An adaptive algorithm for Internet multimedia delivery in a DiffServ environment.

Date of Award


Degree Type


Degree Name



Computer Science


Computer Science.




To meet the Quality of Service (QoS) requirements of multimedia applications and to reduce the network congestion, several service models and mechanisms have been proposed. Among these, Differentiated Service (DiffServ) architecture has been considered as a scalable and flexible QoS architecture for the Internet. DiffServ provides class-based QoS guarantees. Applications in different classes receive different QoS and are priced differently. If network congestion occurs, DiffServ may not be able to guarantee the QoS for the application. Thus, the QoS may not reflect the price paid for the service. A problem of considerable economic and research importance is how to achieve a good price and quality tradeoff even at times of congestion. This thesis presents an Adaptive Class Switching Algorithm (ACSA) which intends to provide good quality with good price for real-time multimedia applications in a DiffServ environment. The ACSA algorithm combines the techniques of Real-time Transport Protocol (RTP), DiffServ, and Adaptation together. It also takes both QoS and price into account to provide users a good QoS with a good price. The algorithm dynamically selects the most suitable class based on both the QoS feedback received and the highest user utility. The user utility is a function of quality, price, and the weight which reflects the relative sensitivity to quality and price. The class with the highest user utility is the class that provides the best quality and price tradeoff. The QoS feedback is conveyed by RTP's Control Protocol (RTCP) Receiver Reports. The results of simulation demonstrate that ACSA can react fast to the current class state in the network and reflects the best QoS and price tradeoff. It always seeks to find a class which provides the highest user utility except when the Internet is congested and the required QoS in all classes can not be satisfied. If this happens, the real-time multimedia flow chooses Best-Effort class with no payment. (Abstract shortened by UMI.) Paper copy at Leddy Library: Theses & Major Papers - Basement, West Bldg. / Call Number: Thesis2005 .F46. Source: Masters Abstracts International, Volume: 44-01, page: 0389. Thesis (M.Sc.)--University of Windsor (Canada), 2005.