A Self-interference Canceller Design Based on Knowledge-driven Neural Network

Le résumé fourni par la source

In-band full-duplex (IBFD) technology, which enables signal simultaneous transmission and receptions in the same frequency band, has been recognized as a key technology in wireless networks. However, self-interference (SI) is a significant problem that hinders its practical application, and effective SI cancellation techniques should be employed. Traditionally, SI uses a polynomial model with a nonlinear adaptive learning algorithm for reconstruction and cancellation, but the number of parameters that need to be estimated is large and difficult to achieve. Therefore, in order to obtain cancellation algorithms with fewer parameters and low complexity, neural network tools have been noticed. In this paper, we propose a knowledge-driven neural network canceller structure, which significantly reduces the computational complexity compared with the polynomial model. Experimental results show that the proposed structure can reduce the number of real addition and multiplication calculations by about 40-60 and 0-40% while achieving similar nonlinear cancellation performance.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.