A predictive algorithm for adaptive resource management of periodic tasks in asynchronous real-time distributed systems

We present a "predictive" resource management algorithm for periodic tasks in real-time distributed applications that are characterized by significant execution-time uncertainties. The algorithm is predictive in the sense that it forecasts the timeliness behavior of the tasks during the resource allocation process and select allocations that yield the optimal forecasted timeliness. The algorithm uses statistical regression theory for predicting task timeliness. The performance of the predictive algorithm is studied by comparing with a nonpredictive resource management algorithm that uses heuristic rules for allocating resources. The experimental results indicate that the predictive algorithm outperforms the non-predictive algorithm when the workload shows fluctuating behavior.