| Summary: | 博士 === 國立清華大學 === 資訊工程學系 === 91 === As wireless technologies and Personal Communications System continue their widespread adoption and explosive growth, gradually increased attention has been focused on issues related to QoS provisioning for supporting high quality audio transmission, web content delivering, and other emerging multimedia applications in mobile environments. The key technical challenge that we want to address in the study of this dissertation is related to the design of Online QoS-controllable Scheduling Algorithm (OQSA), QoS-controllable Scheduling Mechanism (QSM), and their possible real-time applications for a wireless environment. The proposed on-line scheduling algorithms explicitly apply imprecise computation techniques trying to ensure that time-critical services can meet their deadlines and/or QoS requirements.
We attempt to design a coarse-and-fine adjustable QSM in the dissertation. The QSM performs a fine adjustment by employing various OQSA algorithms such as the proposed algorithms MOS, MOP, and MOF to meet different QoS and/or timing requirements. The proposed algorithms are optimal under the Feasible Mandatory Constraint. Meanwhile, the parameter k in the k-tasks-look-ahead substitutable check for the proposed optimal algorithms is designed to be adjustable in QSM to further enhance the overall schedulability. The coarse adjustment of a QSM is made by a regulator via evaluation process. However, we found that it would be difficult to construct a task regulator for the QSM so that the behavior of arrival tasks will meet the Feasible Mandatory Constraint.
In addition, we try to exploit the real-time applications of on-line QoS-controllable scheduling algorithms, such as the real-time processing in GSM-based software radios and QoS management in a LMPS. For the GSM-based software radios, a novel component-binding based architecture was proposed, and for the LMPS, a novel bridging model (Q-Bridge) was proposed for wireless LAN equipped devices. The goal of Q-Bridge design is to enhance both QoS and real-time delivering capabilities of an IEEE 802.11-based access point for a last mile solution. Also, we proposed two heuristic algorithms for the model and the preliminary simulation results were presented and discussed
|
|---|
