A class of tools for analyzing and managing these sets of data is constituted by GIS's, which have been developed to support planning and management of the territory; as a consequence, they can be used profitably in the design of cellular systems, where the spatial components play a key role. Furthermore the possibility of customizing such class of tools in an integrated system allows the designer to perform a quick network planning, giving him the opportunity of having a quasi real time understanding of the current situation and updating the planned structure in a very easy way, as to radioelectrical coverages, frequency plans, channel assignments and interference evaluations.
In an urban area, for example, the more important information to be considered, without forgetting the orography characteristics, is the mobile traffic distribution. This establishes the dimension of the mobile radio structures, i.e. number and size of the cells that implement the radio electrical coverage, achieving the goal of providing a given service quality making maximum use of the available resources.
In rural areas on the contrary, the location of sites, where the base stations have to be installed, establishes another planning methodology: the designer must provide the position and the height for the antennae to guarantee a good radio-electrical coverage.
The main steps of the planning process of mobile radio coverages will be analyzed along with the geographical data base used giving a major emphasis to the electromagnetic propagation aspects that are realized through the RASPUTIN system.
The data that have to be considered to generate such a matrix are usually obtained from a list of parameters:
Each information has to be processed and expressed into a matrix form with the same characteristics of the resultant traffic density matrix, i. e. the values must be referred to the same elementary areas, with the same size and position, and georeferenced in the same geographic coordinates system.
By combining the matrices with proper weights, the resultant matrix of the mobile traffic density can be obtained. It must be noted that the above list of data is not exhaustive, and that data may have different levels of importance. In several cases some types of information are difficult to obtain. As a rule of thumb, however, the more the factors considered in a real situation, the better will be the traffic density estimation.
From an operational point of view each intermediate matrix can be generated through a set of procedures that exploiting GIS capabilities allow to introduce and manipulate information coming from cartography, remote sensing techniques, aerial photography and so on.
For example the built-up density matrix can be obtained from the digitization of the buildings of local authority maps and introduced into ArcInfo as a polygonal coverage. In a second step the function of gridding can produce the built-up matrix with a certain cell size.
In order to provide an estimate of the vehicular mobile traffic a map of the road network can be digitized and introduced as a vector data coverage. If each element is also characterized by the type and the traffic density of the road it belongs to (motorway, main road, secondary road, etc.) an operation (after a partitioning of a town area into a grid of square elements) can sum up the weighted line lengths inside each cell and generate the corresponding vehicular traffic matrix.
The resultant mobile traffic density matrix is then generated combining with proper weights all the various traffic density matrices.
In an urban environment the matrix of mobile radio traffic density is used for selecting the theoretically most suitable sites for dimensioning the corresponding mobile cells. The final result has to ensure the coverage of the whole urban areas with an efficient grade of service. This goal can be reached quickly and with a good accuracy using automatic tools developed in ArcInfo that exploiting GIS capabilities [3] can be used to select potential site for antennae and to determine their coverage radii necessary to cope with the user demand. In fact it is enough to split the urban territory into a grid, where each element represents a potential site for a base station to which a value is associated representing the size of the corresponding mobile cell (got from the matrix of the traffic density); at this point an interactive selection of an antenna site allows to have the corresponding coverage radius displayed on the screen together a cartography map of the town as a background for an easy recognition of the work areas. At the end of this operation all the antennae will be correctly dimensioned in term of site and input transmission power.
In a rural environment the site selection for base stations is obtained first of all specifying the mobile users characteristics distributed on the territory, which can be represented, in this case, in terms of vehicular traffic (motor-ways, main roads, etc.) and villages that have to be covered. While in urban area antenna sites and cell dimensioning are mainly evaluated on the basis of mobile users traffic, in rural areas coverages are predicted taking into account mainly the orography and land usage information: the selection is made pointing out those sites with ensure a good signal propagation. As a consequence, intervisibility analyses (specifying the position of the viewer, the horizontal and vertical angles of view, the viewing direction, and the maximum viewable line-of-sight distance) and 3-dimensional functions, both operated by means of digitized terrain data, are very useful tools to support this part of the planning activities.
The planning phase defines how many base stations have to be activated to satisfy mobile users demand and where they have to be approximately installed to guarantee an adequate electromagnetic coverage but it says nothing about the real characteristics of the antennae and the exact location of installing. It can happen in a real situation that first candidates to host base stations could be not available because, for example, that particular building is not owned by the mobile operator or the rent is not allowed or, in rural environment, optimal site to activate antennae could be not reachable because there are no roads or electric power facilities.
Moreover, it is necessary to check whether the coverage structure, obtained from the phase of planning, is actually confirmed from an electromagnetic propagation stand point, for each cell, according to the environmental characteristics. In fact field-strength computations depend on distance, base station effective antenna height, terrain irregularities and environmental situation (urban, suburban, open, hill areas etc.) [4].
In our case the simulation of radioelectrical area coverages is achieved using RASPUTIN, that is a software tool used to predict radioelectrical field strength.
RASPUTIN stand-alone is a code that was implemented as a rather simple and fast mode software planning tool that can work in an rather general and simple software environment, as it has been developed in both "C" and "FORTRAN77" program languages. Also the input territorial information, namely terrain height, building density and vegetation, is structured in an easy to manage way (a straightforward set of files), without using any particular or sophisticated commercial data base.
RASPUTIN in ArcInfo, on the other hand, is a planning system that allows the user (by means of a sequential set of menus) to properly design a whole mobile radio network in an interactive way. Such an operation is performed through a user friendly interface, which gives the planner the opportunity to have a continuous menu driven interaction with the system, associated with advanced graphical display characteristics of the results, merged with the territorial images managed by the GIS.
In order to generate this kind of information a class of tools for preparing, analyzing and managing altimetry data has been developed in ArcInfo to provide the data that have to be used by the propagation algorithm.
In particular, as an example, a system in ArcInfo that uses a set of digitized maps in form of contour lines and elevation points, obtained from cartography maps at different scale (1:250,000, 1:50,000 and 1:10,000), has been developed for the mobile network design for Greece, country in which the Italian public telecommunications holding is one of the two licensed private GSM operators. This system generates both a set of regular matrices (ASCII files) of altimetric values with an appropriate grid size of 230x230 m containing the average value of the terrain heights in each specific location (used by RASPUTIN stand-alone) and a georeferenced final altimetric grid with the same characteristics of resolution (used by RASPUTIN in ArcInfo) [5].
This particular grid resolution is derived from the consideration that it represents a good compromise in term of mass storage and the precision needed by the propagation algorithms operating at frequencies between 450 and 900 MHz.
As far as the built-up and vegetation area density information the method of generating this particular data base for RASPUTIN consists in producing final grids and a set of files with the same resolution of the corresponding altimetry ones and whose elements identify respectively the percentage of the built-up area and vegetation on the territory.
As a last consideration the altimetry final grid is also used to produce shaded relief images that are used as a background by the user interface of RASPUTIN in order to give the planner the opportunity of having an immediate understanding of the situation on territory, when performing the network planning.
Regarding the land usage treatment, two main factors may be identified: building density and vegetation effects, and their influence has been included in the smooth earth propagation component. The effects due to local urbanization around the mobile unit can be usefully related to the surface parameter "building density" defined as the percentage of area covered by buildings, with reference to a given standard grid size (230x230 m) as it was said above. Such a parameter usually ranges from very low values (<5%) for open (rural) zones, to values above about 60-70% for very densely built-up areas (historical town centers). This continuous-type urbanization approach is of great value, in that provides a quantitative identification of the type of the area under examination, which cannot be achieved when a more or less dense discrete land cover classification is adopted [8]. Furthermore, RASPUTIN evaluates also the effects of vegetation on the signal propagation starting from a suitable knowledge of the vegetated areas distribution.
The shadowing effects due to orographic obstacles are taken into account by a more complex propagation model, which requires a strict interaction with the territorial data base. Starting from a base station site, the algorithm spans the surrounding area along radial directions with a suitable angular step (typically, 0.5 degrees). For each direction, the data base interface provides the radial terrain height profile along which the field strength computation should be performed. It is clear that the use of a detailed territorial data base may be responsible on one hand of operational difficulties due to the large amount of data to be handled, but, on the other hand, ensures more accurate coverage predictions.
RASPUTIN is able to account for the diffraction effects by using a prediction model based on the Huyghens-Fresnel diffraction theory. Accordingly, a single obstacle is assumed to be a perfectly absorbing half-plain screen ("knife-edge" approach), whose diffraction effects can be easily calculated [9], as long as the signal wavelength is negligible with respect to obstacle size and distances from both transmit and receive sites. The approximation consists in supposing the obstacle having no thickness along the propagation direction and being infinitely extended in the orthogonal section, respectively, and neglecting its real electromagnetic properties. Obviously, this approach has been extended to the case of multiple knife-edge obstacles in tandem, by means of a proper combinations of single obstacle contributions. In more details, RASPUTIN can work using two different calculations methods: the Epstein-Peterson method [10] and the Deygout method [11]. The first one is more suitable for situations where the various obstacles are responsible of more or less similar effects, while the second one gives better results when a main obstacle exists, whose effect is predominant with respect to the remaining ones. Nevertheless, irrespective of the method, a crucial aspect of the problem is the obstacle identification technique, starting from the terrain profile along the path between base station and mobile terminal. RASPUTIN works by means of a proprietary selection technique developed in CSELT for the synthesis of a generic terrain height profile [12]. Comparisons with several measurement results have confirmed the reliability of the above synthesis algorithm.
The above features allow the user to have, at each step of the planning process, a real-time feedback about what the current situation can provide, how the same situation might evolve when introducing new RBS(s) and to what extent such new RBS(s) might interact with the rest of the already existing network, in term of global performance improvements (or impairments). Specifically, exploiting GIS facilities, differences between "old" and "new" network shape can be pointed out in a straightforward manner, in term of additional coverage and modifications in the relative or global "best server" configurations. Furthermore, the use of a GIS facilitates either the implementation of new functionality, in a software evolution perspective, or the integration of the current features with other external potentially useful tools.
The RASPUTIN in ArcInfo software can be roughly described as it follows. The starting point (1) is the selection from the data base of the last configuration of the network (which may be eventually a "blank" one), namely the result of the latest planning activity in the area under consideration; this allows to proceed with the development of the network (increase of coverage, cell splitting, "best server" improvement) with the introduction of new RBS(s), and facilitates also the identification of the new site(s), when needed.
At this stage, a proper choice of a geographical working area is required (2), in order to extract from the database just those data corresponding to a region within a square of approximately 200x200 km around the new entry, in which the simulations have to be carried out. Such a selection, which is performed by simply dragging the mouse and properly clicking at the region corners, allows to save computing time, forcing the algorithm to operate on a limited portion of data only (and neglecting the rest of the territory). An example of selection of a working area for a given region of Greece and the relative "old" network configuration (in terms of global coverage) are reported, which can be used as a starting point for the subsequent cellular planning phases. The four different colours correspond to four different field strength levels: 20,30,40,50 dB[microV/m], respectively.
Further to the previous steps, RASPUTIN is allowed to run the simulation algorithm, to produce the area coverage corresponding to a new RBS, after introducing the base station geometrical features and the relevant antenna radiation pattern characteristics (both in azimuth and elevation), as selected from an ad-hoc menu. (3). However, as the electromagnetic simulation algorithm may be quite time consuming (approximately 5 to 10 HP 750 CPU minutes for an omnidirectional antenna, for distances up to 40 km from the base station, with an angular step of 1 degree), a quicker (although rougher) area coverage estimation may be alternatively obtained by a simple intervisibility (optical) analysis (4) (this computation takes only about half a minute). In most cases, such an alternative procedure is enough to decide whether the new RBS would provide a real improvement or not. On the basis of the above results, the users can then go back and restart the computations after properly changing the RBS geometrical and/or radioelectrical parameters, or continue in the simulation with the full electromagnetic RASPUTIN algorithm for an actual field strength prediction (5). An example of the single radioelectrical area coverage generated by a new RBS with the associated 2-D background image (altimetry), located at the center of a 35 km radius circle (maximum operation range for a GSM system) is reported (the colours have the same meaning as explained above).
At this point, the users is given the possibility to display and analyze the results obtained so far, using a visualization menu; the available on-screen outputs are as follows: (6)
As a typical output example, the new global radioelectrical coverage (inclusive of the new RBS) is illustrated and as such should be compared with the "old" network configuration. In order to stress the capabilities offered by ArcInfo in managing a territorial database, the same coverage is presented superimposed to a 3-D representation of the same region. Additional information about the performance of the new configuration can then be obtained; the "relative" best server after the introduction of the new RBS is reported, in that it points locations where the new RBS acts as "best server" in comparison with the all pre-existing network considered as a whole (the global "best server" can also be represented using different colours for a given field strength level according to the RBS which provides in each element of the area under examination the highest field strength value).
The final stage is the acceptance or rejection by the user of the new predicted configuration performance. In case of acceptance (7) the new configuration is saved, updating the original data base with the current new RBS; in other case (rejection, (8)), RASPUTIN allows the users to go back to stage (3), in order to change the base station geometrical and/or radioelectrical characteristics (location, height, antenna pattern and orientation, etc.) and start a new simulation cycle.
At each step of the entire planning work, the users is given the possibility to query the ArcInfo data base through a set of menus in order to obtain information about the characteristics of the coverage structure (for example the value of the field strength on the terrain of the "new" total coverage, by simply clicking the mouse) or the altimetry data base. This is very useful to take decision when it is necessary to accept or not the results of RASPUTIN, not only in term of coverage but also considering the field strength values. As far as the altimetry data base is concerned, the information of the height values of a specific area on the ground, obtained again by simply clicking the mouse, can drive the planner to select adequate prospective sites. At the same time, a customized menu gives the possibility to use or modify the cartography background on the screen (motorways, main roads, railways, rivers, lakes, etc.) according to the specific information required by the designer (this facility can be useful, for example, to have an immediate information whether or not a specific road is covered by the mobile service).
On the basis of the gained experience, it appears that GIS's are unquestionably fundamental tools for a proper development of radio activities that make a large use of territorial data bases. Furthermore, the intrinsic flexibility in managing and processing geographic information, associated with the possibility of generating ad hoc user interfaces represent an optimum reason for using GIS's in mobile network planning activities.
Obviously GIS's deal with general aspects of the territory, as they have not been developed to be specifically applied to problems related to mobile radio systems. Hence, programs exploiting the GIS's capabilities must be adapted to planner requirements, tailored to respond to his specific problems.
[2]..C. Scarati, B. Lingua, G. Minerva, G. Calabrini, "Radio coverage for cellular networks planning: a methodological approach", in CSELT Technical Reports (Vol.XXI, No.5, December 1993)
[3]..G. Bussolino, "Use of geographic information systems and spatial data bases for mobile radio applications", in CSELT Technical Reports (Vol.XXI, No.5, December 1993)
[4]..E. Damosso, "The jump from 450 to 900 MHz: experience and future investigations in land mobile propagation", in Proc. of Int. Symp. in Antennas & Propagation, Kyoto, (J), 1985
[5] G. Bussolino, E. Gilio, E. Lo Gatto, B. Missan, "Generation of an altimetric data base using a Geographic Information System" in Proc. of IEEE 2nd Symp. on Communications and Vehicular Technology in the Benelux, Louvain-la-Neuve (B), November 1994
[6] Y. Okumura, et alii, "Field Strength and its variability in VHF and UHF land mobile radio service", Review of the ECL, Vol.16, 1968
[7] ITU-R P.529-2, "Prediction methods for the terrestrial land mobile service in the VHF and UHF bands", Doc.3/1014, Geneva, 1995
[8] E. Gurdenli, P. Huish, L.E. King, M.J. Nicholls, "Land usage factors in mobile radio propagation prediction", in Proc. 4th Int. Conf. on Land mobile radio, Warwick (UK), 1987
[9] D.C. Livingston, "The physics of microwave propagation", Prentice-Hall,Inc.,Englewood Cliffs, New Jersey, 1970
[10] J. Epstein D. and W. Peterson, "An experimental study of wave propagation at 850 Mc/s", in Proc. IRE, Vol.41,No.5,May 1953
[11] J. Deygout, "Multiple knife-edge diffraction of microwaves", IEEE Trans., Vol.AP-14,No.4, 1966
[12] G. Dellagiacoma, G. De Renzis, L. Stola, "Terrain height profile synthesis for mobile area coverage applications", in CSELT Technical Reports (Vol.XXI, No.5, December 1993)
