This paper considers the capacity dimensioning of intersatellite link (ISL) networks in broadband LEO satellite systems, where the major challenge is the topology dynamics. First, a general method to design convenient ISL topologies for connection-oriented operation is presented, and a reference topology for numerical studies is derived. A permanent virtual topology is then defined on top of the physical one, thus forming a framework for discrete-time dynamic traffic routing. On this basis, heuristic and optimization approaches for the combined routing and dimensioning task, operating on discrete time steps, are presented and their performance is numerically compared. It is shown that minimizing the worst case link capacity is an appropriate target function, which can be formulated as linear minmax optimization problem with linear constraints. Using linear programming (LP) techniques, the dimensioning results are clearly better than with simple heuristic approaches. I. INTRODUCTION F...

The planning of multi-service IP networks requires the consideration of many different and also diverse issues. Since Internet technology has been developed in a rather pragmatic way ("we believe in rough consensus and running code"), various approaches and concepts exist in most areas of networking, which are relevant for network planning (e.g., types of applications, routing schemes, Quality of Service mechanisms). In this paper, the key issues of the network planning process for multi-service IP networks are covered in a structured way. To provide a foundation for any network dimensioning procedure, we first introduce and categorize current IP Quality of Service mechanisms and identify their implications for network planning. Furthermore, a systematic approach for classification and modeling of Internet traffic is suggested and a universal link model together with appropriate dimensioning formulae is proposed. Finally, we go over different network planning scenarios and discuss relevant characteristics and possible solution approaches.

Abstract not found

Abstract. This paper proposes an algorithm for cost-efficient network design. The two main network components are the routers and the links connecting them. As these network elements have discrete capacity values, their costs are modeled by stepwise functions. In the current interpretation of the network design problem the subproblems of topological design, resource dimensioning, and routing of traffic flows are handled simultaneously. Using new network technologies (e.g., MPLS) the operators have the opportunity to reconfigure their networks dynamically. By using this function the periodic (e.g., daily, weekly) changes of traffic volumes and directions can be taken into consideration during network design resulting in lower deployment costs. The algorithm proposed for multi-hour design is based on an efficient, published method. The efficiency of the new algorithm is demonstrated by simulation on random and real network topologies with the help of two reference algorithms. Key words. MPLS, traffic engineering, network planning, multi-hour design. 1

Huge increases in switching and transmission capacity coupled with a need to integrate communication services has lead to the development of the asynchronous transfer mode (ATM), a standard for the transport of all telecommunication services over a common network. This thesis describes optimization techniques used to design call admission controls and call routing strategies which optimize the rate of earning revenue in broadband ATM networks carrying multirate tracs. Several optimization techniques are investigated and a new technique called XFG is presented. XFG is a small ecient algorithm for calculating a virtual path connection network (VPCN) that optimizes the rate of earning revenue. The optimal VPCN provides route separation (each origin-destination pair is connected by a dedicated VPC) and service integration (all service classes share a VPC). The thesis presents the theoretical basis of XFG, followed by applications of the algorithm to several test networks. These results ar...

In this paper, we consider configurable capacity tunnels. Static bandwidth allocation (SBA) assigns the network capacity to the tunnels according to the busy hours of their traffic aggregates. At secondary times, their capacity is underutilized and can not be used to accommodate excess traffic of other tunnels. The contribution of this paper is twofold. Firstly, we propose two mechanisms for adaptive bandwidth allocation (ABA) for the tunnels: complete capacity reassignment (CCR) and selective capacity reassignment (SCR). They both adapt the tunnel capacities to the current traffic demands but differ in their implementation, signaling, and configuration complexity. Secondly, we asses the bandwidth savings of ABA vs. SBA in wide area networks where the transfer rates of traffic aggregates fluctuate over time according to busy hours. Our results show that the capacity savings strongly depend on the traffic model and that they may be increased by time-aware routing. 1