Dynamic Adaptive Compressive Sensing-Based Multi-User Detection in Uplink URLLC

Abstract

Ultra reliable and low latency communication (URLLC) is one of the three typical service scenarios in the fifth generation mobile communications (5G) system, which supports mission-critical machine-type communication. Grant-free non-orthogonal multiple access (NOMA) system is a promising candidate technology for uplink URLLC scenario but it causes the problem of multi-user detection (MUD). In this paper, we propose a dynamic adaptive compressive sensing (DACS)-based MUD algorithm to realize MUD in URLLC scenario by exploiting user activity sparsity. Different from most of the state-of-the-art compressive sensing (CS)-based MUD algorithms, this algorithm needs no input of user activity sparsity level which may be unknown in practical system. Particularly, this algorithm adopts a stage-wise approach to increase estimated number of active users stage by stage for adaptively acquiring the true user activity sparsity level, introduces a backtracking idea to refine the estimated active user set for more accurate detection, and exploits the temporal correlation between active user sets in adjacent time slots for reducing computational complexity. Simulation results demonstrate that, although the proposed DACS-based MUD algorithm lacks the information of user activity sparsity level, it achieves better bit error rate (BER) performance than the conventional CS-based MUD algorithm.