Prediction of the eigenvectors for spatial multiplexing MIMO systems in time-varying channels

In mobile communications, time varying channels make the available channel information out of date. Timely updating the channel state is an obvious solution to improve the system performance in a time varying channel. However, a better knowledge of the channel comes at the cost of a decrease in the system throughput. Thus, predicting the future channel conditions can improve not only the performance but also the throughput of many types of wireless systems. This is especially true for a wireless system where multiple antennas are applied at both link ends. In this paper we propose and evaluate the performance of a prediction scheme for multiple input multiple output (MIMO) systems that apply spatial multiplexing. We aim at predicting the future precoder/decoder directly without going through the prediction of the channel matrix. The results show that in a slowly time varying channel an increase in the system performance by a factor of two is possible.