Decentralized decision-making for task reallocation in a hard real-time system

A decentralized task reallocation algorithm for hard real-time systems is developed and analyzed. The algorithm, which is fast and reliable, specifically considers deadlines of tasks, attempts to utilize all the nodes of a distributed system to achieve its objective, handles tasks in priority order, and separates policy and mechanism. An extensive performance analysis of the algorithm by means of simulation shows that it is quite effective in performing reallocations and that it is significantly better than a centralized approach.<
>