Research On Distributed Information Processing In Wireless Multimedia Sensor Networks

Wireless Multimedia Sensor Networks (WMSNs), developed from traditional Wireless Sensor Networks(WSNs), is a novel sensor network integrating sensation functions of multimedia streaming (e.g. audio, video, image). Compared with traditional WSNs which can only measure simple scalar physical phenomena such as temperature, pressure, humidity, or location of objects, Wireless Multimedia Sensor Networks are able to ubiquitously retrieve multimedia content such as video ,audio streams and still images. Potential applications of Wireless Multimedia Sensor Networks span a wide spectrum from military to industrial, from commercial to environmental monitoring.Wireless Multimedia Sensor Networks have broad application foreground in military, especially in constructing the integrative information system in battle field .Supported by the 863 National High Technology Research and Development Program "xxx image data compression technology"(No.2006AA701121), this dissertation concentrates its attention on the distributed information processing in Wireless Multimedia Sensor Networks. The main contents and contributions of this dissertation are as follows:1. The state of art in the key techniques of WMSNs, including node systems, MAC protocols, routing protocols, multimedia signal processing, is surveyed, and open research issues are discussed in detail.2. A distributed image compression algorithm based on lapped biorthogonal transform (LBT) in the cluster-based Wireless Multimedia Sensor Networks is proposed. In this algorithm, the image is firstly partitioned into several tiles (blocks) that consist of eight rows, and then each block is sent to a node to do 1D LBT. In order to prevent blocking artifacts, this algorithm uses nodes in two levels to enhance the correlation of adjacent blocks. A simplified zerotree coding algorithms is used in "coding nodes" of two levels by turns to reduce the complexity of coding. Simulation results show that this distributed algorithm can prolong the lifetime of the WMSNs compared with "centralized" compression algorithm under a specific image quality requirement. The "node data stream" conception is proposed in order to analyse the distributed algorithm, this new conception and method will be useful in analyzing the data of sensor networks consist of thousands of sensor nodes on a large scale.3. The theory of Distributed Source Coding(DSC) is studied, and two predominant practical coding methods—Distributed source coding using syndromes (DISCUS) and turbo-DSC method based on punctured turbo codes are discussed. Then, this dissertation simulates and tests the turbo-DSC system using binary and Gaussian random sequences and finally gets a lot of useful results.4. The use of practical DSC coding methods in Wireless Multimedia Sensor Networks is discussed and two coding frameworks are proposed. One of the frameworks focuses on image distributed coding in camera array and the other on video coding in single sensor node. In the second framework, a scheme of translation-domain intra-frame encoder and inter-frame decoder is used. Thanks to low complexity and the zerotree structure in LBT coefficient and the excellent performance of trubo code, LBT and turbo-DSC method are used in the two proposed frameworks.