Multimedia applications over the Internet are getting more and more popular. While non-real-time streaming services, such as YouTube and Megavideo, are attracting millions of visiting per day, real-time conferencing applications, of which some instances are Skype and Yahoo Voice Chat, provide an interesting experience of communication. Together, they make the fancy Internet world become more and more amusing. Undoubtedly, multimedia flows will eventually dominate the computer network in the future. As the population of multimedia flows increases gradually on the Internet, quality of their service (QoS) is more of a concern. At the moment, the Internet does not have any guarantee on the quality of multimedia services. To completely surpass this limitation, modifications to the network structure is a must. However, it will take years and billions of dollars in investment to achieve this goal. Meanwhile, it is essential to find alternative ways to improve the quality of multimedia services over the Internet. In the past few years, many endeavors have been carried on to solve the problem. One interesting approach focuses on the development of end-to-end congestion control strategies for UDP multimedia flows. Traditionally, packet losses and delays have been commonly used to develop many known control schemes. Each of them only characterizes some different aspects of network congestion; hence, they are not ideal as feedback signals alone. In this research, the flow accumulation is the signal used in feedback for flow control. It has the advantage of reflecting both packet losses and delays; therefore, it is a better choice. Using network simulations, the accumulations of real-time audio applications are collected to construct adaptive flow controllers. The reason for choosing these applications is that they introduce more control challenges than non-real-time services. One promising flow control strategy was proposed by Bhattacharya and it was based on Model Predictive Control (MPC). The controller was constructed from an ARX predictor. It was demonstrated that this control scheme delivers a good QoS while reducing bandwidth use in the controlled flows by 31 percent to 44 percent. However, the controller sometime shows erratic response and bandwidth usage jumps frequently between lowest and highest values. This is not desirable. For an ideal controller, the controlled bandwidth should vary near its mean. To eliminate the deficiency in the strategy proposed by Bhattacharya, it is proposed to introduce a feed forward term into the MPC formulation, in addition to the feedback terms. Simulations show that the modified MPC strategy maintains the benefits of the Bhattacharya strategy. Furthermore, it increases the probability of bandwidth savings from 58 percent for the case of Bhattacharya model to about 99 percent for this work.

Besides the two classical techniques used to evaluate the performance of a protocol, computer simulation and experimental measurements, mathematical modeling has been used to study the performance of the TCP protocol. This technique gives an elegant way to gain insights when studying the behavior of a protocol, while providing useful information about its performance. This thesis presents an analytical model for the SF-SACK protocol, a TCP SACK based protocol conceived to be appropriate for data and streaming applications. SF-Sack modifies the multiplicative part of the Additive Increase Multiplicative Decrease of TCP to provide good performance for data and streaming applications, while avoiding the TCP-friendliness problem of the Internet. The modeling of the SF-SACK protocol raises new challenges compared to the classical TCP modeling in two ways: first, the model needs to be adapted to a more complex dynamism of the congestion window, and second, the model needs to incorporate the scheduler that SF-SACK makes use of in order to maintain a periodically updated value of the congestion window. Presented here is a model that is progressively built in order to consider these challenges. The first step is to consider only losses detected by triple-duplicate acknowledgments, with the restriction that one such loss happens each scheduler interval. The second step is to consider losses detected via triple-duplicate acknowledgments, while eliminating the above restriction. Finally, the third step is to include losses detected via time-outs. The result is an analytical characterization of the steady-state send rate and throughput of a SF-SACK flow as a function of the loss probability, the round-trip time (RTT), the time-out interval, and the scheduler interval. The send rate and the throughput of SF-SACK were compared against available results for TCP Reno. The obtained graphs showed that SF-SACK presents a better performance than TCP. The analytical model of the SF-SACK follows the trends of the results that are presently available, using both the ns-2 simulator and experimental measurements.