论文部分内容阅读
The optimal rate control problem in networks with unicast and multirate multicast sessions is investigated. A penalty function approach is used to solve a convex program formulation of this problem, and then a heuristic rate control algorithm is derived. The algorithm is distributed, and suitable both for source-driven unicast sessions and receiver-driven multicast sessions. To obtain practical viability, the computational burden on core routers as well as end-hosts is kept very low, also is the overhead of network congestion feedback. Simulation results show that the algorithm guarantees TCP (Transmission Control Protocol)-based unicast sessions coexisting with multirate multicast sessions in a fair and friendly manner. It is also shown that various fairness criteria of resource allocation could be achieved by choosing appropriate utility functions, and resource-utilizing efficiencies would be likewise different.
The optimal rate control problem in networks with unicast and multirate multicast sessions is investigated. A penalty function approach is used to solve a convex program formulation of this problem, and then a heuristic rate control algorithm is derived. The algorithm is distributed, and suitable both for source-driven unicast sessions and receiver-driven multicast sessions. The computational burden on core routers as well as end-hosts is kept very low, also is the overhead of network congestion feedback. Simulation results show that the algorithm guarantees TCP (Transmission Control Protocol) -based unicast sessions coexisting with multirate multicast sessions in a fair and friendly manner. It is also shown as various fairness criteria of resource allocation could be achieved by choosing appropriate utility functions, and resource-efficiency efficiencies would be likewise different.