Fault Localization for Power Line Communications with Topology Inference

Power line communications (PLC) is one of the communication methods which have been deployed and attracted considerable attention on supporting smart grid applications. Furthermore, link failures will have a great impact on the normal operation of PLC network and the causes of link failures are diverse. Therefore, rapid fault location is extremely important for PLC. Fault localization generally requires complete knowledge of the network topology which is usually unknown and needs to be learned by topology inference algorithms. However, it is difficult for routers to collect complete topology information for topology inference. We propose an algorithm based on partial information for topology inference to help the PLC to detect and locate faults normally. The inference algorithm first builds a virtual topology, overestimates the number of network components, and then repeatedly merges the links in this topology to infer the structure of the real network. We conducted simulations and compared our algorithm with the most advanced inference algorithms. The results show that compared with other methods, the topology structure of our inference algorithm significantly improves the performance of the fault location.