粒子滤波理论及其在盲均衡中的应用

Particle filter theory and its application in blind equalization

粒子滤波器的基本思想是用加权的离散随机样本点表示所需要的后验概率密度。粒子滤波理论的盲均衡方法与其他均衡方法相比,其优点在于它不包括对当前估计的线性化,而是利用离散的随机测度来对期望分布进行近似,而且算法收敛快,所需的数据量较少。介绍了粒子滤波理论及其在盲均衡中的应用。仿真结果表明,使用粒子滤波器的盲均衡方法在信噪比较低时也能完成对信道的均衡。

In this paper, particle filter theory and its application in blind equalization are presented. The basic idea of particle filter is the recursive computation of relevant probability distributions using discrete random measures composed of the particles and its weights. Compared with the other methods for equalization, the advantage of particle filtering is that exploited approximation doesn＇t involve linearization around current estimates but approximations in the representation of the desired distribution by discret random measures. The simulation shows that the algorithm is feasible even under the conditions of lower signal-to-noise ratio.