Reduced-rank interference suppression algorithm based on generalized MBER criterion for large-scale multiuser MIMO systems

Guijie Wang,Yunlong Cai,Ibo Ngebani,Minjian Zhao
DOI: https://doi.org/10.1109/ChinaSIP.2015.7230407
2015-01-01
Abstract:In this work, a novel adaptive reduced-rank (R-R) algorithm for large-scale multiuser multiple-input multiple-output (MIMO) systems is presented. The proposed algorithm is based on the joint iterative optimization of filter employing the minimization of the bit error rate (BER) criterion using the generalized Gaussian kernel density estimation. The generalized Gaussian kernel density estimation method can better estimate the probability density distribution of sample data having heavier or lighter tails as compared to the normal kernel density estimation technique leading to improved performance. The proposed optimization technique adjusts the weights of a subspace projection matrix and a RR filter in a joint manner. We develop stochastic gradient (SG) algorithm for the adaptive implementation using the generalized Gaussian kernel. The simulation results show that the proposed adaptive algorithm significantly outperforms the compared schemes.
What problem does this paper attempt to address?