This paper proposes a new measure for network performance evaluation called topology lifetime. The measure provides insight into which one of a set of topologies is likely to last the longest before more capacity must be installed. The lifetime measure is not single valued, but considers growth as a function of a set of demand shifts (perturbation). One network may be better able to support a uniform growth in the traffic, while another may support more growth when unexpected shifts in the load occur. The ability of a network to support unexpected changes in load is becoming more important because of: 1) current practices for installing fiber optics cables; 2) recent advances in dense wavelength division multiplexing; and 3) the increasing popularity of the Internet. The lifetime measure is applied to several topologies; a dual ring, a chordal ring, a Manhattan Street network and an hierarchical network. We also apply the measure to a realistic US IP Backbone network. In this paper, our objective is to show how to apply the measure to different networks, and to explain certain implications for comparisons between networks. We expect this measure to be useful both in the construction of new networks and in selecting between new links that may be added to an existing network.

Optical burst switches (OBSes) have been proposed to improve the utilization of a network of optical cross connect (OXCs). Current studies on OBS assume a network consisting of OBSes alone. While this is a reasonable assumption for evaluating a new technology, the question of how a network of OXCs can be evolved to a network of OBSes has not been studied. In this paper, we propose a hybrid architecture consisting of OBSes at the network edge and OXCs in the network core. This architecture allows carriers to gradually migrate from an OXC-based network to an OBS-based network with an improved network utilization. In addition, we use queueing analysis to study the performance of this new architecture.

We advocate the use of random exponential marking (REM) - an optimization flow control approach - as an extension to the assured forwarding (AF) service of differentiate services Internet that is well-suited to adaptive applications. The main features are that it offers per-flow signaling, hence allows for end-to-end adaptive users, but enables routers to work on aggregate flows so that the network is scalable. It allows the network to offer reliable service assurances via a distributed optimization algorithm, without introducing special protocols or requiring routers to have access of individual user requirements. Apart from these features, it provides an incentive for users to cooperate so that optimum performance can be accomplished.

This paper proposes to use a programmable network to offer a new type of wholesale product. By augmenting the wholesale of carriage with the wholesale of network-based computation, the need for server overlay deployment to support retail service providers is reduced or eliminated. The paper develops an efficient algorithm for allocation and management of processing and transport resources. This algorithm is designed to optimize a large network supporting many retail service providers. The algorithm is based on the Lagrangian relaxation method and provides error bounds so that the accuracy of the solution can be ascertained. Numerical results are presented and the benefit of the optimization for a simple multiservice scenario is assessed.

This paper explores the potential benefits of programmable networks for telecommunication carriers. It particularly focuses on using a programmable network to offer a new type of wholesale product by adding value to the wholesale of pure carriage. By augmenting the wholesale of carriage with the wholesale of network-based computation, the need for server overlay deployment to support retail service providers is reduced or eliminated. The paper develops an efficient algorithm for allocation and management of computation and transport resources within the network infrastructure. This algorithm is designed to optimize a large network supporting many retail service providers and can be used by the wholesaler for the optimal management of scarce computational and transport resources. The algorithm is based on the Lagrangian relaxation method and provides error bounds so that the accuracy of the solution can be ascertained. Numerical results are presented and the benefit of the optimization for a simple multi-service scenario is assessed.

This paper proposes a new measure for network performance evaluation called topology lifetime. The measure provides insight into which one of a set of topologies is likely to last the longest before more capacity must be installed. The lifetime measure is not single valued, but considers growth as a function of a set of demand shifts. The lifetime measure is applied to several topologies: a dual ring, a chordal ring, a Manhattan Street network and an hierarchical network.

In this paper an alternative hybrid network solution is evaluated, which we have called the Internet Service Provider Access Network (ISPAN).

This paper deals with the problem of dynamic allocation of bandwidth in a hierarchical transport platform. By hierarchical we mean that the crossconnect platform is comprised of both broadband and wideband crossconnects which react to fluctuations of traffic on different time scales. The model developed in this paper takes advantage of both the multi-service and multi-hour character of traffic profiles. Obtaining a minimum-cost network for multi-hour traffic profile for multi-service is a large scale optimisation problem. We formulate the problem of capacity variation for a long term time scale as mixed integer programming and for a short term time scale as linear programming problems and examine the problem complexity in terms of the number of variables and constraints.