Compressed Sensing Based Multiuser Detection of Grant-Free NOMA With Dynamic User Activity

Abstract

In this letter, by exploiting both the temporal correlation of active user and signal statistical information, we propose a compressed sensing based multi-user detection for uplink grant-free non-orthogonal multiple access where the user activity varies dynamically in one frame consisting of multiple time slots. In the proposed method, channel estimation and data detection are implemented jointly by efficient slot-wise generalized approximate message passing (GAMP) algorithm, active user detection (AUD) is implemented by message passing (MP) algorithm, and GAMP and MP exchanges information through turbo iterations. Specifically, the user activity in each time slot and active user correlation in one frame are introduced as variable and check nodes, respectively, to tackle the dynamic user activity directly in the factor graph for MP-based AUD. Finally, the performance of the proposed method is demonstrated by computer simulations.