MIMO-Based Multi-LEO-Satellite Cooperative Grant-Free Random Access for IoT Massive Connectivity

The low earth orbit (LEO) satellite communication network has attracted extensive attention owing to its advantages of seamless coverage and low propagation delays, which provides a promising solution to realize massive access for Internet of Things (IoT) devices. In this paper, we study cooperative grant free random access (GF-RA) in LEO satellite communication systems. Specifically, we investigate the joint activity detection and channel estimation (JADCE) problem for multi-input multi-output (MIMO) based massive connectivity. First, we analyze the channel characteristics, and reveal the low-rank and row-sparsity properties of the channel impulse response (CIR) matrix. Accordingly, we transfer the JADCE problem into a low-rank matrix completion problem and a compressive sensing problem, which are solved by a two-stage algorithm efficiently. In the first stage, we design the principal component analysis with adaptive signal space detection (PCA-AASD) algorithm to perform low-rank matrix completion. In the second stage, we employ a sequential sparse Bayesian learning with multiple measurement vector (MMV) algorithm to perform active terminal detection and channel estimation. Finally, a majority voting scheme is utilized to estimate the active terminals by aggregating the estimation of multiple satellites. Simulation results validate that the proposed method achieves lower activity detection error probability and better channel estimation performance than other baseline methods in the literature.