| The studies on network congestion and bandwidth provisioning are often formulated into the Network Utility Maximization (NUM) framework. This framework is characterized by a predetermine network topology and fixed network capacity, with an objective to allocate bandwidth to users while maximizing the aggregated user's satisfaction subjects to network capacity constraints. This dissertation primarily addresses challenges in resolving problems in NUM while considering user demand and quality of service (QoS). We propose a new user utility model that incorporates QoS to support elastic traffic (such as email, ftp, http, etc.) and real time traffic (for e.g., VoIP, etc.). Further, our proposed user utility function also supports multimedia applications, such as video streaming, where the data from these applications is often distributed into various layers of different qualities. Recognizing that finding an optimal solution that maximizes user utility for real-time traffic may require non-negligible amount of execution time given that this is an NP-hard problem, we propose a market- informed congestion control, i.e. a sub-optimal solution to speed up the process of resolving network congestion. Following this, we investigate the consequences of suboptimal solution in real time environment. One that is extensively discussed is user's self-regulation and abandonment of the network when the network is not able to meet their demand for minimum bandwidth or QoS. To address this, we propose network admission control scheme to select a group of users that the network can support. Finally, we discuss the practical aspects and the consequences in implementing the solutions for NUM. |
