Heavy traffic models and control of wireless systems

Thursday, June 30, 2005 - 4:00pm - 5:00pm
EE/CS 3-180
Robert Buche (North Carolina State University)
Heavy traffic methods are useful for wireline queueing analysis and show promise for wireless systems. The main difference compared to the wireline setting is the randomly-varying environment in which wireless systems operate. In the heavy traffic analysis, typically a Brownian driven stochastic differential equation with reflection (SDER) models the queueing dynamics. We will discuss these models and the associated stochastic control problem along with some numerical simulation results. Recently, the importance of modeling long range dependence in wireless systems has been shown. We will give some initial comments on extending the analysis to this case where the SDER is driven by a more general Levy process.