Low-Rank Tensor Completion-Based Channel Database for Cell-Free Massive MIMO Networks

In Cell-Free Massive Multiple-Input Multiple-Output (CF Massive MIMO) systems, channel estimation is one of the key technologies because important components such as linear precoding schemes (e.g., conjugate beamforming and zero-forcing) and transmission power control schemes are based on channel coefficients between access points and users. In this paper, we propose low-rank tensor completion-based channel database for large scale fading coefficients that include path loss and shadowing effects. In the proposed channel database, the large scale fading coefficients are formulated as a low-rank third-order tensor, which is referred to as channel tensor, and these coefficients are stored by means of factor matrices and a core tensor, which are estimated from partially observed coefficients using a low-rank tensor completion technique. The large scale fading coefficients for a newly generated user are obtained from a simple calculation with the core tensor and the factor matrices. The proposed channel database is scalable because the number of elements in the core tensor and the factor matrices can be reduced significantly as compared with the number of elements in the channel tensor. In this paper, we present a basic idea of the proposed channel database and evaluate its basic performance with simulation experiments.