Efficient Training of Kalman Algorithm for MIMO Channel Tracking

In this paper, a Kalman algorithm is applied to track a time-varying flat fading MIMO channel. The importance of training and appropriate initialization in combination with the Kalman tracking algorithm is shown. Adopting a periodical training scheme with a given bandwidth efficiency, a trade-off between investing pilots for good initialization and training the algorithm exclusively leads to the lowest BER. We also introduce a training on request scheme, in order to overcome the error propagation encountered by the Kalman filter after a series of detection errors. For this purpose, two metrics to detect the Kalman filter divergence are developed.We show the effectiveness of the new aperiodical training scheme in reducing the channel estimation error and saving bandwidth at the same time.