In communication systems, all service control nodes, such as web sites or Mobile Switching Centers, can be modeled as a server system with one or more servers processing the incoming requests. To avoid overload at the service node some type of admission control mechanism is usually implemented to guarantee good performance also during high traffic loads.

Nonlinear discrete-time modeling of a web server system is investigated. Server systems typically contain non-linearities such as saturations and bounded queue lengths. The incoming traffic and service rates are best modeled by stochastic processes, well described and analyzed by queuing theory.

In this seminar we investigate, from a control point of view, the nonlinear discrete-time modeling of a server system, and the analysis and design of load control mechanisms based upon this model. Verification of the server model behavior with respect to queue theoretic models are made and the different load control mechanisms are implemented on an Apache web server and experimentally evaluated.