| Summary: | 博士 === 國立臺灣科技大學 === 電子工程系 === 92 === Communication networks beyond the third generation (3G) are enabling full coverage of all-IP services for users anywhere and on the move. End-to-end quality of service (QoS) provision is one of the major challenges in the design of such systems and must be supported by the appropriate solution of applications, networks, and wireless platforms. In the past, circuit-switched service provides well-defined QoS for end-to-end users once the connection is established via the call admission control, but the bandwidth utilization is not optimized. In order to advance the overall system efficiency, packet-switched service was proposed to fulfill the goal of maximized bandwidth utilization. Besides call admission control for establishing the connection for the users, both rate regulation and rate scheduling mechanisms are required to be implemented in the transmission system with respect to the guaranteed QoS.
This thesis proposes a QoS implementation model in terminal nodes, edge nodes and switch nodes thus the service advantages of Internet and multimedia transmissions are made possible. The proposed QoS implementation model is presented in terms of three control mechanisms. First, the call admission control makes the decision of establishing a connection over the network. The call admission control is based on prioritized resource reservation associated with QoS violation assessment, and gives a viable parameter-regulation approach by using resource-starvation analysis and estimation, thus solving the problem of QoS violations. Second, once the call is admitted, a rate control with multi-level queueing is adapted to ensure that the traffic entering the network receives its negotiated QoS and does not adversely affect other traffic. The multi-level queueing is implemented to enable the negotiation of QoS, the traffic input rates are regulated by scaling the quantization parameter according to the feedback of monitoring the input queue of each connection. Third, an interarrival-time based rate scheduling scheme is proposed to reduce the burstiness of the traffic streams and to eliminate the delay jitter in multimedia communications. A time-frame windowing mechanism is designed within the rate scheduling scheme to further improve the rate smoothness, interdeparture delay and bandwidth utilization.
In the call admission model, resource reservation is made through dynamic sharing according to the index of QoS violation assessment. Thus, the system can provide guaranteed resource reservation for users in advance to avoid resource starvation from the bursty arrivals of instantaneous users. After the call requests are accepted, the MLQ rate control provides rate regulation and QoS negotiation according to the scalability of QoS parameters, so that the bandwidth utilization is maximized and fairly shared among video, voice and data classes of traffics. Finally, the end-to-end connections are established by call admission control and associated with the MLQ rate control and the cascade ITWFQ rate scheduling, therefore, the system results the performance in terms of less delay jitter and smoothness of transmission.
The implementation model not only supports the development of QoS negotiation but serves as a formal base for a structured system analysis. We describe the experimental results and queueing analysis methods which are adapted to analyze call blocking, rate renegotiation, packet delay and interdepartures. Simulation and numerical results are also provided to verify the system performance of the model.
|
|---|
