基于混合消息传递和部分近似的联合信道估计MIMO-OFDM接收机

本文提出了一种基于混合消息传递和部分高斯近似(Partial Gaussian Approximation, PGA)的多用户干扰消除方法，并应用到联合信道估计MIMO-OFDM接收机中。由于多用户干扰模型中存在的“乘积-求和”结构，使得选择标准消息传递规则，如置信传播(Belief Propagation, BP)，期望传播(Expectation Propagation, EP)，平均场规则(Mean Field, MF)，或者联合方法时只能在性能或复杂度方面有所取舍。现有根据标准消息传递规则得到的最优性能接收机复杂度高，而近似程度大的低复杂度接收机性能损失严重。本文根据多用户干扰模型的自身特点，对标准消息传递规则进行了修改，提出了一种基于混合消息传递规则和部分高斯近似的多用户干扰消除方法。依据信道估计过程中不同用户的信道权重，采用不同的消息传递规则，可以实现复杂度和性能的均衡调整。仿真结果表明，本文提出的多用户干扰消除方法，在性能接近已知最优接收机的情况下，能够大幅降低复杂度。 

Joint Channel Estimation and SC-FDE Receiver Based on Stretch Factor Graph and Combined BP-MF Framework

Based on the hybrid message passing rule and partial Gaussian approximation (PGA) method, a novel multi-user interference elimination algorithm has been proposed in this paper, and is applied to the joint channel estimation and detection for MIMO-OFDM system. The “multiplication-summation” structure lies in the multi-user model makes the adoption of standard message passing rules difficult, since the performance and complexity cannot achieve simultaneously. The standard message passing rules include belief propagation (BP), expectation propagation (EP), mean field (MF), and the joint approaches as BP-MF or BP-EP. Based on the structure of multi-user interference, a hybrid message passing rule and partial Gaussian approximation based algorithm has been proposed in this paper, which can be seen as an amendment of the standard rule. By the evaluation of channel weight, the proposed method employ different message passing strategies to different users, thus compromise the complexity and performance. Simulation results shown that, the proposed MIMO-OFDM receiver approaching the best one in literature while has much lower complexity.