NET-COOP 2009
November 23-24-25, 2009
EURANDOM, Eindhoven, The Netherlands
|
|
NET-COOP 2009 |
|
November 23-24-25, 2009 EURANDOM, Eindhoven, The Netherlands |
Abstracts Invited Lectures
Eitan Altman
An Anonymous Sequential Game Approach for Battery State Dependent Power Control
The sensitivity of mobile terminals to energy and power limitations keeps posing challenges to wireless technology. The ratio between the useful signal’s power and that of noise and interferences has a crucial impact on the achievable throughputs and on outage aspects, little has been done concerning another central challenge that limited energy poses: that of limitation on battery life. In this paper we study power control in a way that combines the two aforementioned aspects. We propose a modeling approach which extends the Anonymous Sequential Game framework introduced in 1988 by Jovanovic and Rosenthal. The approach is designed for systems that have a very large number of interacting decision makers, so large that they can be modeled as a continuum of players. We introduce an appropriate equilibrium concept for this game (which extends the Wardrop equilibrium by including random individual states with controlled transitions), characterize the structure of the equilibrium policies and provide two efficient equilibrium computation procedures.
JOINT WORK WITH: Piotr Wiecek and Yezekael Hayel
Rami Atar
Control Formulations under the Nondegenerate Slowdown Diffusion Regime
Control of many-server queueing systems has been the subject of much recent research in the Halfin-Whitt heavy traffic regime. The nondegenerate slowdown regime gives rise to diffusion control problems that are of different nature and it is expected that they lead to closed form solutions. I will describe work in progress on the subject, based on collaborations with Itai Gurvich and Nir Solomon.
Hans van den Berg
Self-Optimisation of Mobile Networks
Significant effort is currently targeted towards the research and development of self-optimisation methods for mobile cellular networks, both in standardisation (e.g. 3GPP) and various (international) collaboration projects. Self-optimisation, in conjunction with self-configuration and self-healing, is seen as a promising opportunity to automate wireless access network planning and optimisation, thus substantially reducing operational costs and improving network coverage, resource utilisation and service quality. Fundamental drivers for the deployment of self-* methods are the increasing complexity of access network technologies, the growing diversity in offered services and the need for enhanced competitiveness. In this lecture we present a vision on the deployment of self-* methods in future mobile cellular networks, and describe the key drivers, potential benefits, some relevant use cases and key challenges.
Costas Courcoubetis
Tariffs, Mechanisms and Internet Equilibria
We analyze the interplay between the demand for downloads, choice of congestion control mechanism, and tariff structure at a single link, when users have preferences in terms of average download delay and they are charged according to the number of ECN marked packets they receive. Our model involves a timescale separation approach, where in the fast timescales active flows compete for instantaneous bandwidth share by optimally tuning congestion control parameters in a noncooperative fashion. This is modeled by letting flows choose utility functions within the network utility maximization framework laid down by Kelly [1]. On a slower timescale, users selfishly change their otherwise unrestricted demand for downloads based on the average experienced download delay and charges incurred. We study the equilibrium of this loop of interactions from the point of view of social welfare.
For homogeneous users we find that optimal equilibria are induced when they choose among linear utility functions, while this is not the case for logarithmic, i.e., proportionally fair congestion controllers. We next consider two types of users, web-browsing and bittorrent, where the latter are much less sensitive to download delay than the former. If bittorrent users respond to congestion according to proportionally fair utility functions the charge induced by ECN marked packets does not provide the correct signal for service differentiation and the resulting equilibrium is suboptimal. On the other hand, if flows are charged for the volume of bits they transfer as well, then social welfare maximum is attained for some price per bit. This reveals a new role for bit volume pricing: to provide bittorrent users the correct incentives for choosing congestion controllers that effectively give priority to web-browsing users.
JOINT WORK WITH: Antonis Dimakis
Mor Harchol-Balter
Surprising Results on Task Assignment in Server Farms under High Variability Workloads
It is well-known that when job size variability is high, one needs to prevent short jobs from getting stuck behind long jobs. In a server farm setting, one way to achieve this goal is to allocate short jobs their own server (or set of servers). This is the theory behind the popular Size Interval Task Assignment policy (SITA) for server farms, which assigns each server a unique size range, so that short jobs are given isolation from long ones. The SITA policy is prevalent throughout compute server farms and manufacturing systems, whenever job size variability is high. The higher the job size variability, the more important it is to provide short jobs some isolation from long ones, via a SITA policy, or some variation thereof.
This talk questions the above common wisdom. To understand what's going on, we study the performance of task assignment policies, in the limit, as the variability of job sizes (service demands) approaches infinity. Results in this limiting regime reveal that the SITA policy can be far inferior to much simpler greedy policies, like Least-Work-Left (LWL), for many common job size distributions, including a range of Pareto distributions. Regimes are also defined where SITA's performance is good, and here simple closed-form bounds are proved on its performance. Towards the end of the talk we will also consider the performance of SITA variants/hybrids.
(Parts of this work appeared in ACM SIGMETRICS 2009.) JOINT WORK WITH: Alan Scheller-Wolf and Andrew Young
Ingemar Kaj
Probabilistic Analysis of Hierarchical Cluster Protocols for Wireless Sensor Networks
Johan van Leeuwaarden
Optimal tradeoff between exposed and hidden nodes in large wireless networks
Denis Miretskiy
Rare-Event Simulation for Tandem Queues: A Simple and Efficient Importance Sampling Scheme
JOINT WORK WITH: Werner Scheinhardt and Michel Mandjes
Maurizio Naldi
Understanding and Preventing Tacit Collusion among Telecommunication Operators
JOINT WORK WITH: Patrick Maille and Bruno Tuffin
Yoni Nazarathy
Optimal File Splitting for Wireless Networks With Concurrent Access
JOINT WORK WITH: Gerard Hoekstra, Rob van der Mei and Bert Zwart
José Niño-Mora
A Restless Bandit Marginal Productivity Index for Opportunistic Spectrum Access with Sensing Errors
Philippe Robert
Stability Properties of Networks with Interacting TCP Flows
The asymptotic behavior of a Markovian model describing the interaction of several classes of permanent connections in a network is analyzed. For this model, each of the connections has a self-adaptive behavior in that its transmission rate along its route depends on the level of congestion of the nodes on its route. In this situation Graham and Robert have shown that the invariant distributions are in a one-to-one correspondence with the solutions of a fixed point equation in a finite dimensional space. The purpose of this paper is to investigate the problem of uniqueness of the equilibrium of these networks, i.e., the uniqueness of the solutions of the associated fixed point equation. Uniqueness results of such solutions are proved for different topologies: rings, trees and a linear network and with various configurations for routes through nodes.
JOINT WORK WITH: Carl Graham and Maaike Verloop
Devavrat Shah
Network algorithms made distributed
Resource allocation is inherent to any well engineered network. For example, resolving contention between packets accessing LAN, wireless medium or switch fabric in the router; sharing bandwidth between flows accessing the Internet; or multiplexing computational/memory resources in emerging data centers. Naturally, the algorithms allocating or scheduling resources crucially determine performance of the network. In a nutshell, to achieve efficient network utilization, such algorithms are required to solve a network-wide problem. However, in order for the network to be scalable, algorithms need to be distributed.
Thus, the challenge is that of achieving a global objective through local rules.
In this talk, I will discuss a new method for design of distributed algorithm for a generic network resource allocation problem. It builds upon insights from classical theories of reversible networks and variational principle as well as modern theory of dynamics of Markov chains. Method will be explained by means of two examples: packet-switched model for Medium Access Control in wireless networks and circuit-switched model for Optical Core Network design.
JOINT WORK WITH: Jinwoo Shin
Koen De Turck
Performance of the Sleep-Mode Mechanism of the New IEEE 802.16m Proposal for Correlated Downlink Traffic
JOINT WORK WITH: Stijn De Vuyst, Dieter Fiems, Sabine Wittevrongel and Herwig Bruneel
Gustavo de Veciana
User Association to Optimize Flow Level Performance in Wireless Systems with Dynamic Interference
JOINT WORK WITH: Balaji Rengarajan
Warren Volk-Makarewicz
Quantile Sensitivity Estimation
JOINT WORK WITH: Bernd Heidergott
Damon Wischik
Control of Multipath TCP and Optimization of Multipath Routing in the Internet
There are moves in the Internet architecture community to add multipath capabilities to TCP, so that end-systems will be able to shift their traffic away from congested parts of the network. We study two problems relating to the design of multipath TCP. (i) We investigate stochastic packet-level behaviour of some proposed multipath congestion control algorithms, and find that they do not behave how we might expect from fluid modeling: they tend to flap randomly between their available paths. We explain why, and propose a congestion control algorithm that does not flap. (ii) We consider how the path choice offered by the network affects the ability of end-systems to shift their traffic between a pool of resources. We define a ‘resource poolability’ metric, which measures for each resource how easy it is for traffic to be shifted away from that resource e.g. in the event of a traffic surge or link failure.
JOINT WORK WITH: Mark Handley and Costin Raiciu