文摘
We consider the problem of incorporating control tasks on top of a partially loaded shared computing resource, whose current task execution pattern is characterizable using a window based pattern. We consider that the control task to be scheduled is allowed to switch between multiple controllers, each with different associated sampling rate, in order to adjust its requirement of computational bandwidth as per availability. We provide a novel control theoretic analysis that derives a Timed Automata (TA) based specification of allowable switchings among the different controller options while retaining the asymptotic stability of the closed loop. Our scheduling scheme computes a platform level residual bandwidth pattern from individual task level execution patterns. We then leverage the TA based controller specification and the residual bandwidth pattern in order to synthesize a Linearly Priced Timed Automata for which the minimum cost reachability solution provides realizable multi-rate control schedules. The provided scheduler not only guarantees the asymptotic stability of the control loop but also increases the robustness and control performance of the implementation by maximizing the bandwidth utilization.