Novel Image Rejection Filter Based on Neural Networks

摘要

In this paper a novel image rejection algorithm based on neural networks is proposed. The low-IF receiver architecture and the phenomena of I/Q imbalance (also referred as image interference) are described. The proposed filter is an enhancement of a complex LMS adaptive filter, which separates the desired and image signals, but the recovered signal still suffers from the effects of imbalance parameters. This is corrected by the proposed filter. Simulink simulations were performed in order to prove the functionality of the novel filter. The simulations prove the convergence and stability of the filter. The necessary sample number is given to achieve -60 dB image rejection.