Joint Channel Estimation and MUD for Scalable Grant-Free Random Access

Abstract

A scalable grant-free access, which simultaneously reduces access collision rate and signaling overhead, can be enabled by allowing contending users to randomly select a preamble from a very large set of non-orthogonal preamble sequences. By casting the joint channel estimation and multi-user detection (CE-MUD) problem into a multiple measurement vector (MMV)-based compressive sensing problem and employing a two-stage active preambles search over a narrowed subspace, a scheme proposed in this letter shows significant performance improvement over conventional schemes. Particularly, the two-stage search allows the proposed scheme to support multiple order of more active users while reducing receiver's complexity.