TD(98)005

Abstract
In the radio network planning process a proper description of cell shapes is important for many  tasks, e. g. interference analysis, hand-over planning or traffic calculations. The corresponding cell calculation has impact on network quality. In this paper the performance of a statistical cell calculation method to compute cell assignment probabilities (CAP) based on Markov-processes in a GSM1800 network is analysed. The CAP are computed from transition (or  hand-over) probabilities, which are derived from path loss predictions and a statistical description of the hand-over process. In order to enable a reasonable calculation time a number of simplifications and approximations are necessary. The method is compared to the traditional best server method and verified by test-mobile measurements.  Although a large number of simplifying assumptions were used in the CAP-model a reasonable agreement with measurements was achieved. The impact of the propagation model, the fading parameter and of mobility aspects have been investigated. It is shown that the cell description by a prediction-based cell-assignment probability is more accurate than the traditional best server method.