A modified sequential quadratic programming method for sparse signal recovery problems

• Published in: Signal processing Vol. 207; p. 108955
• Main Authors: Alamdari, Mohammad Saeid, Fatemi, Masoud, Ghaffari, Aboozar
• Format: Journal Article
• Language: English
• Published: Elsevier B.V 01-06-2023
• Subjects: Non-convex optimization; Sequential programming method; SL0 approximation; Sparse signal recovery; Non-convex optimization; Sparse signal recovery; SL0 approximation; Sequential programming method
• ISSN: 0165-1684; 1872-7557
• DOI: 10.1016/j.sigpro.2023.108955

•Sequential quadratic programming for sparse signal recovery provide superlinear convergence.•Theoretical global convergence results are well established.•Exact solution of subproblems can be obtained in a low cost.•The SQP algorithm provide good results for real word problems.
We propose a modified sequential quadratic programming method for solving the sparse signal recovery problem. We start by going through the well-known smoothed-ℓ0 technique and provide a smooth approximation of the objective function. Then, a variant of the sequential quadratic programming method equipped with a new approach for solving subproblems is proposed. We investigate the global convergence of the method in detail. In comparison to several well-known algorithms, simulation results demonstrate the promising performance of the proposed method.