Abstract

Discrete event dynamical systems (DEDS) are a class of man-made systems, which are driven by a set of discrete events. Typical examples include queuing systems, manufacturing systems, telecommunication networks, and so on. In the past two decades, the modeling, simulation, and optimization of DEDS have received considerable attention. The objective of this thesis is to apply the DEDS approaches to study wireless personal communication systems. Two applications are studied: the call request buffering problem and the mobile terminal location tracking problem. In a personal communication service network, a set of channels is assigned to every cell. When a phone call arrives and is assigned a channel, it consumes the channel until the end of the conversation. If no channel is available, the call is dropped/blocked. In many cases, some channels may become available shortly after a call is dropped. Thus, if some buffering mechanism is used, a cell may accommodate more phone calls and thus the channel utilization is increased. This is referred to as call request buffering. The first part of this thesis develops an algorithm to estimate the sensitivity of a Markov process using a single sample path of its uniformization Markov chain, and uses a gradient algorithm to find a locally optimal parameter. After that, a random call request buffering scheme is introduced, which is optimized using the proposed algorithm. In a wireless personal communication service system, in order to deliver a phone call, the network has to track the mobile terminals' locations from time to time. The second part of the thesis proposes a mobile terminal location update model, whose states consist of two components--the time elapsed since last call arrived and the distance the mobile terminal has traveled since last registration.