The Adaptive Base Station Positioning Algorithm (ABPA) is presented, which is based on a neural net approximation of the tra c density in the coverage area of a cellular mobile communication system. ABPA employs simulated annealing, thereby achieving quasi-optimal base station locations depending on the topography of the investigated area. Furthermore, ABPA considers the radio wave propagation within this area for the base station positions. Therefore, a three dimensional digital surface model is used to approximate the topography and two eld strength prediction methods, a line-of-sight (LOS) approach and a ray-tracing technique, are investigated within the context of adaptive positioning. In particular, the results obtained by the ray-tracing technique are encouraging, showing supplying areas, which seem to be similar to those, stemming from real measurements. However, as simulations show, the more realistic eld strength prediction by ray-tracing has a strong in uence on the performance of ABPA, resulting in di erent base station locations. As an outlook, the combination of eld strength prediction using ray-tracing with adaptive tra c prediction on a road graph by neural nets is proposed for further investigation.
[1]
M. Hata,et al.
Empirical formula for propagation loss in land mobile radio services
,
1980,
IEEE Transactions on Vehicular Technology.
[2]
Theodore S. Rappaport,et al.
Propagation measurements and models for wireless communications channels
,
1995,
IEEE Commun. Mag..
[3]
Turner Whitted,et al.
An improved illumination model for shaded display
,
1979,
CACM.
[4]
E. V. A..
Lehrbuch der Hochfrequenztechnik
,
1937,
Nature.
[5]
R.L. Hamilton,et al.
Ray tracing as a design tool for radio networks
,
1991,
IEEE Network.
[6]
Thomas Kürner,et al.
Concepts and Results for 3D Digital Terrain-Based Wave Propagation Models: An Overview
,
1993,
IEEE J. Sel. Areas Commun..
[7]
Thomas Martinetz,et al.
Topology representing networks
,
1994,
Neural Networks.
[8]
T. Fritsch,et al.
An integrated approach to cellular mobile communication planning using traffic data prestructured by a self-organizing feature map
,
1993,
IEEE International Conference on Neural Networks.