Real-Time Resource Management for End-to-End Deadlines

• Main Authors: Chin-Fu Kuo, 郭錦福
• Other Authors: Tei-Wei Kuo
• Format: Others
• Language: en_US
• Online Access: http://ndltd.ncl.edu.tw/handle/63302792327672519565

博士 === 國立臺灣大學 === 資訊工程學研究所 === 93 === In recent years, many modern component-oriented distributed systems, such as phased array radars, are built with commercial-off-the-shelf components, and the functions of many hardware components are also re-implemented by software modules. In such systems, tasks could be modelled as distributed real-time tasks which require end-to-end deadline guarantees and have precedence constraints. In this thesis, we propose a two stage scheduling algorithm with emph{probabilistic timing guarantees} and a hierarchical emph{resource allocation} framework with no totally ordered importance or numeric utility value for every task. We take component-oriented radar systems as a case study to demonstrate the feasibility of our work. Different from most previous work on either algorithms with restrictions in resource utilization or heuristics without analytical ways for schedulability guarantees, we propose a joint real-time scheduling algorithm for both front-end and back-end processor workloads with an analytical framework for off-line probabilistic analysis and on-line admission control. We also develop a hierarchical resource allocation framework. This approach consists of two phases: In the off-line phase, the algorithm computes the sub-optimal resource configuration assignments for few workload states by a dynamic programming approach; in the on-line phase, the algorithm computes resource reconfiguration assignments by a greedy approach, provided that sub-optimal resource reconfiguration assignments for several selected workload states are computed in the off-line phase. Beside component-oriented radar systems, the ideas presented in this thesis could be applied to the designs of many component-oriented distributed systems.