Cellular telephone network planners are facing problems with topographic maps that are not updated often enough to reflect the rapid vegetation land cover changes and urban growth taking place in many expansive markets. In countries where some of the most rapid expansion is taking place, suitable base maps can be outdated, available only in small scale, or difficult to get access to. Insufficient or outdated geographic data may lead to sub-standard radio telephone networks. Up-to-date terrain information from a single, accessible source means savings in time, and allows the network planners to concentrate on optimizing their networks in an increasingly competitive market.

Radio Wave Propagation
Air, land cover, terrain heights and buildings all provide obstacles affecting radio wave signal strength. Compensation for the influences of these factors can be based on a statistical model, or - in more detail - on databases with information on actual heights and land cover.
      Computer analysis programs are now capable of processing databases taking into consideration categories of land cover and obstacle heights to determine radio signal strength at each location. Some of the leading software are MSI Planet (adapted for Ericsson's planning it is named EET), Tornado by Siemens and NPS/X (Nokia). The systems are continuously refined, meaning higher demands on input data to produce desired results.
      "The quality of the network plan depends on the network designer's know-how and the tools that he is using," according to Antti Rahikainen, department manager, planning services for Nokia's customer service center for South East Asia in Bangkok, Thailand. With the help of the planning tool (NPS/X), Rahikainen is able to produce coverage prediction, frequency allocation, and optimization of parameters controlling the dynamic operation of the network. "The base for all the planning is the coverage prediction which depends on radio-wave propagation and the digital map," he says. "If the digital map is incorrect or old the coverage prediction is also not correct and the plan is not the best possible."

Geographic Information From Satellites
With the capability to collect geographic information over practically all parts of the globe using Earth observation satellites, this imagery can be a useful source of information for telecom network planners. Rahikainen notes, "The quality of the network plan is dependent on the quality of the digital map. Satellite imagery as an input for the digital map is very good...."
      Rahikainen further states he likes the use of satellite images because, typically, they are recent photographs and their accuracy is sufficient for planning networks in urban areas. He says different types of satellite imagery are available for both rural and urban planning which require different resolutions from the digital map.
      For best results satellite images should meet the following specifications:
• The images used should be the most recent images available.
• The dates of the images should differ as little as possible.
• The images should contain as few clouds as possible.
• The sensor selected should meet the demand for map scale.
• The sensor radiometric properties must match ground features to be detected.

Satellite Image Maps
In order to produce map products of high quality from satellite images, the data must be processed. The satellite data in digital form is a very dynamic tool, but the output quality depends on the methodology used and the skills of the operators. By applying correct algorithms, information important for interpretation is emphasized.
      SSC Satellitbild, a subsidiary of the Swedish Space Corp., has specialized in producing value-added and specialized satellite image products, such as its Satellite Image Maps (SIMs). The SIMs satellite imagery is produced in a map sheet format. Satellite scenes are, after precision correcting, merged digitally to cover the requested area and subsectioned into map sheet units. Ground control points, required for precision correctioning, can most often be derived from topo maps. GPS points can also be used. The subsectioning and scale are determined by the interpretation requirements. In areas above sea level, ortho correction may be necessary to avoid spatial shifts due to parallax errors. DTMs-derived from digitized point heights or curves in topo maps or from SPOT stereo pairs are used.
      Input data are selected based on the type of network planning situation, acquisition dates and archive availability. For detailed urban applications it is advantageous to use Sensor Composite Satellite Image Maps: SPOT P merged with multispectral SPOT or TM data.
      The land use map is a perfect basis for an inventory. In this way the current state of resources can be defined. If further studies are necessary with greater detail it can be used as a means of stratification. The primary reasons for mapping land use classes are to provide information which can be utilized for analysis, planning, monitoring and decision making.
      The necessity of land use mapping is manifold. In a recent project it would be the foundation for the environmental monitoring. In discussions with the client it became necessary to establish a system for land use classification with the pertinent classes for environmental monitoring.
      Experience has shown that the acquisition of imagery, both using aircraft and satellites, which meets cloud cover specifications for mapping may be difficult in parts of a project area. With satellite imagery it is, however, possible to combine scenes on which different parts are cloud-free to produce a cloud-free result. This is not possible with aerial photographs.
      For instance, if the project area is located along the coast of Mozambique, the climatic conditions are ideal for receiving SPOT or Landsat data. If necessary, data will be received continuously as the mapping is carried out in order to obtain the most recent imagery as a base for the production.

Terrain Types
Based on the firm's experience in fields such as forestry and environmental studies where a number of projects involving interpretation of imagery and detailed classification of land cover have been carried out, SSC Satellitbild has developed special products for cellular network planning.
      Satellite Image Maps are, together with other available sources of geographic information, used to build a database with clutter data - classes of land cover. Road information and other linear features are supplied in separate files. Other input materials are used as supplements.
      For large areas with homogeneous input material, computerized interpretation methods are effective. This situation is however not common in production for telecom. Short production times necessitate innovative use of available information sources, and skilled combinations of visual interpretation and interactive computerized interpretation methods. The human mind is superior to machines in taking into account: tone, color, texture, pattern, shape, size, shadow and context, especially in heterogeneous data. The combination of visual descriptions and automatic classification, resulting in categorization of surfaces into meaningful groups, are hence superior in accuracy in comparison to machine dependent classification only.
      Common interpreted terrain type classes are:
• Water-Open land-Bush / plantation / quasi open
• Forest-Park-Suburban
• Urban-Dense urban

Terrain Heights
Digital elevation models are produced in two different ways: Scanned or digitized from topographic maps, or processed from SPOT satellite image stereo pairs.

Three levels: City, Region, Country
The scope for detailed cell planning is different than the scope for overview planning. Types of image data used, pixel size, scale of presentation and number of classes should be adapted to actual planning situation. Three interpretation situations are suggested:
      The terrain classifications, linear information and height data are converted into a format compatible with the client's network planning system.

Experiences
SSC Satellitbild has cooperated with suppliers and network operators since 1991, when an extensive digital mapping project for wave propagation analysis was carried out for the Swedish National Telecom Agency and the Defence Material Administration. Using 46 Landsat TM scenes and 12 SPOT scenes, 900 digital maps with land cover in 13 classes were produced. A computerized classification with 50m pixels, later also presented in 25m pixels, was used. Digital height data were added to the files.

Cable network planning
Detailed infrastructure information from geometrically correct Satellite Image Maps is also useful for another telecom planning activity: cable network design. Requirements on scale and accuracy are high, and also here the planners are often faced with inadequate local paper maps as their only input besides surveys on the ground.
      The varied landscapes of most of the western Malaysian peninsula, from coastal plains up to the mountains, have been mapped during the first part of 1995. SPOT satellite imagery, supplemented by additional information, were used to produce terrain type, terrain height and road data. With the experience gained on these projects, SSC Satellitbild is contracting for work on supplying the necessary data for cellular network planning projects in other locations in east Asia, south Asia and South America.

About the Author:
Bjorn Ohlson is a sales manager for Spot Asia in Singapore. He may be reached at +46 980 121 40.

Back