Abstract:

Massive multiple-input multiple-output (MIMO) is a solution for efficiently providing connection services for a variety of machine equipment in the Internet of things (IoT), and efficient connection services require accurate channel estimation.Aimed at the problems of high pilot overhead and poor performance of normalized mean square error (NMSE) estimation in downlink channel estimation of massive MIMO systems, based on the compressed sensing (CS) theory, the common sparsity of the channel space domain was combined while using the feature of lower sparsity of adjacent time slot differential channel impulse response (CIR), which leaded to a significant reduction in pilot overhead.In the reconstruction algorithm, a two-stage differential estimation algorithm, which divided the channel estimation in consecutive time slots with time correlation into two stages, was proposed and the idea of adaptive compressed sensing was combined to achieve fast and accurate CIR estimate.The simulation results show that the proposed two-stage differential channel estimation algorithm not only has a significant improvement in the estimated NMSE performance and data transmission rate compared to the existing CS-based multiple measurement vector (MMV) algorithm, but also show a certain reduction in runtime complexity.

Key words: Internet of things, massive MIMO, compressed sensing, channel estimation, sparsity adaptive, differential CIR