Multi-Fault Diagnosis in Twisted-Pair Cables of Networked Control Systems Using Transferometry

In networked control systems, power line communication technique is used to transfer data over existing energy cables. A soft fault degrades the integrity of the signal without impacting the system behaviour. This work develops a transferometry-based method for detecting, localising, and estimating the severity of two simultaneous soft faults in such cables. A soft fault is modelled as a series impedance, and the transmission coefficient (TC) is computed from the ABCD cascade model of the cable. We prove that, under unmatched terminations, the time-domain TC exhibits a five-pulse signature whose peak positions and amplitudes map directly to the two fault positions and their individual severities. A residual signal constructed from this signature yields closed-form estimators for the fault positions and their combined severities; individual fault severities require a bounded nonlinear least-square fit, valid for approximately symmetric, known terminations. We further show that the method extends to n simultaneous soft faults under a combined soft-fault condition, with the (2n+1)-pulse pattern verified in simulation for n∈{1,2,3,4}. A Monte Carlo study using correct localisation probability as the detection criterion establishes a practical SNR threshold of 25 dB; fault-separation resolvability shows intermittent, sidelobe-driven degradation rather than a single threshold. Simulations on a measured 24 AWG cable, extrapolated beyond its characterised band, confirm reliable two-fault diagnosis under additive noise, with reliable multi-fault performance demonstrated for n=1,2, presented as a numerical proof of concept on this extrapolated cable model rather than a characterisation confirmed by measurement over the full simulated band.