Validity of QoS Requirements for Application Using Probabilistic Model Checking

Abstract

In the field of computer applications when be there need for design application to present certain service for specific customer, this application must design in care of how satisfy customer needs, therefore any application must evaluate if satisfy all customer needs or has violation in some needs. Customer needs is customer requirements, and customer satisfaction level sometimes called Quality of Services (QoS), so the customer requirements are main factor use to evaluate QoS. The customer requirements differ according to the type of service. The study was initiated to solve the problem of validity of QoS requirements for application, it is to evaluate if the application satisfy user requirements or not, the user requirements is considered as QoS requirements because it represented according to agreements between customer and service provider which called service level agreement (SLA), the violation in one of QoS requirements means there defect in application, and must redesign of application. To validate QoS requirements the study use probabilistic model checking, this technique depends on probabilistic techniques and model checker software. The probabilistic technique was used to represent system states and their requirements which represented as probability values, the system states represented using Markov chains, these states is represented according to system navigation, mean when system transfer from certain state to another and probability of navigation. The system quality requirements represented using Discrete Time Markov Chain (DTMC) and Continuous Time Markov Chain (CTMC) which specify according to time state, the requirements represented in DTMC transform to Probabilistic Computation Tree Logic (PCTL) formula, and another one transform to Continuous Stochastic Logic (CSL) formula. Then use software PRISM as model checker, in software represent system states and probabilities in modules using model language, and then run module for each one to gain results which specify if the system achieve user satisfaction or not. In this study used the web application used as case study and after follow proposed solution, the outputs of experiments clarify that some of requirements were violated and then the system need to redesign to achieve QoS requirements.