Skip to Main Content
This paper provides an application of a consensus problem with nonlinear performance functions to resource allocation in soft real-time systems. In soft real-time systems, fairness of QoS (Quality of Service) levels plays an important role to guarantee appropriate service under timing and resource constraints. We propose fair QoS control with an agent-based controller where each agent manages an allocated resource and a QoS level of each task. We derive an adaptive resource allocation algorithm and sufficient conditions to achieve fair QoS levels by extending results of a consensus problem.