Diagnosing Failures in Wireless Networks Using Fault Signatures

Detection and diagnosis of failures in wireless networks is of crucial importance. It is also a very challenging task, given the myriad of problems that plague present day wireless networks. A host of issues such as software bugs, hardware failures, and environmental factors, can cause performance degradations in wireless networks. As part of this study, we propose a new approach for diagnosing performance degradations in wireless networks, based on the concept of ``fault signatures´´. Our goal is to construct signatures for known faults in wireless networks and utilize these to identify particular faults. Via preliminary experiments, we show how these signatures can be generated and how they can help us in diagnosing network faults and distinguishing them from legitimate network events. Unlike most previous approaches, our scheme allows us to identify the root cause of the fault by capturing the state of the network parameters during the occurrence of the fault.