This paper studies the quality of service (QoS) provision problem in noncoo
perative networks where applications or users are selfish and routers imple
ment generalized processor sharing based packet scheduling. We formulate a
model of QoS provision in noncooperative networks where users are given the
freedom to choose both the service classes and traffic volume allocated, a
nd heterogenous QoS preferences are captured by a user's utility function.
We present a comprehensive analysis of the noncooperative multi-class QoS p
rovision game, giving a complete characterization of Nash equilibria and th
eir existence criteria, and show under what conditions they are Pareto- and
system-optimal. We show that, in general, Nash equilibria need not exist,
and when they do exist, they need not be Pareto- nor system-optimal. For ce
rtain "resource-plentiful" systems, however, we show that the world indeed
can be nice with Nash equilibria, Pareto optima, and system optima collapsi
ng into a single class. We study the problem of facilitating effective QoS
in systems with multi-dimensional QoS vectors containing both mean- and bur
stiness-related QoS measures. We extend the game-theoretic analysis to mult
i-dimensional QoS vector games and show under what conditions the aforement
ioned results carry over. (C) 2000 Elsevier Science B.V. All rights reserve
d.