Document Type

Technical Report

Publication Date






Technical Report Number



An increasing number of distributed real-time embedded systems face the critical challenge of providing Quality of Service (QoS) guarantees in open and unpredictable environments. For example, such systems often need to enforce CPU utilization bounds on multiple processors in order to avoid overload and meet end-to-end dead-lines, even when task execution times deviate significantly from their estimated values or change dynamically at run-time. This dissertation presents an adaptive QoS control framework which includes a set of control design methodologies to provide robust QoS assurance for systems at different scales. To demonstrate its effectiveness, we have applied the framework to the end-to-end CPU utilization control problem for a common class of distributed real-time embedded systems with end-to-end tasks. We formulate the utilization control problem as a constrained multi-input-multi-output control model. We then present a centralized control algorithm for small or medium size systems, and a decentralized control algorithm for large-scale systems. Both algorithms are designed systematically based on model predictive control theory to dynamically enforce desired utilizations. We also introduce novel task allocation algorithms to ensure that the system is controllable and feasible for utilization control. Furthermore, we integrate our control algorithms with fault-tolerance mechanisms as an effective way to develop robust middleware systems, which maintain both system reliability and real-time performance even when the system is in face of malicious external resource contentions and permanent processor failures. Both control analysis and extensive experiments demonstrate that our control algorithms and middleware systems can achieve robust utilization guarantees. The control framework has also been successfully applied to other distributed real-time applications such as end-to-end delay control in real-time image transmission. Our results show that adaptive QoS control middleware is a step towards self-managing, self-healing and self-tuning distributed computing platforms


Permanent URL: