T. Bu and D. Towsley, "On Distinguishing between Internet Power Law Topology Generators," Proc. IEEE INFOCOM, 2002.  Home/Search   Document Details and Download   Summary   Related Articles   Check

....by and large the possibility of the power laws being the result of coincidence. Therefore, the power laws appear as a necessary though not sucient condition for a topology to be realistic. There may be more topological properties of the Internet topology that are not captured by our power laws [45, 54]. The rest of this paper is structured as follows. In Section 2, we present some de nitions and previous work on measurements and models for the Internet. In Section 3, we present our Internet instances and provide useful measurements. In Section 4, we present our three observed power laws and ....

....with exponent 0.8. Our initial work [20] on power laws has generated signi cant follow up work. Various researchers veri ed our observations with di erent datasets[24, 23, 33] In addition, signi cant work has been devoted in understanding the origin [36] and generating power law topologies [35, 36, 42, 26, 54, 58]. We discuss these approaches for generating power laws in section 6. More recently, several works have focused on describing the topology in a qualitatively way [53, 31, 32, 52] 3 Our Internet Instances In this section, we present the Internet instances we study in our work. We use topologies ....

[Article contains additional citation context not shown here]

T.Bu and D. Towsley. On distinguishing between Internet power law topology generators. Proc. IEEE INFOCOM, 2002.

....distributions about the Internet, leading to the creation of new Internet topology generators. Tangmunarunkit et al. divide network topology generators into two categories [38] Structural and Degree Based network generators. Other recently proposed generators are [1] 14] 39] 40] 41] [42]. The major difference between these two categories is that the former explicitly injects hierarchical strcuture into the network, while the later generates graphs with power law degree distributions without any consideration of network hierarchy. Tangmunarunkit et al. argue that even though ....

....which is well matched to real Internet topology. Characteristics of the Internet topology and its robustness against failures have been widely studied [1] 5] 6] 14] with focus on extracting common regularities from several snapshots of the real Internet topology. On the other hand, [42], 43] have shown that the clustering coefficient of the Internet has been growing and that the average diameter of the Internet has been decreasing over the past few years. However, 43] used this characteristic only as evidence of topology stability. III. NETWORK MODEL AND SIMULATION ....

[Article contains additional citation context not shown here]

T. Bu and D. Towsley, "On Distinguishing between Internet Power Law Topology Generators," in Proceedings of INFOCOM, 2002.

.... may give rise to powerlaw degree distributions in graphs [5, 20, 19] The most prominent model attempting to explain the emergence of power law degree distributions is the Barab asi Albert model (or B A model) 5, 2] In fact, it has been considered in a number of papers as a model for AS graphs [3, 7, 27, 24, 32]. The B A model assumes the network is formed through incre1 mental addition of nodes. In the simplest form of the model, a new node forms a connection to an existing node with probability proportional to the existing node s degree. This preferential connectivity leads to a rich get richer ....

....vary. The work in [27] investigates the case where only a subset of all nodes in the network are available for connection. With only slight modifications to the B A model they show that a power law degree distribution emerges. Additionally, a generalized linear preference model is proposed in [7] that better matches the clustering behavior and path lengths of empirical Internet measurements. These extensions have improved the flexibility of the B A model, albeit with a corresponding increase in complexity. The generation of power laws through random graph models has also received ....

Tian Bu and Don Towsley. On Distinguishing between Internet Power Law Topology Generators. In IEEE INFOCOM, New York, NY, June 2002.

....Computing, Georgia Institute of Technology, Atlanta, GA. email: mihail cc.gatech.edu College of Computing, Georgia Institute of Technology, Atlanta, GA. email: ewz cc.gatech.edu perlink between the corresponding pages, the Internet at the level of Autonomous Systems (a.k.a. inter domain level) [16, 24, 29, 34, 10, 11, 36, 4] where a node corresponds to a distinct routing administration domain (such as a University, a corporation, or an ISP) and an edge represents direct exchange of tra#c between the corresponding domains, and biological networks [20] where, nodes correspond to genetic or metabolic building blocks ....

....the case for the observed numbers of the parameters # and # mentioned above) On the other hand, graph models for complex networks are often expected to pass strict performance requirements. For example, the networking community uses such graph models to simulate a wide range of network protocols [40, 16, 24, 30, 29, 34, 10, 11, 36, 4], and hence the accuracy of the underlying topology model is considered very important. Therefore, the following alternative degree driven approach for generating network topology models has been adopted. First predict the degrees of the graph to be generated by extrapolation from available data, ....

Tian Bu and Don Towsley. On distinguishing between internet power law topology generators. In Proc. Infocom. IEEE, 2002.

.... may give rise to powerlaw degree distributions in graphs [5, 20, 19] The most prominent model attempting to explain the emergence of power law degree distributions is the Barab asi Albert model (or B A model) 5, 2] In fact, it has been considered in a number of papers as a model for AS graphs [3, 7, 27, 24, 32]. The B A model assumes the network is formed through incre1 mental addition of nodes. In the simplest form of the model, a new node forms a connection to an existing node with probability proportional to the existing node s degree. This preferential connectivity leads to a rich get richer ....

....vary. The work in [27] investigates the case where only a subset of all nodes in the network are available for connection. With only slight modifications to the B A model they show that a power law degree distribution emerges. Additionally, a generalized linear preference model is proposed in [7] that better matches the clustering behavior and path lengths of empirical Internet measurements. These extensions have improved the flexibility of the B A model, albeit with a corresponding increase in complexity. The generation of power laws through random graph models has also received ....

Tian Bu and Don Towsley. On Distinguishing between Internet Power Law Topology Generators. In IEEE INFOCOM, New York, NY, June 2002.

....level. Using such data Faloutsos et al. first observed that the degree distribution of the AS level topology is actually consistently highly skewed [11] Consequently, the research community has shown considerable interest in obtaining topology models that better resemble the real data [2] [5], 18] 23] as well as understanding the impact of such network topologies on the performance of network protocols [27] 32] This new generation of synthetic Internet topology models is strongly driven by the observed skewed statistics of the degree sequence and its evolution, and by even ....

....by the observed skewed statistics of the degree sequence and its evolution, and by even further observations of more detailed graph theoretic characteristics of the network. Most notably, following the natural intuition that, for example, geography must be relevant in the real Internet topology, [5] paid special attention to the clustering coefficient; the observation of the significance of geography has been also made in [39] and [21] In this paper we revisit the issue of clustering. As opposed to previous work that has focused on the clustering coefficient, our starting point is the ....

[Article contains additional citation context not shown here]

Tian Bu and Don Towsley. On distinguishing between internet power law topology generators. Infocom 2002.

.... e#orts has been to focus on matching a sequence of easily understood metrics or observed features of interest; e.g. explicitly imposed connectivity properties or hierarchical structures (see [33] and references therein) empirical node degree distributions [21, 23, 1, 7] clustering coe#cients [8], etc. see for example [30] for additional candidate metrics) However, this type of descriptive or evocative modeling can be misleading, since the question of which metrics or features are the most important ones for judging and comparing di#erent Internet topologies remains largely ....

T. Bu and D. Towsley. On distinguishing between Internet power law topology generators. Proc. IEEE INFOCOM'02, 2002.

....topology of the Internet. The AS topology can be represented as an undirected graph: each vertex corresponds to an AS, and two vertices are joined by an edge if there is at least one physical link between the corresponding ASs. The AS topology has been investigated by a number of authors, see e.g. [6, 13, 4, 15]. Di erent ASs exchange routing information using the Border Gateway Protocol (BGP) If an AS X is connected to another AS Y , it announces the routes to a certain set A(X; Y ) of destination addresses to Y . This implies that when Y has a packet with a destination in the set A(X; Y ) it can ....

T. Bu and D. Towsley. On distinguishing between Internet power law topology generators. In Proceedings of INFOCOM'02, June 2002.

No context found.

T. Bu and D. Towsley, "On distinguishing between Internet power law topology generators," in Proceedings of IEEE INFOCOM, 2002.

No context found.

T. Bu and D. Towsley. On distinguishing between Internet power law topology generators. In Proceedings of IEEE INFOCOM, 2002.

No context found.

T. Bu and D. Towsley. On distinguishing between Internet power law topology generators. In Proceedings of IEEE INFOCOM, 2002.

No context found.

T. Bu and D. Towsley,"On Distinguishing between Internet Power Law Topology Generators," in Proc. Infocom 2002, June 23--27, New York.

....model [1] and topology generators such as Inet [21] and generation models in BRITE[27] measured success primarily in terms of graph connectivity properties, such as node degree distributions. An active debate about the merits and limitations of these approaches is ongoing [21] 24] 7] [4]; the jury is still out on which models are best and studies have shown varying conclusions depending on the generators used [31] Our goal is not to propose a new topology generation method in this paper, but to suggest a wider set of bases for the construction of topology generation tools. To ....

T. Bu and D. Towsley. On Distinguishing between Internet Power Law Topology Generators. In Proceeedings of IEEE INFOCOM '02, 2002.

....lattice BRITE is available at http: cs pub.bu.edu brite 10 100 1000 10000 number of infected nodes time simulation B C D Fig. 13. Comparison of results obtained by simulation and approximate analysis on power law graphs generated with the GLP algorithm algorithm described in [16]. A power law graph is defined as a graph in which the ccdf of the node degree d satisfies F (d) # 0. We adopted the GLP algorithm, that was designed to match the power low exponent and the clustering behavior of the AS level Internet topology, just in order to generate topologies very ....

....GLP algorithm, that was designed to match the power low exponent and the clustering behavior of the AS level Internet topology, just in order to generate topologies very different from a ring lattice. The graphs were generated using a constant value 0. 45 for the parameter p of GLP algoritm (see [16]) while trying different values of m, which is the initial connectivity of the nodes as they are added to the graph during the generation process. Figure 13 plots on a log log scale the evolution of the average number of infected nodes in four different cases, comparing results obtained from both ....

T. Bu and D. Towsley,"On Distinguishing between Internet Power Law Topology Generators," in Proc. Infocom 2002, June 23--27, New York.

....model [2] and topology generators such as Inet [20] and generation models in BRITE[25] measured success primarily in terms of graph connectivity properties, such as node degree distributions. An active debate about the merits and limitations of these approaches is ongoing [20] 22] 7] [5]; the jury is still out on which models are best and studies have shown varying conclusions depending on the generators used [29] Our goal is not to propose a new topology generation method in this paper, but to suggest a wider set of bases for the construction of topology generation tools. To ....

T. Bu and D. Towsley. On Distinguishing between Internet Power Law Topology Generators. In Proceeedings of IEEE Infocom, 2002.

....2002 Abstract Recent work has shown the prevalence of small world phenomena [28] in many networks. Small world graphs exhibit a high degree of clustering, yet have typically short path lengths between arbitrary vertices. Internet AS level graphs have been shown to exhibit small world behaviors [9]. In this paper, we show that both Internet AS level and routerlevel graphs exhibit small world behavior. We attribute such behavior to two possible causes namely the high variability of vertex degree distributions (which were found to follow approximately a power law [15] and the preference of ....

....graphs, they have typically short distances between arbitrary pairs of vertices. It has been shown that many networks have similar small world property. Examples include actor collaboration networks, power grids, the World Wide Web links [1] 7] and autonomous system (AS) graphs of the Internet [9]. The second topological feature of recent interest is the skewed degree distributions of network vertices. This feature is present in paper citation databases [26] actor collaboration networks, Web links [5] and the physical connectivity of the Internet [15] In such networks, vertices have a ....

[Article contains additional citation context not shown here]

T. Bu and D. Towsley. On distinguishing between Internet power law topology generators. In Proceedings of INFOCOM, 2002.

....2002 Abstract Recent work has shown the prevalence of small world phenomena [28] in many networks. Small world graphs exhibit a high degree of clustering, yet have typically short path lengths between arbitrary vertices. Internet AS level graphs have been shown to exhibit small world behaviors [9]. In this paper, we show that both Internet AS level and routerlevel graphs exhibit small world behavior. We attribute such behavior to two possible causes namely the high variability of vertex degree distributions (which were found to follow approximately a power law [15] and the preference of ....

....graphs, they have typically short distances between arbitrary pairs of vertices. It has been shown that many networks have similar small world property. Examples include actor collaboration networks, power grids, the World Wide Web links [1] 7] and autonomous system (AS) graphs of the Internet [9]. The second topological feature of recent interest is the skewed degree distributions of network vertices. This feature is present in paper citation databases [26] actor collaboration networks, Web links [5] and the physical connectivity of the Internet [15] In such networks, vertices have a ....

[Article contains additional citation context not shown here]

T. Bu and D. Towsley. On distinguishing between Internet power law topology generators. In Proceedings of INFOCOM, 2002.

No context found.

T. Bu and D. Towsley, "On Distinguishing between Internet Power Law Topology Generators," Proc. IEEE INFOCOM, 2002.

No context found.

T. Bu and D. Towsley, "On distinguishing between Internet power law topology generators," in Proceedings of INFOCOM, 2002.

No context found.

T. Bu and D. Towsley, "On distinguishing between Internet power law topology generators," in IEEE INFOCOM, 2002.

No context found.

T. Bu and D. F. Towsley. On distinguishing between internet power law topology generators. In Proceedings of INFOCOM, 2002.

No context found.

T. Bu and D. Towsley. On distinguishing between Internet power law topology generators. Infocom, 2002.

No context found.

T. Bu and D. Towsley. On distinguishing Between Internet Power Law Topology Generators, IEEE INFOCOM 2002.

No context found.

T. Bu and D. Towsley. On distinguishing between Internet power law topology generators. In IEEE INFOCOM, Apr. 2002.

No context found.

Tian Bu and Don Towsley, "On distinguishing between internet power law topology generators," in Proc. Infocom. IEEE, 2002.

No context found.

Tian Bu and Don Towsley, "On distinguishing between internet power law topology generators," in Proc. Infocom. IEEE, 2002.

No context found.

T. Bu and D. Towsley, "On Distinguishing between Internet Power Law Topology Generators," in Proceedings of INFOCOM, 2002.

No context found.

T. Bu and D. Towsley, "On Distinguishing between Internet Power Law Topology Generators," IEEE INFOCOM, 2002.

No context found.

T. Bu and D. Towsley, "On Distinguishing between Internet Power Law Topology Generators", in Proceedings of the IEEE INFOCOM, 2002.

No context found.

T. Bu and D. Towsley, "On Distinguishing between Internet Power Law Topology Generators," IEEE INFOCOM, 2002.

No context found.

T. Bu, D. Towsley, On Distinguishing between Internet Power Law Topology Generators, in Proceedings of IEEE Infocom, 2002.

No context found.

T. Bu and D. Towsley. On distinguishing between internet power law topology generators. In IEEE Infocom: The 21st Annual Joint Conference of the IEEE Computer and Communications Societies in New York, NY, USA. IEEE Computer Society Press, Los Alamitos, CA, USA, 2002.

No context found.

T. Bu and D. Towsley. On distinguishing between Internet power law topology generators. In IEEE INFOCOM, Apr. 2002.

No context found.

Tian Bu and Don Towsley, "On distinguishing between internet power law topology generators," in Proc. Infocom. IEEE, 2002.

No context found.

Bu, T., Towsley, D.: On Distinguishing between Internet Power Law Topology Generators, in Proc. INFOCOM 2002.

No context found.

T. Bu and D. Towsley. On distinguishing between internet power law topology generators. In Proceedings of IEEE INFOCOM, pages 1587--1596, June 23--27 2002.

No context found.

T. Bu and D. Towsley. On Distinguishing between Internet Power Law Topology Generators. In Proceedings of IEEE INFOCOM, New York, NY, June 2002.

No context found.

T. Bu and D. Towsley. On Distinguishing between Internet Power Law Topology Generators. In Proceedings of IEEE INFOCOM, New York, NY, June 2002.

No context found.

T. Bu and D. Towsley. On distinguishing between Internet power law topology generators. In IEEE INFOCOM, 2002.

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