Advanced Technology Transfer in Software Engineering

The growing diffusion of the Information and Communication Technology among any everyday assets and services is nowadays referred to as the "Pervasive ICT". The term "pervasive" well represents the trend that very soon hundreds of billions of microprocessors will be embedded within objects that do not even appear as ICT products and will govern in practice every aspect of our lives. This true revolution is progressively changing every business and the very way companies exist and compete. In any case, the software will actually constitute the core of any product and confirms itself as the element that is responsible for the satisfaction of the quality requirements of the services and the assets that the modern society calls for. Research in computer science has long recognized the challenges that this technological evolution places on the production of the software; if on one side the latter must be able to satisfy stringent quality requirement, above all in terms of correctness and performances, on the other it becomes more and more complex and liable to increasingly sophisticated attacks and failures. The discipline of Software Engineering answers today to such challenges with structured process models and rigorous, adaptive and automated methodologies of specification, coding and verification. Special care is devoted to the aspects relative to validation and verification, that permeate all phases of product engineering, with the aim of guaranteeing a development that is suitable to the user requirements since the very first stages of specification and design. Advances in such direction are continuous and quite fast, so that the natural gap between practice and science risks to grow in the absence of a continuous process of collaboration and knowledge transfer between academy and industry.
At ISTI, where expertise and research projects in the software domain are many-sided and long established, cooperation with industry is an essential component of research. As of today, in addition to the many projects funded by the Italian Ministers and the European Community, several agreements for direct collaboration are in place with important national and international companies, that stand out in the contemporary situation of limited private investment in research.
We briefly report on a few examples of main successes. The collaboration with "Fiat Auto" is aimed at the identification and the improvement of the level of capability of their suppliers, and includes the monitoring and the improvement of software products through the evaluation of the software process. The activity is based on the screening of the technological capabilities of the companies (Italian and above all foreign) that supply the software and the electronic control units, which are today embedded into the car in high number (up to 70 - 80 for vehicle). The activity started in 2000: in a first phase the assessment of the software development process of some reference suppliers of Fiat Auto has been carried on. Based on the results of such assessments, and of the consequent identification of the critical aspects tied to the software supplies, a reference model and an expected capability profile has been defined, that has been then used in order to submit to evaluation all suppliers, with a methodology shared by all European car manufacturers. So far more than 20 software suppliers of Fiat Auto in Europe and in the United States (among which Delphi, TRW, Marelli Magnets, Bosch, Siemens) have been assessed. The results of such assessments and the approach defined are now integral part of the process of management of the software suppliers by Fiat Auto.
The collaboration with Alstom started in 2003 and is finalized to the study and the realization of a formal development process that can be used for modelling a system of railway signaling, executing on the model all the simulations needed to guarantee its correctness and finally transferring on one hardware platform the code automatically generated. In particular the collaboration between ISTI and Alstom Ferroviaria S.P.A. is aimed at finding solutions for the modeling of interlocking systems in the railway field, using engineered formal specification environments: the purpose of these experiments is to establish a highly automated development process in which apparatuses of railway signaling, which are specific to each national reality, can be directly implemented from their formal description.
Finally, in 2001 the joint laboratory Pisatel (Pisa Initiative on Software Architectures for Telecommunications) between ISTI and Marconi/Ericsson Lab Italy has been established. It also involves collaborations with the University of Pisa and Scuola Sant'Anna. The initiative addresses research and education on software themes of common interest, which are articulated in yearly reviewed projects mainly carried on through the recruitment of young researchers: several grants have been now successfully completed, in addition to six PhDs. The special atmosphere of open collaboration that reigns within the laboratory, not constrained by pressing contractual obligations, but however in line with concrete requirements, has concurred to address advanced themes of frontier between academy and industry from a privileged perspective. The focus of past activities has included advanced methodologies for software testing, real-time systems, and efficient methods of video transcoding for the mobile telephony. The currently faced challenge consists in the identification of effective tools for model-driven engineering (that is the development of software guided by rigorous models that automatically transformed into executable code, thus preventing the problems and the costs of manual programming) that support the early analysis of quality and performance.