Feedback compression for time-correlated MIMO block-fading channels using huffman coding

In this paper, we investigate feedback compression for a general multiple-input multiple-output (MIMO) system over time-correlated Rayleigh block-fading channels. With a channel state transition graph characterized by Markov chain, we first compress channel state information (CSI) feedback through Huffman coding. We then propose a threshold based approach by neglecting the states with small transition probabilities with slight performance loss to further reduce the feedback rate. Finally, we study the relationship between the channel ergodic capacity and feedback rate with water-filling precoding. Analysis and simulation show that the feedback rate can be significantly reduced by the proposed algorithms.