Prediction and Suppression of Twisted-wire Pair Crosstalk Based on Beetle Swarm Optimization Algorithm

Abstract

Based on the theory of multi-conductor transmission lines (MTL), this paper proposes a new method for predicting and suppressing crosstalk of twisted-wire pair (TWP). The per unit length (p.u.l) RLCG parameters change caused by the inconsistent cross-sectional shape of TWP, changes in parameters make it difficult to solve the telegraph equation. In this paper, the method of transmission lines cascade is used. TWP is divided into several segments, and p.u.l parameters of each segment are predicted. Compared with before method, we propose a higher precision algorithm—beetle swarm optimization (BSO) to optimize the weights of back-propagation (BP) neural network, which predict p.u.l parameters at each segment. On this basis, it is divided into two steps: 1) Use MTL frequency domain method combined with lines’ terminal conditions to solve crosstalk and compare with CST simulation results; 2) Use the singular value decomposition (SVD) method to add matrix modules at both ends of lines for suppressing crosstalk. The results show that proposed method in this paper is consistent with the simulation, and the accuracy is higher than before