vii 1 Introduction 1 1.1 Relation to Prior Work : : : : : : : : : : : : : : : : : : : : : : : : : : 2 1.2 Contribution and Organization of the Paper : : : : : : : : : : : : : : 3 2 Problem Formulation 4 3 Message Transmission Problem 5 3.1 Shortest-Widest Paths : : : : : : : : : : : : : : : : : : : : : : : : : : 5 3.2 Properties of Multipaths : : : : : : : : : : : : : : : : : : : : : : : : : 6 3.3 NP-Completeness of MTP : : : : : : : : : : : : : : : : : : : : : : : : 10 3.4 Approximate Routing Algorithm : : : : : : : : : : : : : : : : : : : : 14 3.5 Relation to Maximum Flow Algorithm : : : : : : : : : : : : : : : : : 16 3.6 Simulation Results : : : : : : : : : : : : : : : : : : : : : : : : : : : : 18 3.7 Delay-Bandwidth Product : : : : : : : : : : : : : : : : : : : : : : : : 24 4 Sequence Transmission Problem 25 4.1 Intractability Results : : : : : : : : : : : : : : : : : : : : : : : : : : : 25 4.2 Approximation Algorithm : : : : : : : : : : : : : : : : : : : : : : : : 28 5 Concl...

Routing as well as the management of communication networks that support hybrid types of communications requiring quality of service is a very hard problem. We present here a framework 1 that decomposes the network into a hierarchy of abstract views of the network that summarizes the available bandwidth resources and highlights bottlenecks in the network in order to reduce the complexity of the previously mentioned tasks. This framework can easily be distributed to a hierarchy of intelligent agents. This framework is technology independent and can be applied to any circuit-switched communication network. Keywords: communication network, quality of service, routing, network management, network planning, abstractions. 1 Introduction Many applications, especially those related to multimedia, require that information is transmitted in a communication network with certain quality of service guarantees. Such guarantees can only be given in circuitswitched networks, such as ATM ( [ 6 ] ). ...

Since real-time applications usually require not only timeliness but also fault-tolerance, it is essential to incorporate fault-tolerance into real-time communication services that are indispensable to distributed realtime applications. The techniques for failure recovery in datagram communication are not adequate for real-time communication, because they cannot provide recovery-delay guarantees. To ensure fast recovery of a real-time channel from network component failures, we need to reserve network resources (called "spare resources") along a backup route before failures actually occur. The focus of this paper is on minimizing the amount of spare resources while meeting the fault-tolerance requirement. Specifically, we present resource sharing mechanisms and backup-route selection algorithms, and evaluate their efficiency with extensive simulations. 1 Introduction In an Internet-like multihop packet-switched network, messages experience random delays due to their contention for ne...

Quality of Service (QoS) based routing provides QoS guarantees to multimedia applications and an efficient utilization of the network resources. Nevertheless, QoS routing is likely to be a costly process that does not scale when the number of nodes increases. Thus, the routing algorithm must be simple and a class-of-service routing is an alternative to provide QoS guarantees instead of per-flow routing. This paper proposes and analyzes the performance of a distance-vector QoS routing algorithm that takes into account three metrics: propagation delay, available bandwidth, and loss probability. Classes of service and metriccombination are used to turn the algorithm scalable and as complex as a two-metric one. The processing requirements of the proposed algorithm and those of an optimal approach are compared and the results show an improvement up to 50%.

Quality of Service (QoS) based routing provides QoS guarantees to multimedia applications and an efficient utilization of the network resources. Nevertheless, QoS routing is likely to be a costly process that does not scale when the number of nodes increases. Thus, the routing algorithm must be simple. This paper proposes and analyzes the performance of a link-state QoS routing algorithm that takes into account three metrics: propagation delay, available bandwidth, and loss probability. A heuristic based on the residual loss probability and metric combination is used to turn the algorithm scalable and solvable in polynomial time. The simulation results show that our algorithm is a promising solution to construct paths constrained on three metrics.

In a number of distributed computing applications, messages must be transmitted on demand between processes running at di erent locations on the Internet or a dedicated wide-area network. The end-to-end delays experienced by the messages have a significant "random" component due to the complicated nature of network traffic. We propose a method based on delay regression estimation to achieve low end-to-end delays for message transmissions in distributed computing applications. Two-paths are utilized between various communicating processes in a transparent manner by handling all the required networking operations. The main strengths of our approach are the analytical guarantees on the end-to-end delays based entirely on measurements without requiring probabilistic traffic models. Our scheme is implemented over the Internet by a network of NetLets, which communicate with one another to maintain an accurate "state" of delay-regressions in the network. NetLets handle all network trac between the processes and also perform routing at certain level depending on the underlying network. We present experimental results to illustrate that NelLets provide a viable and practical means for achieving end-to-end delays for distributed computing processes over the Internet.

We consider the transmission of a message of size r from a source to a destination over a computer network with n nodes and m links. There are three sources of delays: (a) propagation delays along the links, (b) delays due to bandwidth availability on the links, and (c) queuing delays at the intermediate nodes. First, we consider that the delays on various links and nodes are given as functions of the message size. If the delay in (b) is a non-increasing function of the bandwidth, we propose O(m 2 + mn log n) time algorithm to compute a path with the minimum end-to-end delay for any given message size r. We then consider that the queuing delay in (c) is a random variable correlated with the message size according to an unknown distribution. At each node, the measurements of queuing delays and message sizes are available. We propose two algorithms to compute paths whose delays are close to optimal delays with a high probability, irrespective of the distribution of the delays. 1. Intr...

Quality of Service (QoS) based routing provides QoS guarantees to multimedia applications and an efficient utilization of the network resources. Nevertheless, the QoS routing algorithm must be simple because a costly procedure does not scale with the size of the network. This paper proposes and analyzes the performance of the Single Mixed Metric (SMM) routing protocol in two versions: one based on distance-vectors and the other on link-states. SMM takes into account three metrics: propagation delay, available bandwidth, and loss probability. A heuristic based on the residual loss probability and metric-combination is used to turn the algorithm scalable and solvable in polynomial time. The simulation results show that SMM outperforms other solutions that implement QoS routing based on three metrics.

This paper proposes a scalable algorithm for Quality of Service (QoS) based routing. This type of routing is one of the main components for supporting new multimedia applications in the Internet. The algorithm takes into account the available bandwidth, the transmission delay and the packet loss probability for each link in route computation. This three-metric algorithm is as complex as a two-metric algorithm and provides a better characterization of application requirements. The mechanism was implemented using a network simulator and the results obtained are 25% better than the classical approach.

this paper, we consider QoS support in the context of group communications.