分组Turbo码的译码性能分析及DSP优化

Decoding Performance Analysis and DSP Optimization for Block Turbo Code

基于分组Turbo码的经典译码算法,分析译码参数与译码复杂度的关系及其对译码性能的影响,以一种(15,11)×(13,9)分组Turbo码为例,在权衡复杂度与性能的前提下,给出其在C55系列数字信号处理器(DSP)上的软件实现方案,并从定点化、编译选项、高级语言与汇编语言多个层面对译码算法进行优化,使译码运算量较未优化时降低89%。

Based on the classic decoding algorithm of Block Turbo Code（BTC）,the relationship between decoding parameters and decoding complexity,and the impact of different parameters on performance,are analyzed.Taking a kind of（15,11）×（13,9） BTC for example,the software implementation scheme on C55 series Digital Signal Processor（DSP） is presented considering the compromise between performance and complexity.The optimization is done through different levels such as fix-point processing,complier options,high-level language and assembly language.Computational complexity has 89% reduction after optimization.