Output details
11 - Computer Science and Informatics
University of Edinburgh
Cost-Sensitive Concurrent Planning Under Duration Uncertainty for Service-Level Agreements
<22> Originality: This was the first paper to apply planning to automate the configuration of service-oriented systems using the stochastic process algebra PEPA.
Significance: Service-level agreements (SLAs) are guarantees about probability of completion of work within a time-bound (e.g. responds within 2 seconds 99% of the time). When SLAs are not met, AI planning can be used to configure the system to meet the desired SLA. The Stochastic-POPF planner developed here is capable of solving such problems effectively using anytime search and a novel temporal-and-cost-sensitive heuristic.
Rigour: The paper appeared in the top conference on automated planning and scheduling.