TOOL-ASSISTED VALIDATION AND VERIFICATION TECHNIQUES FOR STATE-BASED FORMAL METHODS

To tackle the growing complexity of developing modern software systems that usually have embedded and distributed nature, and more and more involve safety critical aspects, formal methods (FMs) have been affirmed as an efficient approach to ensure the quality and correctness of the design, that permits to discover errors yet at the early stages of the system development. Among the several FMs available, some of them can be described as state-based, since they describe systems by using the notions of state and transitions between states. State-based FMs are sometimes preferred since they produce specifications that are more intuitive, being the notions of state and transition close to the notions of program state and program execution that are familiar to any developer. Moreover, state-based FMs are usually executable and permit to be simulated, so having an abstraction of the execution of the system under development. The aim of the thesis is to provide tool-assisted techniques that help the adoption of state-based FMs. In particular we address four main goals: 1) identifying a process for the development of an integrated framework around a formal method. The adoption of a formal method is often prevented by the lack of tools to support the user in the different development activities, as model editing, validation, verification, etc. Moreover, also when tools are available, they have usually been developed to target only one aspect of the system development process. So, having a well-engineered process that helps in the development of concrete notations and tools for a FM can make FMs of practical application. 2) promoting the integration of different FMs. Indeed, having only one formal notation, for doing different formal activities during the development of the system, is preferable than having a different notation for each formal activity. Moreover such notation should be high-level: working with high level notations is definitely easier than working with low-level ones, and the produced specifications are usually more readable. This goal can be seen as a sub-goal of the first goal; indeed, in a framework around a formal method, it should also be possible to integrate other formal methods that better address some particular formal activities. 3) helping the user in writing correct specifications. The basic assumption of any formal technique is that the specification, representing the desired properties of the system or the model of the system, is correct. However, in case the specification is not correct, all the verification activities based on the specification produce results that are meaningless. So, validation techniques should assure that the specification reflects the intended requirements; besides traditional simulation (user-guided or scenario-based), also model review techniques, checking for common quality attributes that any specification should have, are a viable solution. 4) reducing the distance between the formal specification and the actual implementation of the system. Several FMs work on a formal description of the system which is assumed to reflect the actual implementation; however, in practice, the formal specification and the actual implementation could be not conformant. A solution is to obtain the implementation, through refinements steps, from the formal specification, and proving that the refinements steps are correct. A different viable solution is to link the implementation with its formal specification and check, during the program execution, if they are conformant.