Two-Stage Adaptive and Compressed CSI Feedback for FDD Massive MIMO

Channel state information (CSI) feedback is a key technical challenge for frequency division duplex (FDD) massive multiple-input multiple-output (MIMO) systems. We propose a two-stage adaptive and compressed CSI feedback scheme to reduce the CSI feedback overhead and improve the feedback accuracy. Specifically, a binary angle-of-departure (AoD) vector and the associated angular-domain channel coefficients are first extracted from the massive MIMO channel. Then, by exploiting the dynamic burst-sparsity of the massive MIMO channel in the joint time-angular domain, a two-stage differential feedback scheme is proposed to compress and feed back the AoD vector, where the feedback overhead is adaptively adjusted according to the AoD changing rate. In addition, an adaptive-codebook-based feedback scheme is proposed for the angular-domain channel coefficients. Compared with the 5G New Radio (NR) Type-2 feedback scheme in 3GPP R15, the proposed scheme can significantly reduce the feedback overhead and achieve better block error rate (BLER) performance.