摘要
首先简单介绍了非正则LDPC码的结构,给出了一种基于IEEE802.16e直接编码法生成的(576,288)非正则LDPC码的编译码原理。然后详细论述了其在TI定点DSP(TMS320C5510)上的定点化算法实现方式,并在经过C和部分汇编优化后将算法效率提高了70%以上,达到了实时系统要求。最后给出了该LDPC码与(2,1,7)卷积码在AWGN信道下的性能对比,表明这种中短码长的非正则LDPC码较卷积码有较大的纠错性能优势。
First, this paper briefly introduces the structure of irregular LDPC ( Low Density Parity Check) codes and gives coding and decoding principles of the (576,288) irregular LDPC codes based on IEEE 802.16e's direct encoding method. Then, it describes how to implement it on the TI fixed point DSP (TMS320C5510) detailedly. After optimizing LDPC decoder using C and assembly language, it reduces the calculation work by 70% and satisfies the standard of real time system. In addition, this paper gives the performance comparison between the (576,288) irregular LDPC codes and (2,1,7) convolutional codes over the AWGN channel. The result shows that the irregular LDPC codes have good error correcting ability at moderate or short length compared with convolutional codes.
出处
《电讯技术》
2008年第2期68-72,共5页
Telecommunication Engineering
关键词
非正则LDPC码
和积算法
DSP
定点化
优化
irregular LDPC codes
sum - product algorithm (SPA)
DSP
fix - point processing
optimize