| Summary: | The emerging high-rate wireless personal area network (WPAN) technology is capable of
supporting high-speed and high-quality real-time multimedia applications. In particular,
video streams are deemed to be a widespread traffic type, and require quality of service
(QoS) support. However, in the current IEEE 802.15.3 standard for MAC (media access
control) of high-rate WPANs, the implementation details of some key issues such as
scheduling and QoS provisioning have not been addressed. Moreover, the hierarchical
structure of video streams calls for special measures at the MAC layer in order to improve
the QoS.
In the first part of this thesis, we propose a frame-decodability aware (FDA) technique
to make the scheduling algorithms aware of the hierarchical structure and decoding
dependencies in video streams. Simulation results show that the FDA technique can significantly
improve the performance of F-SRPT [1] and EDD+SRPT [2] schedulers by up
to 61% and 60%, respectively. We also compare two common performance metrics and
investigate which one is a more accurate indicator of the QoS given to video streams.
In the second part of this thesis, we first propose a mathematical model for the optimal
scheduling scheme for video flows in high-rate WPANs. Using this model, we then propose a scheduler that incorporates reinforcement learning (RL). Simulation results show that
our proposed scheduler is nearly optimal and performs 42%, 49%, and 53% better than
EDD+SRPT [2], PAP [3], and F-SRPT [1] schedulers, respectively. === Applied Science, Faculty of === Electrical and Computer Engineering, Department of === Graduate
|
|---|
