Compressive Sensing Algorithms for Multiuser Detection in Uplink Grant Free NOMA Systems

Abstract

Non-orthogonal multiple access (NOMA) scheme has been found to exhibit the ability of supporting massive connectivity of multiple users by employing the same, non-orthogonal, spectrum resources. Grant-free transmission mode in NOMA system allows users to randomly initiate transmission at any time slot, and this mode of transmission is highly desirable for the uplink NOMA system. This paper focuses on comparing three compressive sensing algorithms, that track user activity and detect users' signals in a scenario where the active user set is assumed unchanged in several continuous time slots, with another two compressive sensing algorithms that perform similar tasks when the active user sets changes in several continuous time slots within a whole frame. The five algorithms-based multi-user detectors (MUDs) are employed for multiuser detection processes in a grant-free NOMA system. The simulation results indicate that the compressive algorithm employed while the active user sets changes in several continuous time slots within a whole frame, the dynamic compressive sensing (DCS)-based multi-user detection (MUD), outperformed all the other MUDs. The traditional subspace pursuit (SP)-based MUD employed when the active user sets remains unchanged in several continuous time slots within a whole frame performs in a similar way with its variant, the modified SP-based MUD employed when the active user sets changes in several continuous time slots within a whole frame. The worst performing MUDs are the orthogonal matching pursuit (OMP)-based MUD and its variant, the computational efficient OMP-based MUD that exhibits similar performance with the OMP-based MUD.