New Nonlinear Signal Processing Method Based on LS Algorithm and Decision Feedback at Receiver

Abstract

This paper proposes a new low-complexity decision feedback nonlinear equalization architecture. It applies the Least Square (LS) algorithm to the coefficient updating structure of decision feedback blind equalization polynomials and momentum coefficient updating algorithm. It uses a low-complexity memory polynomial model to compensate for the nonlinearity and memory effects caused by amplifiers and channels. Besides, the new structure uses an improved iterative LS algorithm when updating the coefficients of the memory polynomial model, and it uses the decision feedback signal as a standard point to calculate the cost function. The new processing method can handle sizeable nonlinear distortion and memory effects. In the model coefficient updating, the symbol point data block is used to update the memory polynomial coefficients by the improved LS algorithm. Compared with other traditional Least Mean Square (LMS) algorithms or Recursive Least Square (RLS) algorithms, the new method has higher stability. The new algorithm can use shorter data blocks to meet the Low-Density Parity Check Code (LDPC) decoding performance requirements and save a lot of computing and storage resources for satellite receivers. It has excellent reference significance for practical engineering applications.