Wideband FDD massive MIMO downlink channel estimation with new pilot design and hybrid GAMP algorithm

Abstract

This paper proposes a new pilot design for the wideband frequency division duplexing (FDD) massive MIMO channel estimation (CE) associated with the hybrid generalized approximate message passing (GAMP) sparse reconstruction algorithm. During each estimation, one subcarrier is used to carry a long pilot sequence that the channel response as well as the non-zero probability are estimated with GAMP algorithm. Then the property of spatially common sparsity across different subcarriers is used so that the non-zero probability information can be shared by different subcarriers before the CE process with hybrid GAMP algorithm. With the help of the extra non-zero information, a similar CE accuracy performance is achieved with much shorter pilot sequences. Thus, the new pilot pattern consists of one long pilot sequence and lots of short pilot sequences, which have a much better spectrum efficiency.