmission, each total retransmission time shown

Chinese Journal of Electronics

288

here is the statistical average of 20 measurements. From the curves shown in Fig.7, we can get that the total retransmission time scales grows linearly with the increase of the maximum number of retransmission. However, the total retransmission time consumed in application layer is significantly less than that in network layer. That is, we can reduce the total retransmission time even if the maximum number of retransmission is relatively large. For example, when the maximum number of retransmission values 12, the total retransmission time consumed in application layer and network layer are 0.121s and 0.69s, respectively. In this way, we can meet real-time transmission while ensuring transmission reliability and decreasing the whole networks energy consumption.

2016

retransmission is fixed, the packet loss rate will increase quickly with the increase of transmission distance. When transmission distance is no more than 12m, the packet loss rate is zero for different maximum number of retransmission. When transmission distance approaches 40m, the packet loss rate for 8 times retransmission (about 10.2%) is about triple as large as that for 16 times retransmission (about 3.4%).

Fig. 6. The composition of testbed

Fig. 8. Comparison of the packet loss rate between two schemes proposed with different maximum numbers of retransmission

Fig. 7. Comparison of the total retransmission time between two schemes proposed

2. Effect on packet loss rate Then, we discuss the relationship between packet loss rate and transmission distance to show their data transmission performances. Fig.8 gives the effect on packet loss rate with different maximum numbers of retransmission (8, 12, 16, 20) and transmission distance (from 0m to 40m, its interval is 4m). We make the following observations from the results. The difference of packet loss rate using two solutions is fairly tiny, it is mainly because that the maximum number of retransmission adopted in network layer and application layer is same. When the maximum number of

In Fig.9, when the transmission distance is fixed, we observe the relationship between packet loss rate and the maximum number of retransmission by using these two image transmission schemes. The transmission distance has a great effect on the packet loss rate. As transmission distance decreases, the packet loss rate will drop obviously. For instance, when transmission distance are 12m, 24m and 36m respectively, for 8 times retransmission, the packet loss rate are 9.17%, 3.78% and 0% respectively.

V. Conclusions The paper explores image transmission in Zigbeebased sensor networks. In order to guarantee reliable and real-time transmission over wireless links, we present a simple and feasible data retransmission and acknowledgment mechanism in Zigbees network layer and application layer, respectively. Experimental results verify the relationships among the total retransmission time, packet loss rate and the maximum number of retransmission using these two solutions proposed.

Efficient Image Transmission Schemes over Zigbee-Based Image Sensor Networks

289

[12] K. Wang, S. Lee, B. Kim, et al., “Robust JPEG2000 Image Transmission over IEEE 802.15.4.”, The 4th IEEE International Symposium on Electronic Design, Test and Applications, pp.253–257, 2008. [13] Z. Wang, X. Liu, L. Huang, et al., “Image transmission over Zigbee-based wireless sensor networks”, Sensor Letters, Vol.10, No.1-2, pp.205–212, 2012. [14] Z. Zhang, H. Yuan and F. Yu, “A dynamic retransmission algorithm for wireless sensor networks”, Chinese Journal of Sensors and Actuators, Vol.26, No.7, pp.1019–1024, 2013. [15] H. Wang, S. Kwong and C. Kok, “An efficient mode decision algorithm for H.264/AVC encoding optimization”, IEEE Transactions on Multimedia, Vol.9, No.4, pp.882–888, 2007.

Fig. 9. Comparison of the packet loss rate between two schemes proposed with different transmission distances

References [1] H. Ma and Y. Liu, “Correlation based video processing in video sensor networks”, 2005 International Conference on Wireless Networks, Communications and Mobile Computing, Vol.2, pp.987–992, 2005. [2] L. Liu, Y. Song, H. Zhang, et al., “Physarum optimization: A biology-inspired algorithm for the steiner tree problem in networks”, IEEE Transactions on Computers, Vol.64, No.3, pp.819–832, 2015. [3] Y. Song, L. Liu, H. Ma, et al., “A biology-based algorithm to minimal exposure problem of wireless sensor networks”, IEEE Transactions on Network and Service Management, Vol.11, No.3, pp.417–430, 2014. [4] P. Boluk, S. Baydere and A. Harmanci, “Robust image transmission over wireless sensor networks”, Journal Mobile Networks and Applications, Vol.16, No.2, pp.149–170, 2011. [5] H. Zhou, X. Guan and C. Wu, “Reliable transport with memory consideration in wireless sensor networks”, IEEE International Conference on Communications, pp.2819–2824, 2008. [6] Y. Sun, S. Tang and H. Luo, “The QoS guarantee problem for multimedia sensor networks”, Acta Electronica Sinica, Vol.40, No.4, pp.625–631, 2012. (in Chinese) [7] G. Yang, “Design and implementation of data acquisition and transmission reliability for wireless image sensor networks”, Master Thesis, Beijing Jiaotong University, China, 2014. [8] Y. Sun, H. Ma and L. Liu. “An ant-colony optimization based service aware routing algorithm for multimedia sensor networks”, Acta Electronica sinica, Vol.35, No.4, pp.705–711, 2007. (in Chinese) [9] Pekhteryev, Georgiy, Z. Sahinoglu, et al., “Image transmission over IEEE 802.15.4 and ZigBee networks”, IEEE International Symposium on Circuits and Systems, Vol.4, pp.3539– 3542, 2005. [10] Y. Charfi, N. Wakamiya and M. Murata, “Network-adaptive image and video transmission in camera-based wireless sensor networks”, First ACM/IEEE International Conference on Distributed Smart Cameras, pp.336–343, 2007. [11] H. Aghdasi, M. Abbaspour and M. Moghadam, “An energyefficient and high-quality MAC protocol for image transmission in wireless sensor networks”, The 4th IEEE International Conference on Circuits and Systems for Communications, pp.838– 842, 2008.

TAO Dan received the Ph.D. degree in School of Computer Science, Beijing University of Posts and Telecomm., Beijing, China in 2007. She currently works as an associate professor in the School of Electronic and Information Engineering, Beijing Jiaotong University, Beijing, China. Her research interests include wireless sensor network, IoT and multimedia system. (Email: [email protected]) YANG Guangwei is a M.S. candidate in the School of Electronic and Information Engineering, Beijing Jiaotong University, Beijing, China. His research interests include wireless sensor network, embedded system. (Email: [email protected])

CHEN Houjin currently works as a professor in the School of Electronic and Information Engineering, Beijing Jiaotong University, China. His main research areas include digital signal processing, medical image processing and multimedia system. (Email: [email protected])

WU Hao currently works as a professor in State Key Laboratory of Rail Traffic Control and Safety, Beijing Jiaotong University, China. Her main research areas include wireless broadband communication and railway mobile communication. (Email: [email protected])

LIU Pingping currently works as an associate professor in the College of Computer Science and Technology, Jilin University, China. Her main research areas include intelligent information processing and machine vision. (Email: [email protected])