Robust allocation and scheduling heuristics for dynamic, distributed real-time systems

Abstract

A challenge facing real-time computing is the need to deploy real-time systems in dynamic operational environments. The systems have explicit deadline requirements, but their execution times are often affected by unpredictable environmental inputs that cannot be known a priori and have no worst-case estimates. As a result, traditional real-time task allocation and scheduling techniques do not apply. This research proposes a new task allocation and scheduling approach for these dynamic, distributed real-time systems. The approach offers these systems explicit real-time guarantees as well as maximized tolerance (robustness) of unpredictable changes in environmental inputs. This work consists of (1) a real-time computing model that incorporates environmental factors, (2) metrics that characterize robustness, and (3) algorithms that find robust allocations with feasible schedules for local schedulers. Analytical bounds were derived to guarantee the performance of the algorithms. The work produces a dependable foundation for task allocation and scheduling so that real-time systems may be designed and deployed for many time-critical but unpredictable real world environments. © 2008 Springer-Verlag Berlin Heidelberg.