Fast Converging ADMM-Penalized Algorithm for Turbo-Like OVXDM

Abstract

Overlapped x domain multiplexing (OVXDM) encoding is recently proposed with ultra-high spectral efficiency by the shift and overlap of weighted data under low signal-to-noise ratio (SNR). However, the decoding performance of OVXDM is restricted by the exponential increase of decoding complexity with the growth of the spectral efficiency. In this paper, we first propose a novel type of turbo-like (TL) structure of OVXDM encoding (TL-OVXDM) combining the advantages of turbo and OVXDM encoding. Then a modified alternating direction method of multipliers with a penalty term (ADMM-penalized) algorithm is proposed, which can achieve robust decoding performance in only a few iterations and significantly reduce the computational complexity of the decoding process through a faster convergence. Simulation results demonstrate that TL-OVXDM with Log-BCJR decoding algorithm can outperform the upper bound of OVXDM bit error probability. And when overlapping fold is relatively high in TL-OVXDM, the proposed ADMM-penalized decoding algorithm quickens the convergence speed and achieves robust performance in bit error rate (BER) with low computational complexity.