Multi-Sensor-based Method for Multiple Hard Faults Identification in Complex Wired Networks

Wired networks are the backbone of most modern systems. Even with the continuous progress of wireless technologies, cables are still the main media of power transmission and communication. Nevertheless, during manufacturing, assembly, and normal functioning, cables are susceptible to aging leading to defects, which prevent systems from well achieving their intended functions with the expected performance or even lead to total failure and long downtimes, leading thereby to a serious problem of reliability. In this paper, we propose a novel low-cost method based on multi-sensor data fusion to identify hard faults, i.e., open and short circuits in complex wired networks composed of multiple independent subnetworks. It can automatically adapt its parameters to meet the limited frequency of the low-cost embedded system and to the network wires lengths. To experimentally validate the method, a low-cost embedded system has been developed. The proposed method has been applied on a wired network with 18 ports and five independent subnetworks, including three open-circuit faults and four short-circuit faults. The processing time needed to measure the networks' responses and identify the seven defects was 11 ms on an Arduino Mega 2560 board, corresponding to 90 cycles per second. This noticeably short processing time allows this method to detect intermittent faults by monitoring instant changes in the network topology.