Subnet Level Internet Topology Generator - PowerPoint PPT Presentation


PPT – Subnet Level Internet Topology Generator PowerPoint presentation | free to download - id: 569996-Y2UwM


The Adobe Flash plugin is needed to view this content

Get the plugin now

View by Category
About This Presentation

Subnet Level Internet Topology Generator


A Better Model for Generating Test Networks 1996 A multi-tier network topology ... Modeling Internet Topology Dynamics , ACM SIGCOMM Computer ... – PowerPoint PPT presentation

Number of Views:63
Avg rating:3.0/5.0
Slides: 25
Provided by: mba171
Learn more at:


Write a Comment
User Comments (0)
Transcript and Presenter's Notes

Title: Subnet Level Internet Topology Generator

Subnet Level Internet Topology Generator
  • Mehmet Burak AKGUN
  • CS790
  • Complex Networks

  • Introduction
  • Literature Review
  • Subnet Level Generator Design Targets
  • References

  • Internet is not always available for experimental
    purposes. Thus researchers use network
  • In order to evaluate new algorithms and
    protocols, network researchers need realistic
    network topologies to be used in the simulation
  • General purpose of internet topology generators
    is to synthesize realistic and highly
    configurable internet topologies in a reasonable

Why not use Existing topologies?
By Stephen G. Eick - http//
Why not use Existing topologies?
  • Internet is growing in a distributed and
    uncontrolled fashion
  • Achieving a deep understanding of internet
    topology is a challenging task.
  • Operators do not want to publish the details of
    the existing topology
  • AS-level connectivity is based on the complex
    business relationships and routing policies among
    service provider companies

Topology Discovery Studies
  • Many studies were carried out to map the actual
    internet topology through Trace-Route.
  • However no one is completely successful.
  • - Aliasing issues
  • - Load Balancing Routers
  • - Unresponsive Routers
  • -It takes a long time to tracert huge number the

Literature Review
  • Before 1999
  • There is a strong belief that internet is
  • 1999-2001
  • Discovery of internets degree distribution to be
    power law
  • 2001-
  • Attention shifted again from local properties to
    large scale properties which are better
    represented by hierarchical generators.

Routing of Multipoint Connections B.M. Waxman
  • Produces random graphs using Erdos-Renyi random
    graph model
  • Nodes are uniformly distributed on a plane
  • Existence of an edge between two nodes is a
    probabilistic function of the distance between
    nodes. (inversely affected by distance)

Tiers M. Doar. A Better Model for Generating Test
Networks 1996
  • A multi-tier network topology generator
  • Three level hierarchical structure
  • Only one WAN per random topology
  • For each level of hierarchy, user specifies
    number of nodes
  • Minimum spanning tree is calculated

GT-ITM How to Model an Internetwork, E.W.Zagura 1996
  • Two types of hierarchical graphs(n-level, TS)
  • Transit-stub reproduces the hierarchical
    structure of internet.
  • Firstly a connected random graph is generated
  • Each node is considered as a transit domain and
    each transit domain is expanded to form another
    connected random graph
  • After running expanding operation for levels, A
    number of random graphs are generated and
    connected to each node in the network as Stubs

On Power-Law Relationships of the Internet
Topology. C. Faloutsos, P. Faloutsos, and M.
Faloutsos. 1999.
  • Measurements on internet
  • AS level (Autonomous Systems as nodes)
  • Router level (Routers as nodes)
  • Found out that Degree Distribution of these
    graphs are power laws
  • Led to a new generation of topology generators
    which does not model the hierarchical structure
    of internet and focus on the node degrees

Degree Based Topology Generators
  • These generators assume the fact that, it is more
    important to match the local properties of
    internet (like degree distribution) rather than
    large scale hierarchical structure

Heuristics for Internet Map Discovery, R.
Govindan, H. Tangmunarunkit, Infocom 2000
  • Measurements conducted by Bell-labs using 284806
    nodes and 449306 edges.
  • System named MERCATOR
  • All measurements are router level, i.e, no end
    hosts (which may be up to billions)
  • Connection degree is lower than 0,001

A map of the Internet as discovered by the
Bell-labs. Links painted with different colors
represent different geographical locations.
Internet Node Degree Distribution by Bell-Labs
97 correlation with ideal power law distribution
Inet Internet Topology Generator Cheng Jin, Qian
Chen, Sugih Jamin 2000
  • AS level generator
  • User specifies target size
  • Inet assigns degree for each node according to
    the power degree distribution
  • Forms the spanning tree using the number of edges
    decided for each node

BRITE An Approach to Universal Topology
Generation Alberto Medina, Anukool Lakhina,
Ibrahim Matta, and John Byers,2001
  • Argue that preferential connectivity and
    incremental growth are the primary reasons of the
    power law distribution of internet
  • Skewed node placement
  • Area is divided into HSxHS squares and nodes
    distributed (one node is selected for backbone)
  • Each square is further divided into LSxLS squares
  • nodes are uniformly distributed among squares
  • Locality based preferential network connections
    (uses Waxman probabilistic
  • Degree distribution is also preserved for nodes

  • Internet is growing in hierarchical structure
  • Opposite to our intuitions, internet topology
    generators using power law degree
    distribution(BRITE and INET), perform better than
    structural generators(TIERS).

Network Topology Generators Degree based vs
Structural H. Tangmunarunkit SIGCOMM 2002
  • Arguing that modeling large scale structure of
    internet(hierarchical structure) should be more
    important than local properties(degree
  • Defined metrics to compare degree based and
    structural generators.
  • Degree based generators are surprisingly perform
    well for large scale metrics.

IGen Generation of Router-level
Internet Topologies through Network Design
Heuristics B. Quoitin 2009
  • Router level topology generator
  • Most of the previous internet topology generators
    use random networks, which is not a realistic
  • Internet topology is highly engineered, optimized
    for costs and affected by business relations
    among Autonomous Systems.
  • Topology design is a complex issue. O(n5)
  • Engineers use heuristics (MENTOR,
    MENTour,Delaunay triangulation and two-trees.
  • IGen tries to mimic the engineering approach by
    using these heuristics.

(No Transcript)
Subnet Level Topology Generation
  • Objectives
  • Subnet level design (unique for now)
  • Realistic Topologies(may use heuristics as in
  • Compatible to famous simulators (NS-2)
  • Highly configurable to fit various network
    applications and satisfy user needs.
  • User-friendly gui

  • A. Medina, A.Lakhina, I. Matta, J. Byers,
    BRITE Universal Topology Generation from a
    Users Perspective Ninth IEEE International
    Symposium on Modeling, Analysis, and Simulation
    of Computer and Telecommunications Systems
    (MASCOTS01), Cincinnati, Ohio, 2001.
  • B. Quoitin, V.V.D. Schrieck, P. Francois and O.
    Bonaventure, IGen Generation of Router-Level
    Internet Topologies through Network Design
    Heuristics 21st international Teletraffic
    Congress,Paris, France 2009
  • C. Jin, Q. Chen, S. Jamin, Inet, Internet
    Topology Generator , ACM SIGCOMM Computer
    Communication Review, vol.32 Issue 4, pp.147-159,
    October 2002.
  • H. Tangmunarunkit, R. Govinan, S. Jamin, S.
    Shenker, W. Willinger, Network Topologies, Power
    Laws, and Hierarchy, ACM Sigcimm Computer
    Communication Review, vol.32, Issue 1,
    pp.76-76, 2002.
  • V. Paxson, S. Floyd, why we dont know how to
    simulate the internet, Proceedings of 29th
    conference on Winter Simulation, pp.1037-1044,
    Atlanta, Georgia, United States, 1997.
  • H. Haddadi, S. Uhlig, A. Moore, R. Mortier, M.
    Rio, Modeling Internet Topology Dynamics, ACM
    SIGCOMM Computer Communication Review, vol.38,
    Issue 2, pp.65-68, April 2008
  • M.B. Doar, A Better Model for Generating Test
    Networks, GLOBECOM Global Telecommunications
    Coference, pp.86-93, London, Novenber 1996.
  • R. Govindan, H. Tangmunarunkit, Heuristics for
    Internet Map Discovery , INFOCOM 2000
  • B.M. Waxman, Routing of Multipoint Connections,
    IEEE Journal of selected Areas in Communications,
    vol.6, no9, pp.1617-1622, December 1988.

Thank you
  • Questions
  • ?