Software Engineering Problems and Solutions in Cyber – Physical Systems (Special Session)

Cyber-physical systems. For more than 25 years extensive research and development efforts had been devoted to problems in real-time, safety-critical, ad embedded systems. Security issues were considered crucial in mission-critical systems. While all such constraints had been derived as design problems in software control systems for real-world applications they been treated as separate aspects since each of them proved even difficult enough when studied separately. While this resulted gradually in nearly different scientific disciplines it turned out to be increasingly inadequate for the construction of large control systems, in particular for applications with (partially) autonomous entities and decision-making: As a well-known example, optimal real-time performance may well be adverse to optimal overall throughput, and vice versa.

In order to take all constraints equally into account we have therefore, in 2008, defined the concept of cyber-physical systems by requesting that the structures and relationships the software and physical layers be as similar as possible, ideally even isomorphic or congruent. (In an easy example, if the physical system exhibits a certain amount of autonomy for the actors, say in a large traffic system, then the control system should exhibit an explicit concept of distributed control.)

Traditional approaches for modeling and analyzing self-organizing process systems (e.g. optimization or other classical theoretical methods) normally fall short concerning efficient solutions in practice. They often do not really scale: With the system size growing the computation time may well explode (e.g. for problems that can be reduced to TSM), and once there are requirements such as real-time conditions involved the solutions exhibit a practically very cumbersome, if not inadequate, timing behavior.

In several novel application areas like Smart Power Grids, Renewable Energy and power management, routing in today’s and future Internet applications, Transportation Planning or Traffic Control systems in Logistics novel technological developments have posed a class of novel cross – disciplinary R&D where modeling, implementation, and analyzing require a very close collaboration between researchers and developers in the disciplines involved. Solutions have been pursued, and even implemented, which borrow principles from Natural Computing or from Swarm Intelligence. Also, they replace conventional approaches issuing exact solutions which are practically irrelevant, through iterative methods which are faster to obtain and at he same time of top quality (Self-Organizing Maps). Open problems such as on-line organization, consistency, and availability of mobile and dynamic geographic data (in Outer Space, in wide-spread maritime operations etc.) have been identified as particular candidates for self-organizing research (among many others).

For the proposed special session papers are solicited which display, and thoroughly discuss, the various aspects addressed above, bringing researches and engineers from academia and industry together for extensive discussions.

Session Organizers

Wedde Horst Wedde, Informatik III, Technical University of Dortmund/Germany

Program Committee