Minimum BER linear transceivers for MIMO channels via primal decomposition
2005 (English)In: IEEE Transactions on Signal Processing, ISSN 1053-587X, E-ISSN 1941-0476, Vol. 53, no 8, 2866-2882 p.Article in journal (Refereed) Published
This paper considers the employment of linear transceivers for communication through multiple-input multiple-output (MIMO) channels with channel state information (CSI) at both sides of the link. The design of linear MIMO transceivers has been studied since the 1970s by optimizing simple measures of the quality of the system, such as the trace of the mean-square error matrix, subject to a power constraint. Recent results showed how to solve the problem in an optimal way for the family of Schur-concave and Schur-convex cost functions. In particular, when the constellations used on the different transmit dimensions are equal, the bit-error rate (BER) averaged over these dimensions happens to be a Schur-convex function, and therefore, it can be optimally solved. In a more general case, however, when different constellations are used, the average BER is not a Schur-convex function, and the optimal design in terms of minimum BER is an open problem. This paper solves the minimum BER problem with arbitrary constellations by first reformulating the problem in convex form and then proposing two solutions. One is a heuristic and suboptimal solution, which performs remarkably well in practice. The other one is the optimal solution obtained by decomposing the convex problem into several subproblems controlled by a master problem (a technique borrowed from optimization theory), for which extremely simple algorithms exist. Thus, the minimum BER problem can be optimally solved in practice with very simple algorithms.
Place, publisher, year, edition, pages
IEEE Signal Processing Society, 2005. Vol. 53, no 8, 2866-2882 p.
BER, convex optimization theory, decomposition techniques, linear precoder, MIMO channel, transceiver, waterfilling, decision-feedback, optimal designs, precoders, equalization, optimization, information, equalizers, systems
IdentifiersURN: urn:nbn:se:kth:diva-14926DOI: 10.1109/TSP.2005.850373ISI: 000230652800022ScopusID: 2-s2.0-84859826797OAI: oai:DiVA.org:kth-14926DiVA: diva2:332967
QC 201111092010-08-052010-08-052012-09-26Bibliographically approved