A Time Petri Net-based Methodology for Embedded Hard Real-Time Software Synthesis

The problem to be addressed in this thesis is expressed in the following question:can a specification be translated into a computer program, in such a way that it executesin a group of processors with all specified constraints satisfied?This work considers embedded hard real-time systems development methodologies,more specifically, the software generation phase. Regarding real-time systems, thecorrect behavior depends not only on the integrity of the results, but also the time inwhich such results are produced. In hard real-time systems, if timing constraints arenot met, the consequences can be disastrous, including great damage of resources oreven loss of human lives.Nowadays, the human life has become more and more dependent of embedded sys-tems. This includes not only critical systems, such as automotive, railway, aircraft,spaceships and medical devices, but also, household appliances, network printers, au-tomatic teller machines, cellular telephones, among others. Due to this great diversityof applications, the design of embedded systems can be subject to several kinds of dif-ferent constraints, including timing, size, weight, energy consumption, reliability, andcost. An alternative to treat with such problem is the adoption of formal methods.Such methods are important mechanisms for the analysis and verification of properties,as well as, facilitate system validation. However, for the effective use of formalisms,the availability of automatic tools for attending the designer is an important issue andneeds to be considered.In this thesis, the software synthesis takes a specification (composed of concur-rent and communicating tasks) and automatically generates a program source codeconsidering: (i) functionalities and constraints; and (ii) operational support for task’sexecution. Usually, complex systems adopt a general-purpose operating system to sup-port the software execution. However, this solution is excessively general and mayintroduce delays in the execution time. Moreover, it produces a higher rate of memoryusage. The software synthesis is a design alternative for these drawbacks. This methodautomatically generates the program source code, satisfying the desired functionality,the specified constraints, the runtime support, and the minimization of both delaysand memory consumption.The embedded software synthesis has been receiving much attention. However, fewworks treat with software synthesis for hard real-time systems considering arbitraryprecedence and exclusion relations. Code generation for meeting all timing and resourceconstraints is not a trivial task. Thus, this research area has several open issues, mainlyrelated to generation of predictable-guaranteed scheduled code.