Research On Rateless Codes Of Communications In Wireless Sensor Network

Channel coding is an efficient method to resist fading effect, which can improve the transmission reliability of communication in wireless sensor networks. As a new type of low-density rateless channel coding,rateless codes can automatically match the transmission rate of the link without knowing the channel state condition, which is especially suitable for the communication in wireless sensor networks. On the other hand,relay technology is the key to realize mutual communication between nodes since there is no central node in wireless sensor networks. Through mutual cooperation between nodes, relay technology not only expands the scope of communication between nodes, but also improves the communication efficiency of the entire communication system. Rateless codes and relay technology has been a widespread concern of scholars. In this thesis, on the basis of study on the channel coding theory and related disciplines, we investigate the design of degree distribution and the application of rateless codes in wireless sensor networks. While keeping the decoding reliability high, we manage to improve the efficiency of encoding, reduce the redundancy and complexity of decoding.Additionally, based on the related theoretical studies of wireless sensor networks, we study the rateless codes in the relay communication system.We focus on the study of partial recovery and distributed rateless codes.Our work presents the following contributions:(1) In the belief propagation decoding process, on the one hand, the release probability of each encoded symbol directly affects the change of ripple size; on the other hand, ripple size revolution affects the release probability of each encoded symbol. Therefore, it is difficult to calculate the accurate degree distribution directly according to the ripple size revolution. To solve this problem, a precise calculation method is proposed. Our method can effectively solve the interaction between the probability that the released input symbols have been in the ripple and the ripple size revolution in each decoding step. Since the problem that the ripple size tends to be zero in the decoding process has been solved by our proposed method, we can possible to analyze ripple size revolution independently and obtain a more accurate degree distribution.(2) Based on the contribution (1), a rateless codes with decreasing ripple size and feedback has been proposed. Ripple size revolution is accurately simulated by a one-dimensional random walk model.Furthermore, ripple size revolution is fitted by the polynomial fitting method. Finally, a rateless codes with decreasing ripple size is designed.With the introduction of feedback, ripple size revolution caused by the number of input symbols that have been released is analyzed directly. On the basis of the contribution (1), the degree distribution is adjusted accurately. Furthermore, we keep the adjusted degree distribution diverse.Simulations show that our proposed rateless codes with decreasing ripple size not only reduces the redundancy, but also reduces the average degree of encoded symbols, while our proposed rateless codes with decreasing ripple size and feedback can effectively reduce the uncertainty of the source and improve the performance of the transmission system.(3) To improve the intermediate performance of rateless codes, a partial recovery rateless codes is proposed. Since many lots of encoded symbols are released in each encoding step in partial recovery rateless codes, traditional analytical method is no longer applicable. An iterative and small degree first analysis method is proposed to accurately analyze the decoding process. In this analysis method, we assume that only one input symbol is released in each decoding step. The number of encoded symbols with respect to degree required in each decoding step is accurately calculated by an iterative optimization algorithm. Through simulations, we demonstrate that our partial recovery and multi-stage partial recovery rateless codes has a high intermediate symbol recovery rate, which improve the transmission efficiency in the networks of single source multiple relay single destination.(4) A general distributed rateless codes is proposed to improve the transmission efficiency in the network of multiple source single relay single destination. Different from the existing deconvolution based method, our codes is designed by a heuristic iterative optimization approach to minimize the redundancy. Degree distributions in sources and relay are adjusted by using the Jacobian matrix. On the basis of the contribution (2), the ripple size revolutions in sources and the relay are adjusted directly by using the Jacobian matrix. Finally, a general distributed rateless codes with decreasing ripple size is designed.Experimental results show that our codes can effectively improve the transmission speed in the multi-access relay network.