K. Calvert and E. Zegura. Georgia Tech Internetwork topology models. 1996.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....We report ratios to the corresponding SPT metrics, so that we can compare results across di erent graphs and groups. 3. 1 Experiment Results: Delay For these experiments, we use a set of 10 at, random graphs of 50 nodes each, using the Waxman model [8] within the GT ITM topology generator [9]. All edges have unit weights and the average node degree is near 4. For each graph, we generate 50 random groups and measure the delay experienced by group members. We de ne delay as the number of links traversed between one sender and one member in the group. We calculate the maximum delay for ....

Ken Calvert and Ellen Zegura, \Georgia Tech Internetwork Topology Models," http://www.cc.gatech.edu/fac/Ellen.Zegura/graphs.html.

....of the transit network, increasing number of transit networks, or increasing the size of the stub networks) To conserve space, we discuss only representative simulations from one of these networks as indicated in the table. All of these networks were generated with the Georgia Tech ITM software [37], 38] Note that these topologies model only the router portion of the network. For simplicity, we do not model hosts, as they will generally connect to a single router and thus add only a single link to the number of links that need to be visited. Thus our evaluation of heuristics is ....

Ken Calvert and Ellen Zegura, "Georgia Tech Internetwork Topology Models," http://www.cc.gatech.edu/fac/Ellen.Zegura/graphs.html.

....Our test sets consist of 50 graphs for each type. We generated 3 sets of graphs for each type to get some indication as to the stability of our results relative to the random number generator used to create the network graphs. We use the suite of random graph generators available at Georgia Tech [5]. These algorithms all place a set of n vertices on a square in the plane, and then consider each pair of vertices in turn, deciding whether an edge is to be added between them. The purely random scheme uses an equal probability between all pairs of vertices; the locality preference scheme has the ....

K. Calvert and E. Zegura. Georgia Tech Internetwork Topology Models. http://www.cc.gatech.edu/fac/Ellen.Zegura/graphs.html.

....with clear benefits. Given this, we assume client site caches large enough to hold all stocks and study the additional value of caching within the network. We do this with a series of simulations. There are several key parameters in our simulations: ffl T, the network topology. We used GT ITM [5, 6] to generate a variety of random internetwork topologies. Each topology is a transit stub network [30] supporting 1000 end nodes. The end nodes are organized into sites (of average size six) that are connected to the backbone. With an average of four sites per backbone node, the backbone consists ....

K. Calvert and E. Zegura. Georgia Tech Internetwork Topology Models (GT-ITM). Georgia Tech College of Computing. Software on-line: http://www.cc.gatech.edu/fac/Ellen.Zegura/graphs.html.

....work of others on automatic topology generation. For the two topology generation tools described below, we have written and made available conversion programs that allow the generated topologies to be used in ns simulations. The Georgia Tech Internetwork Topology Models (GT ITM) software package [12, 13, 79] can create flat random networks using a variety of edge distribution models, including pure random, exponential, locality, several variations of Waxman s model [75] and the Doar Leslie model [22] Given the size of a grid and the number of nodes desired, GT ITM randomly places nodes on the grid ....

Ken Calvert and Ellen Zegura. Georgia tech internetwork topology models. http://www.cc.gatech.edu/fac/ Ellen.Zegura/graphs.html.

....experiments on topologies of up to 100 nodes. These are reasonably large networks given the resources available. We discuss the topologies used as well as the other simulation parameters below. 4.1. 1 Topologies We used Georgia Institute of Technology s Internetwork Topology Models (GT ITM) [24, 4] tool to generate random transit stub topologies for our simulations. GT ITM generates hierarchical, or transit stub topologies as an approximation of an internetwork. The transit stub topologies consist of a top level transit network (nodes 0 to 3) and second level stub topologies (nodes 4 to 99) ....

K. Calvert and E. Zegura. Georgia tech internetwork topology models. http://www.cc.gatech.edu/fac/Ellen.Zegura/graphs.html.

....a varied set of topologies. We are also interested in measuring the path length of alternate paths computed using the heuristics; they should not be many hops longer than the shortest path. 4.2. 1 Topology We generated various large topologies using the Georgia Tech ITM topology generator [ZCB96, CZ] We used one flat random network of 100 nodes (Figure 9a) using the Doar Leslie edge connection method [DL93] to generate edges that mostly connect nodes near each other. The average degree of connectivity for this network is 4.26. We also created transit stub topologies, which consist of a ....

Ken Calvert and Ellen Zegura. "Georgia Tech Internetwork Topology Models". http://www.cc.gatech.edu/fac/Ellen.Zegura/graphs.html.

No context found.

K. Calvert and E. Zegura. Georgia Tech Internetwork topology models. 1996.

No context found.

K. Calvert and E.W. Zegura. Georgia tech internetwork topology models. Available from http://www.cc.gatech.edu/projects/gtitm/, 1996.

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