Research On Runtime Monitoring Of Temporal Properties For Web Service

Service-Oriented Architecture (SOA for short) is an emerging software engineering paradigm and is used to develop dynamic evolution of Web applications. Service providers and service users can public and search for useful service on public registries. Individual services can be combined into a complex Web service composition and interactions between services are based on messages. Since Web service composition is distributed, autonomous, dynamic evolution, the designers of it do not own the entire service, which leads that the correctness of the service is easy to be affected by third party participants and the institutions interacting with the system in service-based systems.The correctness of the process of service composition can be ensured by using static testing, validation and other technology, but its runtime behavior may still violate the various property requests of the service combination, so it is necessary to implement Web service run-time monitoring, on service-based systems. In practice, it is typically to deploy continuous runtime monitoring method in the system to monitor the whole service composition, to ensure that the runtime behavior of the system is able to comply with some specific properties. However, most existing results only focus on the properties associated with the monitoring of system control process, only pay attention to the correctness of the name of the messages or the order of information interacted between different services. In these methods, many dynamic properties cannot be monitored, and it is difficult to understand how data of interest is interacted between different partner services.Temporal property refers to properties which can dynamiclly binding with different values in the system running process, depending on system behaviors. These kinds of properties need to monitor not only the execution sequence of events, but also the values of event parameters are in real-time system process. It is an effective and efficient method to guarantee the correctness of data-centric service-based system. This research studies the problem of runtime monitoring of temporal behavior properties and puts forward a complete, effective and non-invasive on-line monitoring method, to dynamically analyze runtime temporal properties in service-oriented application which contains multiple participants.First of all, a temporal property description language is expounded in this article which is based on the SPS property pattern system and it is a Semi-formalmethod. Detailed explanation of each part is presented and two core propertys of the Roadside Parking System are descripted by this language as examples. Then, an existing Web services temporal property monitoring model is introduced, and after that, the improved monitoring model proposed by this article is detailed. The model proposes the idea of prediction model, including filtering, events distribution and exception handling. It can realize prediction and detection of monitored events of specific temporal properties. Hash table, prediction list, reversion list and other data structures are used to achieve quick positioning of the event being monitored in temporal property. The stale transition of finite automaton is used to express the changes of temporal properties. A set of algorithms is designed and implemented to completely descript the proposed temporal properties monitoring model. Finally, realizes the core functions of Roadside Parking System as test object. By implementing the proposed monitoring model and another existing method, test and analyze their running results in the actual cases. Judging from the analysis of experiment, the proposed method shows its excellent performance, quick responsing. and limited load to the original system and significant effect on temporal properties monitoring.