Permutational genetic algorithm for the optimized mapping and scheduling of tasks and messages in distributed real-time systems

The mapping of tasks and messages and the assignment of fixed priorities in distributed real-time systems are known to be NP-hard problems, and thus there are no optimal methods to accomplish them in polynomial time. This fact makes them suitable problems to be approached with generic search and optimization algorithms. In this paper we propose a genetic algorithm with a permutational solution encoding, which apart from solving the mapping and the fixed priority assignment problems using a holistic approach, can simultaneously minimize the average use of computing, memory and communication resources, the average worstcase response time of the transactions and the number of the used processors. The experimental results show that this genetic algorithm can find good solutions for industrial size distributed real-time architectures and in reasonable times from the perspective of a complex system design process. Keywords-distributed real-time; holistic; mapping; scheduling; priority assignment; genetic algorithm; linear programming