Weighted $l_{p}$ Minimization for Compressive Channel Estimation in FDD Massive MIMO

The paper proposes a weighted l
_p
minimization algorithm for downlink channel estimation in frequency division duplexing (FDD) massive multiple-input multiple-output (MIMO) system, in which we take the popular l
_p
minimization algorithm named as Iteratively Reweighted Least Squares(IRLS) method for example. The proposed algorithm involves two stages, in which it first diagnoses the downlink support by utilizing the channel sparsity in angular domain and spatial reciprocity for uplink and downlink channels, and then a weighted IRLS algorithm based on the diagnosed support is used for downlink channel recovery. The diagnosed support is used for generating the weighting matrix in the weighted IRLS. The restricted isometry property(RIP) based analysis results for the lp minimization are given out. The results has the general form for 0 <; p ≤ 1 and can be specified to the RIP analysis results for standard l
₁
and weighted l
₁
minimization which are given in other literatures. Hence our analysis results have the more general form. Simulations show that the proposed method can achieve higher estimation accuracy compared with IRLS, weighted subspace pursuit, weighted l
₁
minimization and standard l
₁
minimization.