Hwang, R.-H. Routing in High Speed Networks. Ph.D. Dissertation Prospectus, April 1992.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....are blocked and a call is blocked if the origin node is blocked. If a blocking message is fed back to the upstream node, the circuit reserved on the link from the upstream node to the node is released. A detailed survey of existing routing algorithms for circuit switched networks is given in [49]. We briefly list below some well known routing algorithms. The oldest and most extensively used static routing algorithm in telephone networks is the hierarchical alternate routing algorithm [9, 76] Because of the inefficiency of static routing, a dynamic routing algorithm, Dynamic ....

....delay per packet under high network load or (b) decrease the average packet delay for the same level of network throughput under low network load. 16 During the last three decades, many routing algorithms for wide area networks have been proposed. A survey of those routing algorithms is given in [49]. In this survey, due to the great diversity in existing sophisticated adaptive routing algorithms, adaptive routing algorithms are further classified according to the following two ways. First, according to the route selection mechanism, adaptive routing algorithms are further classified into ....

Hwang, R.-H. Routing in High Speed Networks. Ph.D. Dissertation Prospectus, April 1992.

....of the four routing policies. Section 6 concludes our study. 2 The MDP Approach Towards Routing In this section, we first briefly describe a formulation of the routing problem in multirate networks as a continuous time Markov decision process [8] The detailed formulation can be found in [11, 18]. We then show how the link independence assumption can be used to decompose the resulting Markov decision process into separate link level Markov processes. 2.1 Network Model Consider a network consisting of a set of nodes, N , a set of links (trunk groups) L, and a set of all possible ....

....to provide a systematic way for obtaining the values of ff k s, we have found, from our experiments, that the choices of ff k s are sensitive only to the number of traffic classes present in the network. A complete discussion of how values of the ff k s are chosen is deferred to section 5.1. In [18], we also have shown that when b=C is sufficiently small, as in the NSFNET example presented in section 5, the performance of MDP based routing algorithms is insensitive to the presence or absence of the compensation parameters while computing the link shadow prices. 4 Routing Policies Four ....

R.-H. Hwang, "Routing in High-Speed Networks," Ph.D. Dissertation, Dept. of Computer Science, University of Massachusetts at Amherst, 1993.

....of 12 nodes, 30 T3 links and 132 source destination pairs. Each node is assumed to have the same processing capacity. We assume the exogenous call arrival rate for each source destination pair is proportional to the traffic load at each node reported in NSFNET [22] The reader is referred to [23] for the detailed traffic model setup. In the following sections, we first validate our analytical models by simulation results. We then compare the performance of the five routing schemes discussed in previous sections and study the effects of call processing delay and the admission control ....

R.-H. Hwang, "Routing in High-Speed Networks," Technical Report 93-43, Dept. of Computer Science, University of Massachusetts at Amherst, 1993.

No context found.

Hwang, R., "Routing In High-Speed Networks," Ph.D. Dissertation, University of Massachusetts, May 1993.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC