Low-Complexity $$l_0$$l0-Norm Penalized Shrinkage Linear and Widely Linear Affine Projection Algorithms

摘要: In this paper, we propose an $$l_0$$l0-norm penalized shrinkage linear affine projection ($$l_0$$l0-SL-AP) algorithm and widely ($$l_0$$l0-SWL-AP) algorithm. The proposed algorithms provide variable step-size by minimizing the noise-free a posteriori error at each iteration introduce constraint to cost function. $$l_0$$l0-SWL-AP also exploits noncircular properties of input signal. contrast with conventional AP algorithms, increase estimation accuracy for time-varying sparse system identification. A quantitative analysis convergence behavior verifies capabilities algorithms. To reduce complexity, dichotomous coordinate descent (DCD) iterations ($$l_0$$l0-SL-DCD-AP $$l_0$$l0-SWL-DCD-AP) in paper. Simulations indicate that $$l_0$$l0-SL-AP faster speed lower steady-state misalignment than previous APA-type $$l_0$$l0-SL-DCD-AP $$l_0$$l0-SWL-DCD-AP perform similarly their counterparts but reduced complexity.