Fig.1: The D2 network - a cellular network 
2. MICROCELL PLANNING 
Due to the growth of the subscribers of the D2 network, the 
resulting high traffic load and the use of low performance 
mobile phones, the network must continuously adjusted. 
One possibility to expand the capacity, that means more 
subscribers are able to set up calls at the same time, is the 
introduction of smaller cells with their own BTS. The 
capacity (Erlang/km?) is proportional to the number of cells 
within a specified area. 
However, the growing number of cells (and BTS) leads to a 
problem. Each BTS uses a specified frequency for the 
connection to and from the mobile. However, the number of 
available frequencies is limited, therefore each frequency 
must be used as often as possible. On the other hand the 
same frequencies sent from different BTS 's may not 
disturb each other to avoid interference problems and 
dropped calls. Therefore the knowledge of the topography 
is very important. 
Especially in the major cities the topographic data (100 m 
by 100 m pixelsize) which has been used up to now is no 
longer sufficient. In order to plan optimized coverage and 
capacity precise digital data of the terrain elevation, 
including the location and the height of the buildings (city 
structure data) is currently needed. The knowledge of city 
structures is a requirement for the planning of further 
antenna locations for the so-called microcells. The height 
above ground of these antennas will not project 
neighbouring buildings and the coverage performance of a 
microcell will be within a radius of about one kilometer and 
less. 
For modelling the urban areas, sophisticated fieldstrength 
798 
propagation models have been developed. Mannesmann 
Mobilfunk is using the so-called 3D-Urban-Micro-Model 
(Cichon et al., 1993) describing multipath wave propagation 
by three components (Vertical Plane, Transversal Plane 
Model and Multipath Scattering Model). These models 
calculate the propagation loss for over-roof-top 
propagation, propagation through street canyons and 
propagation via reflectiong walls of buildings. The 
computer-aided planning software bases on the use of 
precise raster data, i.e. building data. 
2.1 Demands on the data 
A 3D building dataset includes a geocoded height 
representation of a city surface. The height representation 
includes the terrain and also the buildings. The modelling 
of the terrain surface can be considered as sufficiently 
precise or can evaluated during the measurement. 
Therefore the greatest attention must be given to the 
generation of the buildings. The following main questions 
must be answered in advance: 
- Which areas are important (dense urban, urban, dense 
suburban, suburban) ? 
- in which horizontal accuracy the buildings must be 
evaluated and what pixel size is sufficient ? 
- in which vertical accuracy the buildings must be 
evaluated, in relation to the points of support and in 
relation to different heights within a building ? 
- how should irregular shaped buildings be handled ? 
- at which height above ground a building must be 
evaluated ? 
- at which size a building must be evaluated - in other 
words, are isolated buildings necessary or are rows of 
buildings sufficient ? 
- js it necessary to generate the different shapes of the 
roofs, gables, projections, front attributes, passages 
and if so, how can they be stored ? 
- js it necessary to also take into account precise 
information about single trees or line of trees, and how 
this information can be stored ? 
- which methods can be used for the generation, what 
are the advantages and disadvantages ? 
2.2 Methods 
For the generation of the microcell planning data several 
methods are available in theory. The methods must be 
considered with respect to: 
-  up-to-dateness 
- accuracy 
- rights 
- costs 
- controlling and updating 
- availability 
- degree of specification (detail of buildings) 
- homogeneity 
- reliability 
In addition the data must be generated or transformed in 
the same coordinate system as other data used at 
Mannesmann. This means Gauss-Kruger coordinates with 
Bessel ellipsoid and Potsdam datum, which must be 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996 
  
  
— A A emmy e I^ PN $e NPN 
(8 
na 
e o9 (mS. 050 Lo e r3 00 0 rn 
nS TP - T - 
-—— gu 
CY Cy: g) '-€ 'r-—:--T- ge OB 
— c 
a ue 
DO