BISECTION-BASED SPARSITY UPDATE SUBSPACE PURSUIT ALGORITHM FOR SPARSE SIGNAL RECOVERY

Kin Tim Lao ； Kelvin Kuang-Chi Lee

《電機工程學刊》 29卷1期 (2022/02) Pp. 7-12

https://doi.org/10.6329/CIEE.202202_29(1).0002

摘要

Sparse signal reconstruction algorithms have attracted considerable research attentions lately for the fifth generation (5G) mobile communication networks. As many critical parameters in 5G applications are observed to exhibit sparse characteristics, low-complexity compressed sensing (CS) techniques are promoted to acquire the sparse signals, given that exact sparsity is known a priori. Nonetheless such assumption is often unrealistic in practice. Hence, several sparsity approximation methods, e.g., backtracking adaptive orthogonal matching pursuit (BAOMP) and sparsity update subspace pursuit (SUSP) are proposed to address this issue. In this paper, a bisection-based sparsity update subspace pursuit (BSUSP) algorithm is presented, which aims to accelerate the convergence of sparsity estimation and recover sparse signals without the need of exact sparsity. To justify its efficacy, simulations are conducted where the preliminary results show that better normalized mean square error (NMSE) and location accuracy ratio (LAR) can be achieved compared to other greedy-based benchmark methods.

關鍵字

Sparse signal recovery ； compressed sensing ； bisection ； subspace pursuit ； NMSE ； LAR