Research And Implementation Of Parallel Decoding Algorithm Of LDPC Codes

With the deepening and popularization of the mobile Internet applications, broadband mobile data communication is developing rapidly, the outstanding performance on efficiency and stability of the LDPC codes, make it become the preferred scheme of mobile communication. In the process of information transmission, decoding work has become the bottleneck of the communication speed, therefore, how to improve the decoding speed become a focus in the communication study. In order to improve the decoding speed, this paper uses parallel method to explore, study and realize the LDPC codes decoding algorithms. The following parts are included in this article:(1) LDPC codes' theoretical basis has been introduced, the computing power of GPU under different architecture has been researched. Multithreading management way, storage model and programming model of CUDA platform has been introduced in detail.(2) The basic principle and implementation method of LDPC codes' mainstream decoding algorithms have been analyzed, based on this, the parallel feasibility of decoding algorithms have been studied. The belief propagation decoding algorithm which can be parallelized has been studied in depth.(3) Belief propagation decoding algorithm has been optimized, on the basis of the original decoding algorithm, calculation model has been analyzed. This paper uses logarithmic likelihood ratio strategy to decrease the discrete degree of the data processing and increase the efficiency of decoding.(4) In view of the high demand of LDPC codes transmission efficiency, the parallel decoding method of belief propagation decoding algorithm has been designed. To further improve the efficiency of decoding of LDPC codes, merge access and shared memory access has been optimized, and parallelization and optimization of belief propagation decoding algorithm has been realized on CUDA platform.Experiments show that, the optimized belief propagation decoding algorithm can effectively reduce the computational complexity, increase the decoding efficiency and throughput, and improve the ability of mobile communication.