Fault localization in general refers to a technique for identifyingudthe likely root causes of failures observed in systems formed fromudcomponents. Fault localization in systems deployed on mobile ad hocudnetworks (MANETs) is a particularly challenging task because thoseudsystems are subject to a wider variety and higher incidence of faultsudthan those deployed in fixed networks, the resources available toudtrack fault symptoms are severely limited, and many of the sources ofudfaults in MANETs are by their nature transient. ududWe present a suite of three methods, each responsible for part of theudoverall task of localizing the faults occurring in service-basedudsystems hosted on MANETs. First, we describe a dependence discoveryudmethod, designed specifically for this environment, yielding dynamicudsnapshots of dependence relationships discovered through decentralizedudobservations of service interactions. Next, we present a method forudlocalizing the faults occurring in service-based systems hosted onudMANETs. We employ both Bayesian and timing-based reasoning techniquesudto analyze the dependence data produced by the dependence discoveryudmethod in the context of a specific fault propagation model, derivinguda ranked list of candidate fault locations. In the third method, weudpresent an epidemic protocol designed for transferring the dependenceudand symptom data between nodes of MANET networks with lowudconnectivity. The protocol creates network wide synchronizationudoverlay and transfers the data over intermediate nodes in periodicudsynchronization cycles.ududWe introduce a new tool for simulation of service-based systems hostedudon MANETs and use the tool for evaluation of several operationaludaspects of the methods. Next, we present implementation of the methodsudin Java EE and use emulation environment to evaluate the methods. Weudpresent the results of an extensive set of experiments exploring audwide range of operational conditions to evaluate the accuracy andudperformance of our methods.
展开▼
