The paper proposes a decentralized diagnosis approach for a set of choreographed BPEL Web services, where a local diagnoser is associated to each BPEL service and cooperates with a coordinator. The local diagnosis is based on a Colored Petri Nets model enriched with I/O data dependency relations represented with color propagation functions. By applying the multiset marking calculation equation, a diagnosis inequations system is constructed and solved to retrieve a local diagnosis. The coordinator updates the global diagnosis until reaching a final consistency. 1

Abstract—This paper aims to present a method of creating architectures which allow monitoring occurrence of failure in Service oriented Architectures (SoA). The presented approach extends Discrete Event Systems techniques to produce a method of automated creation of Diagnoser Service which monitors interaction between the services to identify if a failure has happened and the type of failure. To do so, a formal representation of business processes is introduced, which allows modeling of observable/unobservable events, failure and the type of failure. The paper puts forward a set of algorithms for creating models of Diagnoser Service. Such models are then transformed into new Services implemented in BPEL, which interact with the existing services to identify if a failure has happened and the type of failure. The approach has been applied to an example of diagnosis of Right-first-time failure in Services used in telecommunications. I.

Abstract: The paper contributes to modeling an orchestrated complex Web Service (BPEL) with Colored Petri Nets (CPNs) for diagnosis. In the CPNs model, colored tokens are used to represent the faults in a BPEL process. A uniform fault model is introduced to represent both the faulty input data and external faulty Web services called by the BPEL activities. We use three I/O data dependency relations for each BPEL activity. To represent the fault propagation in colored Petri nets, we define the color propagation functions for each data dependency relation. We give a concrete translation from a BPEL service to a CPNs model. Model-based diagnosis framework is then used. Based on the evolution equation in Petri nets theory, we construct an inequations system as a diagnosis problem and solve it with an algebra algorithm.