Fault diagnosis in service-oriented computing through partially observed stochastic Petri nets

Abstract

Service composition, interoperability, loose coupling and distributed nature make service-oriented computing (SOC) enhanced for small-scale to large and complex software systems, though it tends to increase the vulnerability of faults. The properties of SOC pose newer challenges in fault diagnosis. To deal with this problem, we need to enhance the understandability that assists in fault analysis of SOC. In this paper, we have proposed a model-based fault diagnosis approach for SOC adopting the concept of partially observed stochastic Petri nets. In this work, Web services are transformed into Petri nets and stochastic Petri nets using the existing constructs. Then, the calculated reachability graph of the modeled Petri nets is used to diagnose the fault in the observable sequence with the help of labeled Petri nets. Experiments are conducted for the illustration of the proposed fault diagnosis model. The analyzed performance of the proposed model guarantees the accessibility of our approach and suggests the inspection of the model into real-world environments.