极低误码率下Turbo码的性能估计

Turbo码在中低信噪比下具有接近信道容量极限的优异性能,但在误码率极低、信噪比较高下性能曲线下降平缓,并且难以进行大计算量的仿真。使用输入信息序列片段长度递增筛选法,逐步筛选出所有产生自由距码字的信息帧,从而得出Turbo码完整的自由距特性,并用来估计极低误码率下的性能,该方法有效地减小了完整枚举法的计算量,且保证了精确性。

Performance Estimation of Turbo Codes at Extremely Low Error Probability Level

In comparison with its well-known near-capacity excellent performance, turbo codes represent a flattened "error floor"in relatively high signal-to-noise ratio and extremely low error probability region of their performance curves, also beyond the capability of ordinary computer simulation. In this paper, we employ a simplified algorithm to obtain complete free distance characteristic by selecting all possible input partial sequences that may generate free distance codewords while increasing their length. Thus computing complexity is decreased to moderate level without loss of accuracy.