Joint Source-channel Coding Based On Sliding-window Belief Propagation

Nowadays, the Source-Channel coding is applied to the multi-point multi-hop system frequently; it involves many hot research areas including distributed source coding. The main goal of distributed source coding is to compress the data from multiple sources that distributed in different locations and spaces, which are commonly applied to applications such as the city multipoint video monitoring system, Chang E detectors, sensor networks and so on. In many distributed coding systems, the source or channel is modeled as a random process with known parameters, which means the source or channel is stable. However, the local statistical characteristics of the source or channel are generally unknown in practical applications, i.e. the source or channel is unstable. Hence, the estimation of source parameters is the key factor that affects the performance of distributed source coding and the estimation of channel noise is an essential aspect of data transmission on the noisy channel. To solve the rate decrease problem of data transmission on the noisy channel in distributed source coding system, joint source-channel coding adjusts the allocation of source coding and channel coding appropriately. In this paper, the joint source-window belief coding algorithm based on sliding-window belief propagation is introduced to solve the above problems. The main contents of this research are as follows.(1) To accomplish the functions of data compress and the fault-tolerance of channel simultaneously, the joint source-channel coding is implemented with low-density parity check coding. Traditional Slepian-Wolf coding generally needs specific source codes to eliminate redundant bits in code sequences to achieve data compress. If noise occurs in transmission channel, it needs to add special channel codes to provide fault-tolerance function which can decrease or avoid the occurrence of error codes. In this paper, the output of encoder is obtained as the input of accumulator to get compressing syndrome operators which are relative to code units, so that the codes have the anti-interference feature.(2) An improved method based on estimation of sliding-window source parameters and channel noise is proposed in this paper. The running time of this algorithm is improved 1.5 times by decreasing the inefficiency iterative process with coefficients in decoding. A channel noise estimation method based on sliding-window belief coding algorithm is designed. In order to estimate the local statistic parameters of channel noise, the noise parameters are estimated by decoder using local partial probability of adjoin syndrome node after belief propagation iteration each time, which is similar with source coding.(3) A joint source-channel coding scheme based on sliding-window belief propagation algorithm is designed. The initial parameters of decoding are obtained from the both sides of source local statistic estimation and channel noise local partial probability so that the coding efficiency and the performance of the whole system are improved more than 70%. The performance of this algorithm is validated by different crossover probabilities and the coding lengths. Compared with the existing algorithms, the proposed algorithm has the advantages of robustness to initial parameters, the low linear time complexity and the ease to implement.