V.B. Kreyndelin, L.A. Varukina

Nowadays people-s activities greatly depend on information availability, and soon one of the major data access facilities will be 3G and beyond mobile systems. A prospective technique facilitating mobile broadband access is a space-time coding in Multiple-Input-Multiple-Output (MIMO) systems. Implementation of MIMO demodulation algorithms is a sophisticated computational problem. The Maximal Likelihood method allows the highest performance demodulation at an inherent huge computational intensity. To implement this method for the M Rx/ M Tx K-QAM system it requires real time computational operations at a multiple of KM, that makes the method realization impossible with the modern digital signal processors at large K and M. Other known demodulation methods, e.g. Minimum Mean Square Error and V-BLAST, have a significantly lower computational complexity but a high performance loss comparing to the Maximal Likelihood method. With this paper a quasi-optimal iterative algorithm is proposed: an input stream is splitted into two substreams, and an optimal estimation of an original signal is computed for each of which substream. This method enables to get high demodulation performance achieving nearly the Maximal Likelihood algorithm performance at a significantly lower computational complexity.