Compressive Sensing Based Grant-Free Random Access for Massive MTC

Abstract

Massive machine-type communications (mMTC) are expected to be one of the most primary scenarios in the next-generation wireless communications and provide massive connectivity for Internet of Things (IoT). To meet the demanding technical requirements for mMTC, random access scheme with efficient joint activity and data detection (JADD) is vital. In this paper, we propose a compressive sensing (CS)-based grant-free random access scheme for mMTC, where JADD is formulated as a multiple measurement vectors (MMV) CS problem. By leveraging the prior knowledge of the discrete constellation symbols, we develop an orthogonal approximate message passing (OAMP)-MMV algorithm for JADD, where the structured sparsity is fully exploited for enhanced performance. Moreover, expectation maximization (EM) algorithm is employed to learn the unknown sparsity ratio of the a priori distribution and the noise variance. Simulation results show that the proposed scheme achieves superior performance over other state-of-the-art CS schemes.