Network QoS games: stability vs optimality tradeoff

  • Authors:
  • Andrew Lomonosov;Meera Sitharam;Kihong Park

  • Affiliations:
  • CISE, University of Florida, Gainesville, FL;CISE, University of Florida, Gainesville, FL;Department of Computer Sciences, Purdue University, West Lafayette, IN

  • Venue:
  • Journal of Computer and System Sciences
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

We study noncooperative games whose players are selfish, distributed users of a network and the game's broad objective is to optimize Quality of Service (QoS) provision. Our classes of games are based on realistic microeconomic market models of QoS provision (Proceedings of the First International Conference on Information and Computation Economics ICE'98, 1998) and have two competing characteristics--stability and optimality. Stability refers to whether the game reaches a Nash equilibrium. Optimality is a measure of how close a Nash equilibrium is to optimizing a given objective function defined on game configuration. The overall goal is to determine a minimal set of static game rules based on pricing that result in stable and efficient QoS provision. We give a new and general technique to establish stability and demonstrate a close trade-off between stability and optimality for our game classes. We also state several open problems and directions together with initial observations and conjectures.