基于粒子滤波的Turbo盲均衡

摘要

Particle filter ( PF) , which is based on the Bayesian theory, is particularly useful in dealing with the blind channel identification and blind equalization for its fast convergence and its outstanding performance of resisting multiple-path fading channels. Under the low SNR conditions the bit error rates ( BER) of Turbo blind equalization are much lower. In order to get good BER performance in multiple-path fading channels, the particle filter algorithm for blind equalization is modified. The important sampling function of particle filtering exploits the prior information of the particles and the soft input soft output (SISO) particle filter equalization algorithm is proposed. Considering the structure of Turbo blind equalization, a new Turbo blind equalization based on particle filter is proposed, which makes use of the channel coding gain. Therefore, the performance of equalization and channel identification are improved. The simulation result shows that compared to the particle filter equalization algorithm, the bit error rates (BER) of the proposed algorithm have a gain of about ldB, and the frame error rates ( FER) have a gain of above 3dB. By analyzing, there are hardly error bits under the condition of accurate estimation of the channel coefficients.%粒子滤波是一种基于贝叶斯估计的算法,在信道盲辨识和盲均衡问题上具有快收敛、抗深衰信道等优势.Turbo盲均衡在低信噪比条件下有较好的误码性能.为了在深衰信道下使通信具有良好的误码性能,对粒子滤波盲均衡算法进行改进,改进算法的重要性采样函数利用了粒子的先验信息,得到一种软输入软输出的粒子滤波盲均衡算法.依据Turbo盲均衡的框架结构实现了一种基于粒子滤波的Turbo盲均衡算法,该算法利用信道编码带来的编码增益,提高了均衡和信道辨识的性能.仿真结果表明相比粒子滤波盲均衡算法本文提出算法的误码率性能提高1dB左右,误帧率性能则提高了3dB以上,经分析可知在信道系数估计较为准确的条件下,系统数据帧几乎没有误码.