Proc. of IEEE Global Telecommunications Conference, Washington, DC, USA, pp. 3499-3503, 26-30 Nov. 2007.
Block diagonalization is a precoding technique for multiuser MIMO systems that pre-cancels inter-user interference at the transmitter side. When there are a large number of base station antennas but a limited number of RF amplifiers, the system performance can be significantly improved by switching a subset of antennas to the RF chains and exploiting antenna selection diversity. The optimal antenna subset can be obtained by exhaustively searching over all possible antenna combinations. This brute-force search, however, is prohibitively complicated and impractical. To reduce the complexity, in this paper we propose several low-complexity suboptimal transmit selection algorithms that minimize a symbol error rate (SER) upper bound or maximize a capacity lower bound. Simulation results show that our proposed algorithms perform very close to the optimal exhaustive search, while the complexity is much lower.
This paper is available as a IEEE Xplore .