Stochastic prediction of wire coupling interference

Many EMC analyses of complex systems frequently result in a statement that insufficient knowledge is available to describe accurately the internal relationships of the system's components. This lack of information precludes any rigorous deterministic prediction and, in principle, requires that we express the uncertainties within the model. This paper shows the practical feasibility of stochastic prediction, as an alternative to deterministic simulation, applied to a class of EMC problems intrinsically affected by randomness. The evaluation of the crosstalk in standard cable bundles, in which several wires are tightly and randomly wrapped together, is the concrete problem that we investigate in this context. We developed a technique based on solving the nonuniform multiconductor transmission lines (MTL) for many randomly generated wires' geometries to obtain many crosstalk samples for a single frequency. Finally we validated the method, setting up a case study with published experimental results.