Title :
Large bursts do not cause instability
Author_Institution :
Coordinated Sci. Lab., Illinois Univ., Urbana, IL, USA
fDate :
1/1/2000 12:00:00 AM
Abstract :
Fluid models of queuing networks are among the simplest models to analyze, owing to the fact that calculus can be applied. At the same time, wider classes of network models are more flexible for modeling real traffic. It is thus useful to reduce questions about the more realistic models to questions about related fluid models. Such a reduction was achieved by J.G. Dai (1995), who showed that stability of a fluid model implies stability (in the sense of Harris recurrence) of related multiclass networks with random service and interarrival processes of renewal type. The purpose of this paper is to similarly reduce the question of stability for networks with input traffic satisfying deterministic constraints in the sense of R.I. Cruz (1991) to a question of stability for a fluid model. It is shown that the stability of networks with fluid traffic implies stability of networks with deterministically constrained traffic
Keywords :
fluid dynamics; queueing theory; stability; Harris recurrence; deterministic constraints; deterministically constrained traffic; fluid models; interarrival processes; multiclass networks; network models; networks with fluid traffic; queuing networks; stability; Automatic control; Bonding; Control systems; Control theory; Friction; Robotics and automation; Servomechanisms; Velocity control; Welding;
Journal_Title :
Automatic Control, IEEE Transactions on