An Efficient, Extensible Design
of a PCS Network Simulation

John T. Douglass, Devidas A. Gupta,
Brian A. Malloy and David A. Sykes
[email protected]
Department of Computer Science
Clemson University
Clemson, SC 29634

1 Introduction

This decade has witnessed a proliferation of personal communication systems where cellular phones provide accessibility to people in transit. In a personal communication system (PCS), the geographic location being serviced is partitioned into cells with each cell containing a broadcast station. Each station is assigned a set of channels used to service calls from cellular phones located in the cell. A given channel can be used concurrently in geographically distinct cells if and only if the cells are sufficiently far apart. This geographic separation guarantees that the level of interference is low for calls assigned to the same channel. When a PCS user crosses a cell boundary, the call must be transferred (handoff) to the station in the new cell. That station must assign a channel to the call1. If the station in the new cell does not have an available channel, then the call is blocked. A call is also blocked if a call arrives for a PCS user located in a cell and the corresponding station does not have a channel available for the call. To provide adequate service, the ratio of blocked calls to the number of attempted calls and attempted handoffs must remain below one percent[1].

A technique for gauging PCS performance is to simulate the actions of the cellular phones for the geographic location under study. However, there are two important problems associated with PCS simulation. The first problem is that PCS simulation is computationally intensive and demanding of the memory hierarchy so that many simulations contain fewer than 50 cells in the network[6, 13]. However, small-scale networks can induce inaccuracies into the simulation due to the effect of handoffs occurring at the boundary 1There are two major channel allocation schemes. In fixed channel allocation, a set of channels is assigned to the cell and only calls located in the cell may use those channels. In dynamic channel allocation, the channel being used by a call may actually be located in a nearby cell.