An application of univariate marginal distribution algorithm in MIMO communication systems

Abstract

The paper discusses a sequence detector based on univariate marginal distribution algorithm (UMDA) that jointly estimates the symbols transmitted in a multiple input multiple output (MIMO) communication system. While an optimal maximum likelihood detection using an exhaustive search method is prohibitively complex, it has been shown that sphere decoder (SD) achieves the optimal bit error rate (BER) performance with polynomial time complexity for smaller array sizes. However, the worst-case complexity of SD is exponential in the problem dimensions, this brings in question its practical implementation for larger number of spatial layers and for higher-order signal constellation. The proposed detector shows promising results for this overly difficult and complicated operating environment, confirmed through simulation results. A performance comparison of the UMDA detector with SD is presented for higher-order complex MIMO architectures with limited average transmit power. The proposed detector achieves substantial performance gain for higher-order systems attaining a near optimal BER performance with reduced computational complexity as compared with SD. Copyright © 2009 John Wiley & Sons, Ltd. A novel approach to MIMO detection based on univariate marginal distribution algorithm (UMDA) from the family of estimation of distribution algorithms (EDAs). The paper details the working of the optimization algorithm to solve complex search problems and also identifies the grey performance areas. Compared to the state of art exiting algorithms, UMDA achieves a near optimal BER performance with significantly reduced computational complexity in MIMO systems employing relatively higher number of transmit antennas and using larger signal constellations. Copyright © 2009 John Wiley & Sons, Ltd.