End-to-End Utilization Control for Aperiodic Tasks in Distributed Real-Time Systems

An increasing number of DRTS (Distributed Real-Time Systems) areemploying an end-to-end aperiodic task model. The key challenges of suchDRTS are guaranteeing utilization on multiple processors toachieve overload protection, and meeting the end-to-end deadlines ofaperiodic tasks. This paper proposes an end-to-end utilization controlarchitecture and an IC-EAT (Integration Control for End-to-EndAperiodic Tasks) algorithm, which features a distributed feedback loopthat dynamically enforces the desired utilization bound on multipleprocessors. IC-EAT integrates admission control with feedbackcontrol, which is able to dynamically determine the QoS (Quality ofService) of incoming tasks and guarantee the end-to-end deadlines ofadmitted tasks. Then an LQOCM (Linear Quadratic Optimal Control Model) ispresented. Finally, experiments demonstrate that, for the end-to-endDRTS whose control matrix \pmb G falls into the stable region, theIC-EAT is convergent and stable. Moreover, it is capable of providingbetter QoS guarantees for end-to-end aperiodic tasks and improving thesystem throughput.