Adaptive multi-step channel prediction in spatial channel model using Kalman filter

Abstract

In Time Division Duplex (TDD) System, the downlink Channel State Information (CSI) can be acquired by using the channel reciprocity, after Base Station (BS) has acquired uplink CSI through channel estimation. Nevertheless, due to the mobility of Mobile Station (MS), the rapidly varying channel gain may result in the outdated-CSI at transmitter end. Besides, without any feedback mechanism, BS cannot correct the downlink CSI using channel estimation information at MS end, if there exist no perfect channel reciprocity. This paper proposes a novel adaptive multi-step channel prediction mechanism to combat these problems. By deriving the expression of prediction error, BS can adaptively predict CSI of multi-step ahead while keeping the Mean Squared Error (MSE) under a certain threshold. CSI of the rest samples can be acquired through linear interpolation. The complexity at BS end is largely alleviated correspondingly. Moreover, the proposed scheme takes Spatial Channel Model into consideration to reflect real transmission scenario. Simulation results illustrate that the proposed scheme can effectively reduce the complexity while maintaining the MSE under a certain threshold.