Stochastic feedback-based control of QoS in soft real-time systems

This paper investigates application of feedback based control mechanisms to the problem of scheduling soft real-time tasks on a shared CPU. Each task has an execution time stochastically varying in time and it has to be provided with a specified level of Quality of Service (QoS). The problem of feedback control is formalised in the stochastic domain, by expressing QoS requirements in terms of properties to be satisfied by the stochastic process describing the evolution of the system state. We present several control schemes and address the fundamental problem of stochastic stability. Experimental results collected by a modified version of the Linux operating system show the effectiveness of the approach and its practical feasibility.