Abstract
Wireless sensor networks have seen a wealth of research efforts and practical implementations in recent years. With the advance of CMOS cameras and microphones, wireless multimedia sensor networks are proposed to be a promising solution to a variety of applications in surveillance and intrusion detection, smart traffic control, automated health care, environmental monitoring, and so forth. It is composed of wirelessly connected small devices, which are able to capture multimedia data from the surrounding environment, besides its capabilities of processing the multimedia data using the embedded CPU and transmitting data to the sink through wireless communication. In this chapter, we firstly address features of WMSNs and challenges facing the realization of WMSNs by introducing two experimental applications of WMSNs. As energy efficient routing and distributed source coding are two critical components for the success of WMSNs, we discuss existing work in these two areas subsequently, in order to reveal details about the challenges and potential solutions to the problems posed by WMSNs. At the end of this chapter, open problems on cross-layer design, quality of information and privacy and security are briefly discussed.
TopRegarded as one of the most promising technologies to extend human presence to interested physical environments, wireless sensor networks (WSNs) have been the subject of a surge of research activities in recent years. Moreover, low-cost, miniaturized image sensors have enabled WSNs to capture multimedia information from the environment. This in turn has introduced a set of new challenges for wireless multimedia sensor networks (WMSNs). Among them, the most critical problems lie on supporting multimedia sensing streaming capability and prolonging the network lifetime using conventional battery and low data rate radio interface. Multimedia sensor networks also demand novel coding scheme, in-network processing strategy, and routing technique in relieving the network bottleneck. In this section we will discuss the key aspects of WMSNs and describe certain experimental platforms.
Key Aspects Of WMSNS
Key aspects of WMSNs include power consumption, bandwidth limits, unreliable wireless link, QoS requirements, multimedia in-network processing, and multimedia source coding. Unique challenges in WMSNs are mainly due to the inherent conflict between the abundant data generated from the sensor nodes and the constraints on bandwidth, power supply and computing capability.
QoS Requirement
Multimedia data delivered by WMSNs present similar QoS requirements as in other multimedia applications. Specifically, the streaming of multimedia data requires consistent connection and has little tolerance toward out-of-order transmission. QoS requirement poses additional challenges to WMSNs since it is mostly application-specific, involving tradeoffs among various metrics like energy consumption, estimation distortion, transmission delay, and so forth.
Bandwidth Demand
Multimedia data usually consumes bandwidth that is orders of magnitude higher than that is supported by normal WSNs. High data rates of the sensor nodes make it almost impossible for the network to transmit the raw data from one node to another. Besides the bandwidth demand, transmitting large amount of multimedia data leads to significant energy consumption compared to transmissions among the traditional scalar sensor nodes.
In-Network Processing
In-network processing has been proposed in WSNs as a powerful methodology to save energy. The application-specific parameter or objectives could be estimated in a distributed way through in-network processing. The most common example is taking average of the sensor data along the route from nodes to the sink. For example, in surveillance WMSNs, the nodes could exchange result of intrusion detection instead of reporting the images or streaming the videos to the sink. As energy efficiency becomes more critical in WMSNs due to high data rate, in-network processing plays a critical role in avoiding transmissions of raw sensor data.
Distributed Source Coding
Compression techniques are indispensable to multimedia applications given abundant redundancy within one frame or among successive frames. However, it is not practical to integrate compression techniques directly in WMSNs because compression algorithms may be too complicated for the resource constrained node platform: distributed source coding, instead, may serve as a desired solution for WMSNs as the complexity of encoder is shifted to decoder. Therefore, distributed source coding provides WMSNs a solution to reduce communication costs by reducing the redundancy of the multimedia sensor data.
Existing sensor node platforms mostly are in the area of video sensor networks, for example surveillance applications. Main concern of these platforms is the applicable system architecture for WMSNs in addition to the efforts of alleviating the conflict between the resource constraints and the abundant sensor data generated by the nodes.
Key Terms in this Chapter
Compress Sensing: Selection of sensor nodes and samples to be taken by the sensor nodes benefits WSNs in terms of energy saving.
Wireless Sensor Networks: Networks composed of small devices equipped with sensors, embedded processor, and radio. The devices are usually called sensor nodes, which are able to sense the physical environment, process the sensing data and communicate with each other through wireless links.
Adaptive Routing: Routes from the source nodes to the sink are determined adaptively according to dynamics of the nodes in the network. In particular, energy efficient routing could be adaptive considering the fusion costs of the nodes.
Wireless Multimedia Sensor Networks: A type of wireless sensor networks. The sensor nodes are capable of capture multimedia data about the environment. Besides the scalar sensor data, the nodes may generate video data, image data, or audio data about the monitored area.
Target Tracking: Moving object in the monitored area is detected by the sensor nodes. Its trajectory will further be predicted by the sensor nodes based on their observations on the movement of the target.
In-Network Processing: Sensor nodes of WSNs or WMSNs conduct aggregation on the sensor data collected from its neighbors or its own sensor data.
Distributed Entropy Coding: Encoding rate for correlated sources remains the same even the side information is not available at the encoder but the decoder. Slepian-Wolf theorem guarantees the lossless encoding rate, while Wyner-Ziv theorem extends the conclusion to lossy encoding.