The present invention is concerned with the identification of a failed communication network component of an industrial communication network and of a substation automation system in particular. This is achieved by supervising a limited number of all the communication paths (Dl-D4, D2-D5) constituting the communication network, wherein each of the supervised paths includes a distinct set of network components (Nl; N2, L3, N3) and is delimited by two end-devices (Dl,D4; D2,D5). A failure status or operating status of each communication path is determined and reported. Any network component that is part of an operating path is itself regarded as operating, i.e. failure-free, whereas all other network components are considered as non-operating or potentially failed. If there is more than one non-operating component, additional suitable communication paths are evaluated, until one single failed network component is identified and subsequently reported as requiring repair or replacement. The invention is particularly applicable in cases where some of the network components, i.e. the lines or some nodes of the communication paths, are passive components or non-intelligent devices that are unable to report an operating status themselves.