Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
The provisioning of dynamic forms of services is becoming the main stream of today’s network. In this paper, we focus on services assisted by network servers and different forms of associated sessions. We identify two types of services:... more
The provisioning of dynamic forms of services is becoming the main stream of today’s network. In this paper, we focus on services assisted by network servers and different forms of associated sessions. We identify two types of services: transparent, where the session is unaware of the server location, and configurable, where the sessions need to be configured to use their closest server. For both types we formalize the problem of optimally placing network servers and introduce approximated solutions. We present simulation result of approximations and heuristics. We also solve the location problem optimally for a special topology. We show, through a series of examples, that our approaches can be applied to a variety of different services. Placing Servers for Session-Oriented Services Sumi Choi and Yuval Shavitt
Research Interests:
Developing an evolution model of the Internet has been a long standing research challenge. Such a model can improve the design and placement of communication infrastructure, reducing costs and improving users' quality of experience.... more
Developing an evolution model of the Internet has been a long standing research challenge. Such a model can improve the design and placement of communication infrastructure, reducing costs and improving users' quality of experience. While communication infrastructure is tightly coupled to geographical locations, Internet modelling and forecasting in the last decade used network elements that are only loosely bounded to any geographical location. In this paper we set the foundations for developing an evolution model of the Internet based on the Point of Presence (PoP) level. As PoPs have a strong geographical grip they can better represent the evolution of the Internet. We annotate the PoP topologies of the Internet with geographical, economic and demographic information to achieve an understanding of the dynamics of the Internet's structure, in order to identify the constitutive laws of Internet evolution. We identify GDP as the strongest indicator on the country level, and ...
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
Research Interests:
There is an increasing need to quickly and efficientlylearn network distances, in terms of metrics such as latency orbandwidth, between Internet hosts. For example, Internet contentproviders often place data and server mirrors throughout... more
There is an increasing need to quickly and efficientlylearn network distances, in terms of metrics such as latency orbandwidth, between Internet hosts. For example, Internet contentproviders often place data and server mirrors throughout the Internetto improve ...
Research Interests:
The stochastic knapsack problem is a stochastic version of the well known deterministic knapsack problem, in which some of the input values are random variables. There are several variants of the stochastic problem. In this paper we... more
The stochastic knapsack problem is a stochastic version of the well known deterministic knapsack problem, in which some of the input values are random variables. There are several variants of the stochastic problem. In this paper we concentrate on the chance-constrained variant, where item values are deterministic and item sizes are stochastic. The goal is to find a maximum value allocation subject to the constraint that the overflow probability is at most a given value. Previous work showed a PTAS for the problem for various distributions (Poisson, Exponential, Bernoulli and Normal). Some strictly respect the constraint and some relax the constraint by a factor of (1 + ). All algorithms use Ω(n1/ ) time. A very recent work showed a “almost FPTAS” algorithm for Bernoulli distributions with O(poly(n) · quasipoly(1/ )) time. In this paper we present a FPTAS for normal distributions with a solution that satisfies the chance constraint in a relaxed sense. The normal distribution is part...
Research Interests:
Research Interests:
This work presents a thorough investiga- tion of the structure of multicast trees cut from the Internet and power-law topologies. Based on both generated topolo- gies and real Internet data, we characterize the structure of such trees and... more
This work presents a thorough investiga- tion of the structure of multicast trees cut from the Internet and power-law topologies. Based on both generated topolo- gies and real Internet data, we characterize the structure of such trees and show that they obey the rank-degree power law; that most high degree tree nodes are concentrated in a low diameter neighborhood; and that the sub-tree size also obeys a power law. Our most surprising empirical finding suggests that there is a linear ratio between the number of high- degree network nodes, namely nodes whose tree degree is higher than some constant, and the number of leaf nodes in the multicast tree (clients). We also derive this ratio analytically. Based on this finding, we develop the Fast Algorithm, that estimates the number of clients, and show that it converges faster than one round trip delay from the root to a randomly selected client.
Research Interests:
Research Interests:
Research Interests: Engineering, Computer Science, Technology, Computer Networks, Routing, and 14 moreConvergence, Internet Routing Protocol, Routing protocols, Simulation, Convergence Rate, Amplified spontaneous emission, Time Complexity, Empirical Study, Routing Protocol, Dynamic Panel Data, BGP, Border Gateway Protocol, Service Provider, and Internet
Research Interests:
The IDMaps project aims to provide a distance map of the Internet from which relative distances between hosts on the Internet can be gauged [1]. Many distributed systems and applications can benefit from such a distance map service, for... more
The IDMaps project aims to provide a distance map of the Internet from which relative distances between hosts on the Internet can be gauged [1]. Many distributed systems and applications can benefit from such a distance map service, for example, a common method to improve user perceived performance of the Internet is to place data and server mirrors closer to clients. When a client tries to access a mirrored server, which mirror should it access? With IDMaps, the closest mirror can be determined based on distance estimates between the client and the mirrors. In this paper we investigate both graph theoretic methods and ad hoc heuristics for instrumenting the Internet to obtain distance maps. We evaluate the efficacy of the resulting distance maps by comparing the determinations of closest replica using known topologies against those obtained using the distance maps. I. INTRODUCTION The IDMaps project [1] aims at providing a distance 1 map of the Internet from which relative distanc...
Research Interests:
Distance estimation is important to many Internet applications. It can aid a WWW client when selecting among several potential candidate servers, or among candidate peer-to-peer servers. It can also aid in building efficient overlay or... more
Distance estimation is important to many Internet applications. It can aid a WWW client when selecting among several potential candidate servers, or among candidate peer-to-peer servers. It can also aid in building efficient overlay or peer-to-peer networks that dynamically react to change in the underlying Internet. One of the approaches to distance (i.e., time delay) estimation in the Internet is based on placing Tracer stations in key locations and conducting measurements between them. The Tracers construct an approximated map of the Internet after processing the information obtained from these measurements.
Research Interests: Algebra, Computer Science, Distributed Computing, Data Mining, Performance, and 14 moreTopology, Internet Measurement, Algorithm, World Wide Web, Distance Estimation, Peer to Peer, Rich Internet Application, Network Topology, Web Server, Internet, Electrical And Electronic Engineering, Www, Time measurement, and Time delay estimation
Most research on Internet topology is based on active measurement methods. A major difficulty in using these tools is that one comes across many unresponsive routers. Different methods of dealing with these anonymous nodes to preserve the... more
Most research on Internet topology is based on active measurement methods. A major difficulty in using these tools is that one comes across many unresponsive routers. Different methods of dealing with these anonymous nodes to preserve the connectivity of the real graph have been suggested. One of the more practical approaches involves using a placeholder for each unknown, resulting in multiple copies of every such node. This significantly distorts and inflates the inferred topology. Our goal in this work is to unify groups of placeholders in the IP-level graph. We introduce a novel clustering algorithm based on semisupervised spectral embedding of all the nodes followed by clustering of the anonymous nodes in the projected space. Experimental results on real internet data are provided, that show good similarity to the true networks. 1.
Research Interests:
Research Interests:
Several important management applications, e.g., billing, generate a relatively high volume of traffic which is not delay sensitive. Both the management and the user traffic use the same network infrastructure. However, the management... more
Several important management applications, e.g., billing, generate a relatively high volume of traffic which is not delay sensitive. Both the management and the user traffic use the same network infrastructure. However, the management traffic does not yield any immediate benefits for the users, and is thus an overhead from their point of view. In a previous work it was suggested to store such traffic inside the network in case of congestion, e.g., by using the power of active networks to look for the least expensive place one can backtrack to. In this paper, we suggest instead to use active networks in order to deflect the time insensitive management traffic from congested areas. To optimize the deflection we use the recent discovery about the power law characteristics of the Internet, and show that this knowledge improves our deflection performance. We further show that by applying learning techniques on locally gathered deflection information performance improves.
Research Interests: Engineering and Power Law
Research Interests:
Research Interests:
In Chapter 3, we formulated IDMaps Tracer placement as a graph-theoretic problem and established a framework for building overlays with bounded latency distortion. On the Internet, a direct application of our theoretical results is not... more
In Chapter 3, we formulated IDMaps Tracer placement as a graph-theoretic problem and established a framework for building overlays with bounded latency distortion. On the Internet, a direct application of our theoretical results is not feasible. First, the Internet topology is unknown and likely to remain so in the future. The sheer size of the Internet makes any effort to discover its topology extremely difficult. Second, even if the Internet topology were known, it is unlikely that one can place Tracers at locations dictated by a placement algorithm. Placement of Tracers on the Internet is not merely a technical problem. There are financial constraints such as the cost associated with buying network connection and bandwidth, and logistical constraints such as the administration of Tracers once they are installed. It is more likely that these constraints will determine where Tracers can be placed on the Internet.
Research Interests:
It is increasingly the case that a network service will need to communicate over some type of logical or overlay network, built on top of a physical network. For example, in user-level multicast, there is a need to construct an overlay... more
It is increasingly the case that a network service will need to communicate over some type of logical or overlay network, built on top of a physical network. For example, in user-level multicast, there is a need to construct an overlay network to support one-to-many or many-to-many communication. With one-to-many communication, there is often a trade off between communication cost and end-to-end performance. At one extreme, one can obtain excellent performance by having direct pair-wise communication channels, hence foregoing the benefit of multicast. At the other extreme, a minimum spanning tree can be established among participants to minimize communication cost, but the end-to-end performance may be very poor. Past studies [Deering et al., 1996, Wei and Estrin, 1994] have clearly highlighted this trade off, and existing designs tend to favor lower communication cost because of scalablity. However, minimizing communication cost alone is not always enough to have a scalable system. If packet forwarding involves many nodes in a multicast network such as in a minimum spanning tree, it would be difficult to build a service because the service requires the deployment of many nodes with forwarding capability. Thus, for a scalable service, minimizing the number of service nodes should also be an objective.