In this paper we study the statistics of the realized throughput of elastic document transfers, accounting for the way network bandwidth is shared dynamically between the randomly varying number of concurrent flows. We first discuss the way TCP realizes statistical bandwidth sharing, illustrating essential properties by means of packet level simulations. Mathematical flow level models based on the theory of stochastic networks are then proposed to explain the observed behavior. A notable result is that first order performance (e.g., mean throughput) is insensitive with respect both to the flow size distribution and the flow arrival process, as long as "sessions" arrive according to a Poisson process. Perceived performance is shown to depend most significantly on whether demand at flow level is less than or greater than available capacity. The models provide a key to understanding the effectiveness of techniques for congestion management and service differentiation.

/pdf/statistical-bandwidth-sharing-a-study-of-congestion-at-flow-385wz3auim.pdf

Statistical bandwidth sharing: a study of congestion at flow level

Dynamic multi-path routing has the potential to improve the reliability and performance of a communication network, but carries a risk. Routing needs to respond quickly to achieve the potential benefits, but not so quickly that the network is destabilized. This paper studies how rapidly routing can respond, without compromising stability.We present a sufficient condition for the local stability of end-to-end algorithms for joint routing and rate control. The network model considered allows an arbitrary interconnection of sources and resources, and heterogeneous propagation delays. The sufficient condition we present is decentralized: the responsiveness of each route is restricted by the round-trip time of that route alone, and not by the round-trip times of other routes. Our results suggest that stable, scalable load-sharing across paths, based on end-to-end measurements, can be achieved on the same rapid time-scale as rate control, namely the time-scale of round-trip times.

/pdf/stability-of-end-to-end-algorithms-for-joint-routing-and-4zm492ljeq.pdf

Stability of end-to-end algorithms for joint routing and rate control

In this paper, we develop a framework for user association in infrastructure-based wireless networks, specifically focused on flow-level cell load balancing under spatially inhomogeneous traffic distributions. Our work encompasses several different user association policies: rate-optimal, throughput-optimal, delay-optimal, and load-equalizing, which we collectively denote α-optimal user association. We prove that the optimal load vector ρ* that minimizes a generalized system performance function is the fixed point of a certain mapping. Based on this mapping, we propose and analyze an iterative distributed user association policy that adapts to spatial traffic loads and converges to a globally optimal allocation. We then address admission control policies for the case where the system is overloaded. For an appropriate system-level cost function, the optimal admission control policy blocks all flows at cells edges. However, providing a minimum level of connectivity to all spatial locations might be desirable. To this end, a location-dependent random blocking and user association policy are proposed.

Distributed $\alpha$ -Optimal User Association and Cell Load Balancing in Wireless Networks

Based on an analysis of the statistical nature of IP traffic and the way this impacts the performance of voice, video, and data services, we question the appropriateness of commonly proposed quality-of-service mechanisms. This paper presents the main points of this analysis. We also discuss pricing issues and argue that many proposed schemes are overly concerned with congestion control to the detriment of the primary pricing function of return on investment. Finally, we propose an alternative flow-aware networking architecture based on a novel router design called cross-protect. In this architecture, performance requirements are satisfied without explicit service differentiation, creating a particularly simple platform for the converged network.

Internet traffic, QoS, and pricing

Modeling integration of streaming and data traffic

It is proposed to apply an integrated admission control scheme to both streaming flows and elastic flows. It is assumed that streaming flowpac kets are served with priority in network queues so that admission control ensures minimal delay for audio and video applications while preserving the throughput of document transfers. An implicit measurementbased implementation is proposed where admissibility is based on an estimation of the bandwidth a newelastic floww ould acquire. An analytical fluid flowmo del provides insight and guides the choice of admission thresholds. Detailed packet level simulations of TCP and UDP connections showthat the proposed algorithms work satisfactorily in the range of admission thresholds predicted by the fluid model.

Integrated Admission Control for Streaming and Elastic Traffic

Quality of service and flow level admission control in the internet

We propose and evaluate an implicit measurement-based scheme for elastic flow admission control in the Internet. We first discuss the nature of IP traffic and present a simple fluid flow model of statistical bandwidth sharing on an isolated bottleneck link. This model is used to guide our choice of admission control algorithms. The main contribution of the paper is to demonstrate by means of detailed packet level simulations that the proposed scheme is efficient and perfectly feasible.

S. Ben Fredj

Papers

Integrated Admission Control for Streaming and Elastic Traffic

Quality of service and flow level admission control in the internet

Measurement-based admission control for elastic traffic