CS Research On Source And Channel Coding In WSN

With the continuous expansion of the wireless sensor network (WSN) applications, the promotions and requirements to information technology will be showing up. In reality, WSN has been deep into all aspects of society and plays a key role in many areas, such as:aerospace, security, universe detection, emergency command, etc. which has been gradually turning into the foundation of human information society. The perception object of the classical WSN are mainly confined to low-dimensional data parameters, such as temperature, humidity, soil composition, or speed and direction of moving objects. With the widespread of WSN applications, the amount and dimensions of collected data increased rapidly, the inherent characteristics of the classical WSN cause the energy of front node and calculation conditions limited. A new front-end encoding method with small coding amount and low computational complexity for WSN is very urgent, compressed sensing theory, just to meet this demand. Compressed sensing is not based on the characteristics of signal bandwidth encoding, but according to the signal "inner dimension" encoding, which breaks the Nyquist rate that Nyquist sampling theorem required. Signal is encoded by the linear projection, which can cause the complexity of computational encoding much lower than that of Nyquist sampling theorem.The main research works and contributions of this thesis are outlined as follows.1. Using deterministic Toeplitz sensing matrix, source coding and channel error correction coding are combined into a linear projection, which can reduce the encoding complexity. Not only is the trade-off between source coding and channel coding taken into account, but also different methods are proposed corresponding to the characteristics of the two channels, respectively. In this thesis, compressed sensing Gauss Channel method and compressed sensing Rayleigh Channel method are proposed to process Gauss channel and Rayleigh channel respectively. Simulation results show that the channel noise, fading, delay and other factors don’t affect the probability of successful recovery significantly. Under the same conditions of coded information amount, due to consideration of some practical factors in CSGC and CSRC coding method, the corresponding recovery probability is slightly lower than that of traditional compressed sensing theory.2. A new distributed source-channel coding method is proposed based on distributed source-channel joint coding in compressed sensing and "Real" Slepian-Wolf Codes (RSWC) coding methods, which not only take advantage of SW coding method that that doesn’t require cooperative communication, but also reducing the encoding computational complexity. Under the same measurement value conditions, utilizing of spatial correlation of adjacent data can improve the performance of recovered signal to noise ratio effectively showed in simulation results.3. A cooperative approach based on compressed sensing theory is studied. As the measurement matrix could not be replaced, this thesis use projection sequences to measure the interesting object. At the back-end recovering, the measurement matrix and its corresponding node are determined according to the sparsity of recovered data. As the weak correlation between the signal itself and measured value, the feasibility of the method is verified by theoretical analysis and formula derivation4. The method of network data acquisition and data transfer by refinance approximation is studied. Each measurement node active involved can be determined by the scale of the object and accuracy requirements at back-end. Simulation results and the theoretical proof of the data collection method by refinance approximation can effectively recover the original data.