Performance Analysis of Polynomial Matrix SVD-Based Broadband MIMO Systems

Abstract

Singular-value decomposition (SVD) is well-established in multiple-input multiple-output (MIMO) signal processing where a broadband MIMO channel is transformed into a number of weighted single-input single-output (SISO) channels. However, applying SVD to frequency-selective MIMO channels results in unequally weighted SISO channels requiring complex resource allocation techniques for optimizing the channel performance. Therefore, a different approach utilizing polynomial matrix singular-value decomposition (PMSVD) for removing the MIMO interference is studied, outperforming conventional SVD-based MIMO systems in the analyzed channel scenarios. As shown by the bit-error rate (BER) simulation results as well as the obtained spectral efficiencies, the proposed PMSVD-based solution seems to be a good alternative to conventional SVD-based MIMO systems.