基于随机集模型的MIMO-OFDM信道估计算法

在多径分量数确定的前提下，MIMO-OFDM系统采用传统的基于导频辅助和盲信道估计算法能获得较好性能。实际无线环境中，多径分量数目与幅度都是时变的，则传统信道估计方法不再适用。该文采用随机集理论建模MIMO-OFDM系统信道多径分量数的变化和MIMO信道。基于此模型提出了集中粒子空间重采样方法(CRS)，在保留大概率粒子抽样样本的同时主动抛弃小概率抽样样本，以获得更为准确的真实样本逼近。并提出了基于集中重采样Rao-Blackwellised粒子滤波的信道估计方法(RBPFC)。仿真结果表明:所提出的RBPFC方法信道估计性能最好，基本Rao-Blackwellised粒子滤波方法次之但优于基本粒子滤波算法，卡尔曼滤波的信道估计方法性能最差。

An Estimation Algorithm of MIMO-OFDM Channels Based on RST

The typical pilot-aided and blind estimation method for MIMO-OFDM channel can achieve good performance when the number of multi-path components is constant. However, in the practical wireless environment, the number of channel taps and amplitude are all unknown and time-varying in whole process, thus typical estimation methods are not suitable. In this paper, the channel-taps’ varying condition and a new channel model are established by using RST theory. Based on this model, the re-sample method by Concentrating particle Resample Space (CRS) is proposed. By abandoning low probability samples and reserving high probability samples,  more accurate approximation is obtained at each iteration. And then the channel estimation method using Rao-Blackwellised Particle Filtering with CRS (RBPFC) is proposed. Simulation results show that the performance of RBPFC is the best, the performance of Rao-Blackwellised particle filtering scheme follows but is better than that of the basic particle filtering scheme, and the performance of Kalman filter-based scheme is the worst.