Descartes-Agent: Verifying Formal Specifications Using the Model Checking Technique

Abstract

The field of software engineering is seeking ways to close the gaps that exist between the phases of software development, right from requirements specification until testing to deliver reliable software systems. The concepts discussed in this paper intends to reduce that gap using formal methods. This paper introduces new methods and tools that potentially would significantly reduce the time and cost of developing software systems while increasing their reliability. This paper describes a methodology using formal methods to verify software specifications. The use of formal methods during the different phases of a software development life cycle has proven advantages of resulting in a reliable software system. Understanding the importance of using formal methods to specify software systems and then to verify the specifications for correctness forms the first sub problem. Tools and framework needed to formally verify software specifications written for agent systems is also discussed in this paper. This research explores using the model checking methods along with the executable Descartes – Agent specifications to provide a basis for formal verification of agent systems and, generally reliable systems. The Descartes – Agent grammar previously developed was used to implement an algorithm that converts the agent specifications into an intermediate form in linear temporal logic form. Popular model checking tools such as Promela and SPIN were also used to provide a complete framework of verifiable formal specifications.