THESIS
1998
xv, 93 leaves : ill. ; 30 cm
Abstract
Given the popularity of multicast services and applications, we study the problem of state dependent multicast call routing for single rate loss networks. We formulate the problem and discuss the difficulties to implement the optimurn solution. Because of these difficulties, a heuristic approach is proposed: we apply Minimum Spanning Tree (MST) searching with a suitable link cost function to search a connected tree for a multicast call connection request such that its normalized revenue loss can be reduced. If there is no connected tree available, the call connection request will be rejected....[
Read more ]
Given the popularity of multicast services and applications, we study the problem of state dependent multicast call routing for single rate loss networks. We formulate the problem and discuss the difficulties to implement the optimurn solution. Because of these difficulties, a heuristic approach is proposed: we apply Minimum Spanning Tree (MST) searching with a suitable link cost function to search a connected tree for a multicast call connection request such that its normalized revenue loss can be reduced. If there is no connected tree available, the call connection request will be rejected.
We develop analytical models of Least Load Multicast Routing (LLMR), which is a well-known multicast routing algorithm, for fully connected networks. The analytical models that we develop for calculating blocking probabilities are based on the Reduced Load Approximation (RLA) with the link independencle assumption. For symmetrical networks, our analytical models include both state aggregation, i.e. ALLMR, and alternative routing for point-to-point calls. The agreements between the simulation and analytical results in both scenarios are very good and we find that the agreements are not affected significantly by the link independence assumption.
Four new link cost functions are proposed to improve the network performance in different ways: Aggregated Least Load Multicast Routing (ALLMR) gives a simpler implementation, lower signaling traffic for establishing conncction requests and lower sensitivity to the design parameters, compared with LLMR. Least Load Multicast Routing with Maximum Occupied Circuits (LLMRMOC) and Least Load Multicast Routing with Minimum Measured Blocking Time (LLMRMMBT) are modified LLMR algorithms which have moderate improvement with minimum additional cost, compared with LLMR. We also discuss their implementation issues. The Maximum Mean Number of New Calls Accepted Before Blocking Multicast Routing (MCBMR) algorithm can more accurately capture the current and future loading of a network. Simulation results show that this algorithm, compared with LLMR, not only has a smaller network revenue loss, but also results in smaller call blocking probabilities for all classcss of traffic. We also discuss the implementation issues of this proposed algorithrn and develop two approximation methods, state approximation and curve fitting, which can reduce the measurement complexity significantly with only a slight performance degradation.
Post a Comment