用于Turbo迭代译码的近似Log-MAP算法研究

本文运用逼近论,研究Log-MAP算法的近似算法.本文提出了校正函数的一阶、二阶和三阶逼近多项式,并对近似式的逼近精度进行了分析和比较.本文将近似Log-MAP算法用于WCDMA turbo译码器中,对译码器在AWGN信道和平坦慢衰落信道上的纠错性能进行了仿真.仿真结果表明：一阶近似Log-MAP算法将Max-Log-MAP turbo译码器的纠错性能改进了0.2~0.3dB,二阶及三阶近似Log-MAP算法与原Log-MAP算法性能等价,优于Max-Log-MAP 算法0.3~0.5dB.

Approximated Log-MAP decoding algorithms for turbo codes

In this paper, approximation theory is used to study Log-MAP algorithm. Based on the optimal uniform approximation method and the least square approximation method, first order, second order and third order polynomials are presented to approximate the correction function of Log-MAP algorithm, and the approximation precision is compared between the algorithms. BER performances of WCDMA turbo decoders based on the approximated Log-MAP algorithms are simulated and compared over AWGN channel and flat slow fading channel. Simulation results show that the coding gain of first-order approximated Log-MAP turbo decoder is 0.2-0.3dB higher than that of MAX-Log-MAP turbo decoder, while BER performances of second-order and third-order approximated Log-MAP algorithms are equivalent to the original Log-MAP algorithm and have 0.3-0.5dB additional coding gain when compared with MAX-Log-MAP one.