Linear prediction of time-varying MIMO systems using Givens rotations

Abstract

MIMO systems require eigenmode information at the transmitter to exploit the multiplexing gain of the channel. In practice, the eigenmode information available at the transmitter after feedback is noisy and outdated. One of the efficient feedback methods used in MIMO systems involves the decomposition of unitary matrices into Givens rotations. This entails the possibility to update the eigenmodes by predicting the Givens rotations. In this paper, two MIMO precoder prediction schemes based on channel prediction at the mobile station (MS) and Givens parameters prediction at the base station (BS) are proposed. The system capacity of the two schemes is evaluated and compared by using a simulation on a time-varying MIMO channel. Simulations show that both prediction schemes improve the system performance; however, channel prediction at the MS has better performance than Givens prediction at the BS. Despite a lower performance, Givens prediction at the BS reduces the number of parameters to be tracked and is suitable for variable delay systems.