Asia-Pacific Conference on Communications
On the Performance of Iterative Order Recursive Least Square for Compressive Sensing-based Multi-user Detection
Ameha T. Abebe, Chung G. Kang,
PDF download (1.5MB)
This paper provides the performance of an iterative order recursive least square-based (IORLS) orthogonal matching pursuit (OMP) algorithm in comparison with various greedy algorithms. The IORLS algorithm significantly reduces the complexity of compressive sensing (CS) detection for a frame-wise sparse signals. The algorithm is designed putting a synchronous CS-based delay-intolerant machine type communication (MTC) in mind where decoding complexity must be kept as low as possible. In a frame-wise sparse multi-user detection (MUD) problems, IORLS algorithm provides higher performance by collecting node activity information from all the symbols transmitted in a frame. In addition, by replacing the least square estimation in the greedy CS algorithms with iterative order recursive least square estimation, IORLS reduces the computational time required for detection. Furthermore, the proposed algorithm is shown to be robust against noise, achieving near-oracle detection performance.