Grant-Free Random Access in Massive MIMO Based LEO Satellite Internet of Things

Abstract

Low earth orbit (LEO) satellite based Internet of Things tend to exhibit unique advantages for broad coverage over the earth with relatively low latency. This paper investigates the random access problem in massive multi-input multi-output (mMIMO) systems for LEO satellite communications (Satcom). Specifically, a training sequence based grant-free random access scheme is adopted to deal with the joint activity detection and channel estimation. Considering the limited power supply and hardware cost onboard, a quantized compressive sensing algorithm is developed to mitigate the distortion introduced by low-resolution analog to digital converters. Expectation maximization algorithm is then employed to learn the unknown parameters in the prior assumption. Simulation results verify the effectiveness of our proposed scheme.