Training students in parallel and distributed thinking by completion of a computer control lab project

Thinking in parallel and distributed categories is imperative for todays computer applications. It is shown how students train this way of thinking by carrying out a laboratory project. Students develop a software system for the computer control of a model plant. The control maximizes the throughput of produced parts. The problem is analyzed using the process flow graph. The analysis returns a structure of process actions which partly run in parallel, partly sequentially. This structure can be mapped into a task system. Basic task functions like synchronisation and communication are solved with mailboxes and semaphores. At first a manual control is implemented, in which basic hardware and task functions are tested. In the next two steps the tasks are integrated into an automatic-mode task system. Both decentralized and hierarchical task organizations are considered. The design of parallel, distributed and realtime task systems provides students with valuable experience and a basic understanding of modern software architectures