This paper analyses the schedulability of real-time systems with uncertain knowledge. Fuzzy numbers are used to model parameters describing tasks, providing information about observation of the tasks deadlines at early phases of development. In this paper fuzzy schedulability and a fuzzy optimality criteria are defined and the fuzzy optimality of Deadline Monotonic Scheduling algorithm is proved. It is shown that conventional crisp schedulability analysis can be extended applying both fuzzy arithmetic and interval arithmetic to compute the fuzzy schedulability considering fuzzy execution times, periods, deadlines, blocking times, release jitter, and the existence of aperiodic servers.
Keywords: Real-time systems, uncertainty, fuzzy scheduling, static priority assignment.
International Conference IMPMU'2002 Congress. Annecy, France. 1-5 de Julio 2002
Publicado: julio 2002.