P. Huang, D. Estrin, and J. Heidemann. Enabling large-scale simulation: Selective abstraction approach to the study of multicast protocol. In Proceedings of the 6th International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, page 241. IEEE Computer Society, 1998.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....the context of networks, traditional packet level discrete event simulations do not scale to larger networks, and several simulation techniques based on approximations have been studied. The approaches can be broadly classified into two categories: i) techniques that abstract the level of traffic [14, 17, 20] (i.e: treating a set of packets as a single unit or abstracting a flow of packets using a fluid model) and (ii) techniques that reduce the number of packet arrival events either by simulating the system at fixed time intervals or by using arrival distributions [15, 29] While these techniques are ....

P. Huang, D. Estrin, and J. S. Heidemann. Enabling Large-Scale Simulations: Selective Abstraction Approach to the Study of Multicast Protocols. In MASCOTS, pages 241--248, 1998.

....and de aggregation maybe performed statically or dynamically. Static transformations occur prior to commencement of simulation while dynamic transformations occur during the course of simulation. Earlier research activities related to static CA technique have been reported in the literature [2, 4, 14], to address issues related to performance of large scale simulations. On the other hand, dynamic CA and CDA not only encompass the utility of their static counterparts but also provide a number of additional advantages [11] Further details about CA and CDA are available in the literature [11] ....

HUANG, P., ESTRIN, D., AND HEIDEMANN, J. Enabling large-scale simulations: Selective abstraction approach to the study of multicast protocols. In In Proceedings of International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Networks (Oct. 1998).

....[6] # and# [16]# outline# fundamental# goals# and# tradeoffs# to consider# when# simulating# a# network# layer# abstraction. Additionally# these# papers# provide# important# insights# into understanding#the#effects#of#model#abstraction#and#testing for#possible#simulator#invalidation. [8]# looks# at# two# abstraction# techniques:# centralized computation#and#abstract#packet#distribution.##Centralized computation# saves# memory# consumption# and# time# by centrally#computing#protocol#states#to#reduce#the#workload and# complexity# of# performing# these# computations# for ....

# Huang,# Polly.# Deborah# Estrin,# John# Heidemann.# et# al., "Enabling# Large-scale# Simulations:# Selective# Abstraction Approach# to# the# Study# of# Multicast# Protocols." USC/Information# Science# Institute,# University# of# Southern California.#Marina#del#Rey,#CA.##July#1998.

....algorithm as the default routing. The RouteLogic Wavelength supplies the WAssignLogic with route information to route incoming traffic. New wavelength routing algorithms can be easily implemented as derived classes based on the RouteLogic Wavelength object. 3.1. 3 Abstraction in OWns Abstraction [17] is a useful technique that helps in implementing large scale scenario simulation and complex protocols. By abstracting necessary details, memory and time consumption can be greatly reduced without sacrificing accuracy. In WDM circuit switched scenarios, each lightpath needs to be mapped by a ....

P. Huang, D. Estrin, and J. Heidemann, "Enabling Large-scale Simulations: Selective Abstraction Approach to The Study of Multicast Protocols," in Proceedings of the International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, (Montreal, Canada), IEEE, July 1998.

....such as [16, 15, 12, 11, 8, 4, 20, 1] 2. We can reduce the memory required to represent the state of the model. We observe that in typical simulations, a) the superfluous states (that are not needed by the simulations) often occupy excessive memory, which can be safely eliminated (e.g. [14, 6]) b) the necessary states (that are needed by the simulations) often contain significant amount of redundant information, which can be easily compressed aggregated (e.g. 14] In this paper, we focus primarily on techniques for reducing the state memory requirements to enable large scale ....

P. Huang, D. Estrin, and J. Heideman. Enabling large-scale simulations: selective abstraction approach to the study of multicast protocols. In Proceedings of the International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, July 1998.

....networks using a network of Dual CPU Pentium processors connected by an Ethernet network. USSF has been demonstrated on network models of hundreds of thousands of nodes. A selective abstraction approach to modeling networks using the ns simulator is discussed by Huang, Estrin and Heidemann in [22]. In this method, decreased detail in the modeling of end to end data flows is traded for increased computational efficiency. Using this approach, the ns simulator has been demonstrated on multicast groups consisting of thou sands of end hosts. The preceding summary indicates that some progress ....

P. Huang, D. Estrin, and J. Heideman, "En- abling large-scale simulations: selective abstraction approach to the study of multicast protocols," in Proceedings of the International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, July 1998.

....bais against long RTT connections. With both RED and Drop Tail queue management, a connection s goodput is inversely proportional to its RTT. 8. Related Work Large scale performance analysis has been an active research area recently. Many researches focus on building scalable simulators, such as [3, 9, 21, 23, 25, 26, 6, 46, 47]. For example, 9] speeds up simulation by using parallel simulation techniques. It does not change the underlying simulation paradigm. Other approaches focus on raising the level of simulation abstraction. In particular, 3] proposes to abstract an entire group of closely spaced packets as a ....

....abstract an entire group of closely spaced packets as a packet train. This enables their simulator to represent network traffic with considerably fewer events than a traditional simulator. Two other abstraction techniques: centralized computation and abstract packet distribution, are suggested in [23], which are shown to apply to multicast simulation. The combination of these two methods leads to orders of magnitude of performance improvement without sacrificing accuracy. Another approach is to use fluid models to approximate real traffic at coarser time scales. For instance, 25, 26] track ....

P. Huang, D. Estrin, and J. Heidemann. Enabling Large-scale Simulations: Selective Abstraction Approach to the Study of Multicast Protocols. Proc. Sixth International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems (MASCOTS'98), Montreal, Canada, July 1998.

....is relatively stable and is clustered within a band close to (the width of the band is usually one unit in log2 scale) 6. Related Work Large scale performance analysis has been an active research area recently. Many researches are currently focused on building scalable simulators, such as [1, 5, 15]. Analyzing simulation results to estimate TCP performance, as done in this project, is a very different approach from building a scalable simulator. The strategy taken by [9] is the closest to ours. It studies how TCP throughput, loss rates, and fairness are affected by changing the number of ....

P. Huang, D. Estrin, and J. Heidemann. Enabling Large-scale Simulations: Selective Abstraction Approach to the Study of Multicast Protocols. USC-CS Technical Report 98-667, January 1998.

....influenced by the competition for wavelengths by different nodes. Therefore, it is imperative to simulate an entire multinode network rather than a single node. 5. 1 VINT ns Simulation Model The VINT ns tool is a popular simulation package used for evaluating protocols in large scale networks [Hua98]. VINT ns performs packet level simulation according to specified set of protocols. It has extensive facilities for the purposes of gathering data and testing functionality, and it is a valu able tool of many protocol designers. Most importantly for our work, it accepts as inputs log files of ....

P. Huang, D. Estrin and J. Heidemann, "Enabling Large-scale Simulations: Selective Abstraction Approach to the Study of Multicast Protocols", Proc. IEEE MASCOTS '98, pp. 241--248, Montreal, Jul. 1998.

....away some of the detail of packet level simulators by grouping packets with closely spaced send times together into a single event called a packet train. This is motivated by the observation that links on real networks are frequently occupied solely by packets from the same ow. In addition, [19] develops two abstractions for simulating large scale mutlicast networks. The rst abstraction is to eliminate maintenance messages that maintain the multicast tree. The second abstraction, and the one most related to our work, is that they do away with hopby hop routing and instead schedule ....

P. Huang, D. Estrin, and J. Heidemann. "Enabling Large-scale simulations: selective abstraction approach to the study of multicast protocols," In Proceedings of the International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, pp. 241-248. Montreal, Canada, IEEE. July, 1998.

....a better insight into the way that protocols behave in di erent environments. Ideally, we would like as few parameters and metrics in the model as possible. We have not been able to nd any such simpli cation e orts for wireless networks, although there has been work in other environments [8, 2, 6]. The problem we address in our work is best illustrated by the following example. Let us assume that we want to compare the performance of two networks A and B. Network A has an arrival rate of 30 users per second, 80 utilization, and 5 loss (i.e. blocked and dropped users) Network B has an ....

....that attempts to simplify wireless network simulations. Some research has been done in the general simulation environment. Most research that aims for a simple or standardized model attempts to hide as many low level details as possible. The idea of simulation abstraction has been considered in [8]. They try to abstract unnecessary details from a simulation. Fall [6] suggests decreasing the number of objects in a simulation by aggregating some of them. He also discusses the idea of coarsening the resolution of some parts of the model. A standard framework for simulation has been suggested ....

P. Huang, D. Estrin, and J. Heidemann. Enabling large-scale simulations: Selective abstraction approach to the study of multicast protocols. Proceedings of the Sixth International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunications Systems (MASCOTS '98), Montreal, Canada, July 1998.

....networks using a network of Dual CPU Pentium processors connected by an Ethernet network. USSF has been demonstrated on network models of hundreds of thousands of nodes. A selective abstraction approach to modeling networks using the ns simulator is discussed by Huang, Estrin and Heidemann in [22]. In this method, decreased detail in the modeling of end to end data ows is traded for increased computational eciency. Using this approach, the ns simulator has been demonstrated on multicast groups consisting of thousands of end hosts. The preceding summary indicates that some progress is ....

P. Huang, D. Estrin, and J. Heideman, \Enabling large-scale simulations: selective abstraction approach to the study of multicast protocols," in Proceedings of the International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, July 1998.

....concludes. Simulation granularity Key abstraction technique Target domain(s) ns [19] packet transport level protocols and below Packet trains [1] groups of packets treating closely spaced series of packets as a large packet transport level protocols on packet networks Selective abstraction [12] packet selectively move transient details from simulated world to simulator transport level protocols and below Fluid network simulation [15] flow changes map network as pipes, packet streams as fluid flows transport level protocols Time stepped Hybrid Simulation (TSHS) 10] packet time ....

....than real time, which may prove unwieldy for experiments that require a day s worth or more of application execution time on a machine cluster to produce a valid data point. To address these issues, many researchers have turned to analytical modeling and or simulation for answers. Several efforts [1, 5, 8, 11, 12, 15, 17] discuss the performance tradeoffs one encounters when mapping real life environments to a simulated environment and back again. They all sound a common theme choosing an appropriate level of abstraction is key, both to ensure continued accuracy of results and to enable the simulator to scale ....

Polly Huang, Deborah Estrin, and John Heidemann. Enabling large-scale simulations: selective abstraction approach to the study of multicast protocols. In Proceedings of the International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, pages 241--248. IEEE, July 1998.

....OPNET [4] and ns 2 [10] aid designers in determining whether a particular protocol change would improve the overall behavior of the network. Research in the area of network simulation has been focused on developing fast software simulation techniques for Internet scale networks (for instance, see [2, 3, 5]) as well as the design of flexible and compositional component based simulation architectures (for instance, see [3, 14] Wireless ad hoc networks present a somewhat different point in the design space for network simulators. A wireless channel is more complex than a traditional wired ....

....required among processors introduces overhead in the parallel simulation that limits speedup [2] Abstraction and Analytic Techniques. Abstraction is a technique whereby certain details of the underlying network are ignored in the interests of improving the performance of the simulation [5]. Since abstracting away details might change the results of the simulation, care must be taken to identify the parts of the simulation that can be safely ignored. This approach can also be combined with analytic techniques [12] where multiple steps of the simulation are mathematically analyzed ....

Polly Huang, Deborah Estrin, and John Heidemann. Enabling Large-scale simulations: selective abstraction approach to the study of multicast protocols. Proceedings of the International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, pages 241--248, July 1998.

No context found.

P. Huang, D. Estrin, and J. Heidemann, "Enabling large-scale simulations: selective abstraction approach to the study of multicast protocols," in 6th Int. Symp. Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS '98), July 1998, pp. 241--248.

....matrix, D. Mapping of delay matrix into complex topologies is out of scope of this document. This topology generator is probably representative of a standard tool for topology generation used in networking research. Using GT ITM we have covered most topologies used in several SRM studies [18] [19]. We faced difficulties when choosing the lossy link for the random topologies in order to maximize the number of responses. This is an example of the difficulties networking researchers face when trying to stress networking protocols in an ad hoc way. SRM response timer values are selected ....

P. Huang, D. Estrin, and J. Heidemann. Enabling Large-Scale Simulations: Selective Abstraction Approach to The Study of Multicast Protocols. Sixth International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS '98), July 1998.

.... [5] Parsec [2] and SSF [8] provide detailed protocol implementations, but what level of detail is required in new protocols, or in adapting existing protocols to model new hardware Some simulators ease the cost of changing abstraction with multiple, selectable levels of detail (for example, ns [17]) but the design choice must still be made. Choices about detail are particularly difficult for wireless network simulations. Wide experience with the important components of wired networks over the last 30 years allows significant abstraction. For example, pointto point links are often ....

....a single detailed quantitative result. A more abstract simulation can also make the effects of a change in algorithm distinct, where they would be obscured by other effects in a more detailed simulation. Finally, omission of simulation detail can improve performance by multiple orders of magnitude [17]. Memory and run time improvements due can offer results sooner, or allow larger or longer experiments, revealing different aspects of protocol behavior. For example, the relative performance of ad hoc routing protocols differs with large numbers of nodes [10] The primary risk of simulation ....

[Article contains additional citation context not shown here]

Polly Huang, Deborah Estrin, and John Heidemann. Enabling large-scale simulations: selective abstraction approach to the study of multicast protocols. In Proceedings of the International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, pages 241--248, Montreal, Canada, July 1998. IEEE.

....Whereas random topologies simulation generated almost 10 responses in the worst case. 11 The topology generator is probably representative of a standard tool for topology generation used in networking research. Using GT ITM we have covered most topologies used in several SRM studies [18] [19]. 12 We faced di#culties when choosing the lossy link for the random topologies in order to maximize the number of responses. This is an example of the di#culties networking researchers face when trying to stress networking protocols in an ad hoc way. 13 SRM response timer values are selected ....

P. Huang, D. Estrin, and J. Heidemann. Enabling Large-Scale Simulations: Selective Abstraction Approach to The Study of Multicast Protocols. Sixth International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS '98), July 1998.

.... [3] Parsec [2] and S3 [7] provide detailed protocol implementations, but what level of detail is required in new protocols, or in adapting existing protocols to model new hardware Some simulators ease the cost of changing abstraction with multiple, selectable levels of detail (for example, ns [16]) but the design choice must still be made. Choices about detail are particularly difficult for wireless network simulations. Wide experience with the important components of wired networks over the last 30 years allows significant abstraction. For example, pointto point links are often ....

....a single detailed quantitative result. A more abstract simulation can also make the effects of a change in algorithm distinct, where they would be obscured by other effects in a more detailed simulation. Finally, omission of simulation detail can improve performance by multiple orders of magnitude [16]. Memory and run time improvements due can offer results sooner, or allow larger or longer experiments, revealing different aspects of protocol behavior. For example, the relative performance of ad hoc routing protocols differs at higher scales [9] The primary risk of simulation abstraction is ....

[Article contains additional citation context not shown here]

Polly Huang, Deborah Estrin, and John Heidemann. Enabling large-scale simulations: selective abstraction approach to the study of multicast protocols. In Proceedings of the International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, pages 241--248, Montreal, Canada, July 1998. IEEE.

No context found.

P. Huang, D. Estrin, and J. Heidemann. Enabling large-scale simulation: Selective abstraction approach to the study of multicast protocol. In Proceedings of the 6th International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems, page 241. IEEE Computer Society, 1998.

No context found.

HUANG, P., ESTRIN, D., AND HEIDEMANN, J. Enabling Large-scale Simulations: Selective Abstraction Approach to the Study of Multicast Protocols. In In Proceedings of the Sixth International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS '98) (Montreal, July 1998).

No context found.

HUANG, P., ESTRIN, D., AND HEIDEMANN, J. Enabling Large-scale Simulations: Selective Abstraction Approach to the Study of Multicast Protocols. In In Proceedings of the Sixth International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Systems (MASCOTS '98) (Montreal, July 1998).

No context found.

Huang, P.; Estrin, D.; Heidemann, J. - "Enabling large-scale simulations: selective abstraction approach to the study of multicast protocols" - Proceedings of IEEE Sixth International Symposium on Analysis and Simulation of Computer and Telecommunication Systems, 19-24 July 1998, Montreal, Que. Canada, pages: 241 - 248

No context found.

HUANG, P., ESTRIN, D., AND HEIDEMANN, J. Enabling large-scale simulations: Selective abstraction approach to the study of multicast protocols. In In Proceedings of International Symposium on Modeling, Analysis and Simulation of Computer and Telecommunication Networks (Oct. 1998).

No context found.

P. Huang, D. Estrin, and J. S. Heidemann. Enabling Large-Scale Simulations: Selective Abstraction Approach to the Study of Multicast Protocols. In MASCOTS, pages 241--248, 1998.

First 50 documents

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