Massive MIMO-OTFS Satellite Communication Channel Estimation Based on Improved 3D-SOMP

To address the issues of low estimation accuracy and high computational complexity caused by the high-dimensional characteristics of three-dimensional (3D) channels and dynamic Doppler spread in low Earth orbit (LEO) satellite massive MIMO- OTFS systems, this paper proposes an improved 3D structured orthogonal matching pursuit (3D-SOMP) algorithm. By exploiting the joint sparsity in the delay-Doppler-spatial domain, a dynamic 3D redundant dictionary is designed to mitigate the fixed grid mismatch problem. Leveraging prior knowledge of satellite channel coherence time, an adaptive atom search step size strategy is constructed to optimize atom matching paths. Furthermore, a residual energy variation rate threshold mechanism is introduced to dynamically control the iteration count. Simulation results demonstrate that, compared to the traditional 3D-SOMP algorithm, the proposed method achieves an improvement of 1.8 dB in normalized mean square error (NMSE), along with reduced computational complexity and pilot overhead.