S. McCanne and S. Floyd, "The LBNL net- work simulator." Software on-line: http://www.isi.edu/nsnam, 1997. Lawrence Berkeley Laboratory.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....that it is simply impossible to create a model of the Internet. This is primarily due to four reasons. 1. We do not have tools capable of modeling networks of that size. All existing tools have bounded capacity, from a few hundred nodes for OpNet[1] to a few thousand nodes for GloMoSim[2] and ns[3], to a few tens of thousands of nodes for T ) 4] up to a few hundred thousand nodes for pdns[5] SSF[6] and USSF[7] 2. Even if we had the capability of constructing large enough models, we would not know how to construct models that accurately reflect the topology and performance of the ....

....290,000 seconds for each second of simulation time (almost four days) 2. The total memory required for the model is 2. 9 x 10 TM bytes (nearly three hundred ter abytes) This calculation uses measured memory requirements for nodes, links, connections and packets using the ns network simula tor [3]. We did not include memory needed for simulated routing tables at each node. See [10] for a discussion of this memory utilization and a potential solution. 3. The total disk space required for logging the simulation results is 1.4 x 1013 bytes for each second of simulation time. This calculation ....

S. McCanne and S. Floyd, "The LBNL net- work simulator." Software on-line: http://www.isi.edu/nsnam, 1997. Lawrence Berkeley Laboratory.

.... purpose parallel simulation kernel that provides the infrastructure for parallel simulation [10] anse provides the necessary tools and models for modeling and simulation of active networks [15, 11] NS is a sequential network simulator that enables modeling and simulation of conventional networks [5, 14]. Brief descriptions about these components are presented in the following subsections. 3.1 WARPED warped is a general purpose parallel discrete event simulator [10] It uses the Time Warp mechanism for parallel synchronization. Time Warp is a optimistic synchronization mechanism based on the ....

....during interpretation. anse also provides a web based interface to ease simulation of active networks over the World Wide Web (WWW) Further details on anse and PLAN are available in the literature [3, 12] 3. 3 NS NS is a network simulator developed at the University of California at Berkeley [5]. It provides substantial support for simulation of TCP, routing, and multicast protocols. NS is written in C with an OO Tool Control Language (OTcl) interpreter as its front end. A NS simulation is organized as a set of interacting agents that are interconnected using links . NS provides ....

[Article contains additional citation context not shown here]

S. McCanne and S. Floyd. The LBNL network simulator, 1997. (on-line: http://www.mash.cs.berkely.edu/ns).

....path. Of course, a more efficient traceroute program could be developed that records the route at each hop, as has been done with the multicast traceroute program mtrace. B. Simulation Model We evaluated the expanding ring local search heuristic within version 1 of the LBNL network simulator ns [35]. Because its performance depends heavily on the topology, we used two different network models: a flat random network and a transit stub network. Flat networks are created using the Doar Leslie edge connection method [36] so that edges mostly connect nodes near each other. Transit stub networks ....

S. McCanne, S. Floyd, and K. Fall, "LBNL Network Simulator," http://ee.lbl.gov/ns.

....sessions with and without TCP traffic in the background. 2.2.1 Multiple Video Multicast Sessions We study what happens between two competing layered video multicast sessions on the bottleneck link , as shown in the topology in Figure 2. All simulations were done using LVMR [LPA98] built on ns [MF]) but the results are applicable to RLM and other layered video rate control protocols. In this topology, there are two receivers, C 1 and C 2 , and two senders S 1 and S 2 . Two video sessions are set up: session 1 from S 1 to C 1 and session 2 from S 2 to C 2 , each running with different ....

..... S1 S2 Sk RA1 RA2 RAk RB RC RD1 RD2 RDk . C11 Ck1 C1n1 . C21. C2n2 . Cknk Figure 11: Simulation Topology 2 5 Simulation and Analysis We simulated the congestion control schemes proposed in the previous section with modules of LVMR implemented in ns [MF]. We used 36000 frames of an MPEG II trace to produce 4 layers with the following layering scheme: given the GOP pattern of IB 1 B 2 PB 3 B 4 PB 5 B 6 PB 7 B 8 , I frames make the base layer, P frames the first enhancement layer, B 1 ; B 3 ; B 5 and B 7 the second enhancement layer and the rest of ....

Steven McCanne and Sally Floyd. The UCB/LBNL network simulator. software online. http://www-mash.cs.berkeley.edu/ns/.

....6.3.1. Hierarchical distribution of session messages in SRM can have an impact on the estimation of propagation delay between the members. We will study this protocol impact and investigate mechanisms that can be adopted to reduce the impact. We will implement SRM in the LBNL Network Simulator ns [31] for our simulation studies. The proposed distribution mechanisms will be implemented for SRM. Simulation studies will compare the proposed approach with global distribution and other mechanisms proposed in [27] for scaling session messages. This simulation environment will also be used to ....

S. McCanne and S. Floyd. The LBNL Network Simulator, 1996. Software On-line.

....4 or r 5 in the second iteration, i.e. the rates falling into the range [r lt 1 ; r lt 2 ] 5 Simulations In this section, we examine the performance of our scheme under a variety of network topologies and group characteristics. We implemented our protocol in the LBNL network simulator ns 2 [11]. In the simulation, the inter group protocol is called every 20 seconds. The sender invites the B group receivers to join the V group whenever the number of B group receivers exceeds 10 of the total and at least one receiver in the B group is not experiencing loss. Moreover, a receiver that ....

S. McCanne and S. Floyd, "The LBNL network simulator, " Lawrence Berkeley Laboratory, Software on-line: http://www-nrg.ee.lbl.gov/ns.

....but some general discussions are available in section V. We believe with all the mechanisms proposed in section V, the control overhead of PIM DT can be comparable to PIM SM. B. Simulation Description We also simulated the PIM DT with source specific trees on the the LBNL network simulator ns [7] to validate the basic protocol behavior of PIM DT and its effectiveness in state reduction. The simulation implementation is based on ns Multicast Extension developed by Daniel Zapala at University of Southern California. The performance of PIM DT is compared to PIM SM. In the simulation all the ....

S. McCanne and S. Floyd. The LBNL Network Simulator. software online(http://www-nrg.ee.lbl.gov/ns).

....RTG[10] As a worst case scenario, all routers were considered Candidate BSRs. 9 Simulations were performed using PIMSIM[11] a discrete event driven packet level simulator based on the Maryland Routing Simulator(MaRS) 12, 13] Future simulations will use the Network Simulator (ns) version 2 [14]. 10 This function could be extended easily to accommodate various priorities as well. ffl Group balancing under any multicast address allocation scheme. By this, we mean that when the number of groups is large, no single RP is serving significantly many more groups than any other RP in the ....

Steven McCanne. The UCB/LBNL network simulator(ns), November 1996.

....Far from other Representatives and Close to Representative Table 3: Evaluating Appropriateness 5 Simulations of the Scalable Session Messages In this section we examine the performance of scalable session messages using simulation. We modified the SRM implementation in the network simulator (ns) [29] to implement the proposed hierarchy configuration and session message distribution mechanisms. Our simulations study three aspects of scalable session messages in SRM. Section 5.1 examines how the loss recovery performance of the original SRM (with flat session messages) is affected by use of ....

Steve McCanne and Sally Floyd. "LBNL Network Simulator". http://ee.lbl.gov/ns.

....representative based on some function of the local scopes reported by various representatives. 5 Simulations of the Scalable Session Messages In this section we examine the performance of scalable session messages using simulation. We modified the SRM implementation in the network simulator (ns) [28] to implement the proposed hierarchy configuration and session message distribution mechanisms. Our simulations study three aspects of scalable session messages in SRM. Section 5.1 examines how the loss recovery performance of the original SRM (with flat session messages) is affected by use of ....

Steve McCanne and Sally Floyd. "LBNL Network Simulator". http://ee.lbl.gov/ns.

....use of the RTCP bandwidth especially for receivers connected through low bandwidth links. In addition, our scheme directs important information included in RRs to an entity that can make valuable use of them. In the next phase of our work, we are simulating S RTCP using the network simulator (NS) (McCanne, 1998) and we will report the results in due course. Also, we intend to investigate more functions that the Manager can do, analyse the limitations of our design, and try to refine it. 7 ACKNOWLEDGEMENT We wish to acknowledge the support of BT Labs for sponsorship of Randa ElMarakby s Ph.D. ....

McCanne, S. and Floyd, S. (McCanne, 1998) "LBNL Network Simulator", http://mash.cs.berkeley.edu/ns, 1998 version.

....statistics, then sends these statistics to a Manager. The Manager will perform other statistics to evaluate the network performance and to estimate regions which are suffering highly from congestion. In the next phase of our work, we will simulate our scheme using the network simulator (NS) [3]. Acknowledgements We wish to acknowledge the support of BT Labs for sponsorship of Randa El Marakby s PhD programme. She was sponsored previously by the British Council in Egypt. ....

S. McCanne and Sally Floyd, "LBNL Network Simulator", http://mash.cs.berkeley.edu/ns

....and vice versa. We discuss self evaluation of appropriateness later in Section 5. 4 Simulations of the Scalable Session Messages In this section we describe the simulation studies we conducted to study the scalable session messages. We modified the SRM implementation in network simulator (ns) [23] to use hierarchy configuration and proposed session message distribution. We conducted three sets of studies using the simulator. The first set of experiments compares the loss recovery performance of original SRM with use of scalable session messages in SRM. We then studied the improvement in ....

Steve McCanne and Sally Floyd. "LBNL Network Simulator". http://ee.lbl.gov/ns.

....the overhead of all of the heuristics to a small number of third party queries. a) 100 node flat random network b) 100 node transit stub network Figure 9: Generated Networks 4. 2 Simulation Model We have implemented the route construction algorithm given above within the LBNL network simulator [MFF] to evaluate effectiveness in finding alternate paths. Our primary goal is to characterize the performance of the heuristics according to 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 ....

Steve McCanne, Sally Floyd, and Kevin Fall. "LBNL Network Simulator". http://ee.lbl.gov/ns.

....losses are detected at different times. This typically allows the receivers closer to the source to suppress the NACKs from receivers further away. One of the key points in our investigation is how the setting of the timer constants affects this behavior. We used the VINT network simulator ns [83] for our work. In its original form, ns turned out to have prolific memory usage with heavy weight nodes, links, and multicast routing infrastructure, and could not support more than a few hundred nodes on an ordinary workstation. However, we took advantage of ns s extensible object oriented ....

....stand alone implementation of the soft state transport protocol libsstp is available from: http: www mash.cs. berkeley.edu mash software srm2.0 Our extensions to ns 2 to perform large scale simulations to study the asymptotic scaling behavior of timer based recovery is available from the ns 2 [83] distribution, available from: http: www mash.cs.berkeley.edu ns All the applications written have been written within the context of the MASH programming environment and are available from the MASH web page: http: www mash.cs.berkeley.edu mash 8.3 Summary In this dissertation, we ....

Steven McCanne and Sally Floyd. The LBNL Network Simulator. University of California, Berkeley. http://www-mash.cs.berkeley.edu/ns/.

....converges, divided by N . For example, a launch overhead factor of 1.5 implies that 2.5N HMs were created before the algorithm stabilized at N HMs. Figure 6 plots the relationship between the ratio of the adaptation to announcement intervals and the launch overhead factor obtained in an ns [24] simulation of the HM deployment algorithm on a cluster of 40 machines with a target population size of 10. The error bars designate the standard deviation of our simulations. As the ratio of adaptation to announcement intervals decreases, the number of redundant launches increases. On the other ....

MCCANNE, S., AND FLOYD, S. The UCB/LBNL Network Simulator. University of California, Berkeley. Software online 2 .

....Assuming the session bandwidth, B, is 128 kb s, proportional bandwidth allocation results in S 0 transmitting at 128 Theta 0.95 Theta 0.7 = 85.12 kb s and S 1 at 36.48 kb s. To better illustrate the dynamics of this distributed algorithm, we implemented it in the ns network simulator [15] and ran a simple experiment consisting of four sources and three receivers across a single bottleneck link. All sources were placed on one side of the link while the receivers ACM Multimedia November 1997 Seattle, WA 5 Time (sec) 0 100 200 300 400 1.5 2 2.5 3 3.5 4 4.5 5 5.5 1.5 2 2.5 ....

MCCANNE, S., AND FLOYD, S. The UCB/LBNL Network Simulator. University of California, Berkeley. Software online 6 .

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