基于近似投影的水平分层调度译码算法

基于交替方向乘子法（ADMM）的线性规划（LP）译码模型因其不会出现错误平台和具有最大似然认证的优点，广受译码研究者的关注。目前大多数ADMM算法采用的是泛洪调度策略（FL），该算法存在译码收敛速度过慢的问题。基于水平分层调度的交替方向乘子法的低密度奇偶校验（LDPC）码译码算法能够加速译码收敛速度，然而目前水平分层调度算法中的投影算法采用的为精确投影算法，复杂度较高。针对该问题，文中将近似投影算法和水平分层调度算法结合，提出基于近似投影的ADMM水平分层调度译码算法以提高译码的性能。仿真实验表明，相比其他算法，本文提出的算法的译码性能可提升0.1~0.3dB，迭代次数可降低约19%~40%，平均译码时间可减少大约21%~65%。

Horizontal Layered scheduling of ADMM Decoding for LDPC Codes based on Approximate Projection

Liner programming (LP) decoding model based on alternating direction method of multipliers (ADMM) has attracted much attention of decoding researchers because of its advantages of no error floor and maximum likelihood authentication. At present, most ADMM algorithms adopted the flooding scheduling, which has the problem of slow convergence rate of decoding. The alternating direction method of multipliers (ADMM) decoding for LDPC codes based on horizontal layered can speed up the decoding convergence rate. However, the projection algorithm used in the horizontal layered scheduling algorithm is accurate projection algorithm, which has a high computation. To solve this problem, the approximate projection algorithm was combined with the horizontal layered algorithm, and the horizontal layered scheduling of ADMM based on approximate projection was proposed to improve the decoding performance. Experimental simulation results show that, compared with other algorithms, the decoding performance of the proposed algorithm is improved by 0.1~0.3dB, the number of iterations is reduced by about 19%~40%, and the average decoding time is saved by about 21%~65%.