Mitigating the Impact of Channel Aging in Cell-Free MIMO Systems Using a Channel Predictor Based on Extended Kalman Filter

Abstract

Several previous works have pointed out that the channel aging effect may severely degrade the performance of wireless systems. This effect may be more pronounced in cell-free networks due to two main reasons. First, designing centralized precoders for cell-free (CF) systems requires the exchange of a large amount of channel state information (CSI) data between access points and a central processing unit, which may cause considerable processing delays. Secondly, the CF architecture requires keeping track of multiple aging channels, since user equipment devices are served by multiple cell sites. In this paper, we consider a CF multiple input multiple output (MIMO) system that uses joint coherence transmission and operates in the presence of channel propagation clustering under high Doppler frequencies. Unlike previous works, which focus on analyzing the impact of channel aging, this paper proposes a Kalman filter-based directional CSI prediction scheme that relies on a set of underlying $P$-th order autoregressive processes that correspond to the active links of the CF system. The extended Kalman filter is chosen with the purpose of predicting the channel path parameters. Using the predicted directional CSI, we design centralized hybrid precoding. Our results indicate that the proposed CSI prediction and precoding schemes outperform related benchmarks schemes and a 3rd order AR process model is a good engineering trade-off between model complexity and CSI prediction performance in practically relevant CF scenarios.