Low-complexity MMSE receivers based on weighted matrix polynomials in frequency-selective MIMO systems
Guido K.E. Dietl, Ingmar Groh, Wolfgang Utschick · 2006
Compared to the Matrix Wiener Filter (MWF), reducedrank signal processing in the Minimum Mean Square Error (MMSE) sense is a well-known method for the design of low-complexity receivers. In this paper, we reveal the relationship between different reduced-rank receivers based on weighted matrix polynomials approximating the MWF in a Krylov subspace, viz., the Multi-Stage Matrix WF (MSMWF), the parallel implementation of Multi-Stage Vector WFs (MSVWFs), and Polynomial Expansion (PE). Besides, we present PE where the weights are approximated based on Random Matrix (RM) theory assuming the application to a frequency-selective Multiple-Input Multiple-Output (MIMO) system. Simulation results show that the MSMWF outperforms the considered reduced-rank methods if their order of computational complexity is the same. 1.