基于滑窗置信传播算法的联合信源信道编码

针对联合信源信道编码中信源统计特性和信道噪声参数未知情况下,解码算法性能急剧下降的问题,提出一种基于滑窗置信传播算法(SWBP)的联合信源信道编码,简称滑窗算法.研究采用不规则重复累积(IRA)码来实现基于滑窗置信传播算法的联合信源信道编码,IRA码可以取得与低密度奇偶校验(LDPC)码同样优越的性能,但编码复杂度远远低于LDPC码.在解码端引入滑动窗口,通过实时地估计不断变化的信源统计特性和信道噪声参数,从而提高解码速率.实验结果表明该算法具有接近相关参数已知情况下的理想性能、不依赖于初始参数、复杂度低且易于实现等优点.

Joint Source-Channel Coding Based on Sliding-Window Belief Propagation

To solve the problem of unknown source statistical properties and channel noise parameters and that of the decoding performance degradation, a new joint source-channel coding scheme based on the Sliding-Window Belief Propagation (SWBP) algorithm was proposed. In this paper, irregular repeat accumulate (IRA) is adopted to achieve the SWBP algorithm in joint source-channel coding. IRA coding can obtain the same superior performance with low density parity check (LDPC) coding, but less than complexity. A sliding-window is introduced in the decoder, and the decoding rate is improved by estimating the changing source statistical properties and channel noise parameters real-timely. The experiment results show that the performance of the proposed algorithm is close to the ideal curve when correlation parameters have been given. While the algorithm shows other advantages, including the robustness to initial parameters, the low linear time complexity, and the ease to implement.