Identification of Communication Cables Based on S-Parameters and K-Nearest Neighbors Algorithm

Abstract

Cable identification has a significant role in cable maintenance and fault detection. Before replacing defective cables in a network, they must be properly identified. In this paper, a novel method for coaxial communication cables identification based on scattering parameters is proposed. The input port reflection's magnitude measured by a Nano Vector Network Analyzer at 101 frequencies for 10 coaxial communication cables are used as features for the K-Nearest Neighbors algorithm. The investigation is held on cables of various lengths, dimensions and connector types. The cable's length, type and connectors are considered as a unique class. The classification accuracy reached is 99% for a test set composed of 100 measurements.