Abstract
This dissertation introduces several traffic management algorithms that represent the building blocks of a traffic management framework for stream multicasting in active networking environments. First, a multicast routing algorithm that can build the minimum cost multicast tree that is both end-to-end delay and inter-receiver delay-variation constrained is presented to provide a near optimal multicast tree cost. The algorithm can reduce the multicast tree cost by an average of 18% compared to the corresponding tree built using the best-known algorithm for building the constrained multicast tree. The algorithm is then extended to include active nodes in the multicast tree, which provide the networking environment required to manage the multicast traffic. Second, an algorithm that can maintain adherence to these constraints with the changing network dynamics is presented. The algorithm can adaptively correct the violations in the end-to-end delay as well as the inter-receiver delay variation constraints while minimizing the cost required for keeping these constraints within their limits. Third, a layer coordination mechanism is presented, which can implicitly coordinate the layer subscriptions in arbitrary non-cumulative layered multicasting environments. The coordination mechanism increases the subscription overlapping among the receivers in the multicast session by more than 35%, and at the same time decreases the number of joinings and leavings performed by each receiver for each rate change by 10% to 29% with almost no overhead on the network. Enhancements to this coordination mechanism are possible if there are active nodes in the multicast tree that can participate in the coordination strategy. Subsequently, several algorithms are introduced to further reduce the number of joinings and leavings required by the receivers for each rate change. Finally, a congestion control algorithm is introduced for addressing the congestion problem in non-cumulative layered multicasting environments. This algorithm takes into account the semantic information of the layered stream, which reduces the effect of the congestion on the perceived quality of the received stream. Also, it minimizes the effect of receivers intentionally ignoring the congestion control algorithm, which provides tighter control of the receivers, and a bounded TCP-friendly behavior.