Wireless Communication Unes in with Satellite Imagery
The appeal of wireless communications expands the demand for more detailed land cover maps.
By Kevin P. Corbley

The next time you complete a cellular or PCS telephone conversation that isn't disrupted by static or a sudden disconnection, you may owe your thanks to a remote sensing satellite. In addition, you can expect the quality of wireless calls to improve as new satellites with higher spatial resolution are launched.
     Telecommunications designers increasingly are turning to multispectral and panchromatic satellite imagery to map the land cover and terrain features which affect radio transmissions that carry wireless conversations. Designers are finding that as image resolution increases, so does the accuracy of their maps, which translates into clearer signals and fewer dropped calls for customers.
     Wireless engineers have been using satellite imagery for several years to design cellular coverage areas, relying on interpretation skills from value-added firms to provide ground clutter and topographic maps. Some vendors, however, wondered aloud if the cellular boom would end quickly for the remote sensing industry once all of the major metropolitan areas in the world were mapped.
     In fact, just the opposite is true. In the fiercely competitive telecommunications industry, companies have an insatiable appetite for more detailed maps that enable them to tweak their systems and upgrade their quality of service. Shifting population densities and changing land use patterns make this constant fine tuning a necessity. It is this same growth that has spawned the advent of new wireless technology and increased the demand for clutter maps.
     "Eventually, every city in the world that now has cellular coverage will redesign for the new personal communication system (PCS) technology," said Richard Wade, director of wireless communications technologies for Earth Information Systems (EISYS) Corp. of Austin, Texas, a provider of image-derived maps for wireless communication design. EISYS has been involved in wireless design projects in North and South America, Eastern Europe and throughout the Pacific Rim.
     The appeal of wireless communication has extended beyond the city. Developing countries that have never had widespread telephone service are forgoing the expense of stringing wires and erecting poles in favor of the more economical cellular service. And in North America and Europe, telecommunications companies are scrambling to provide uninterrupted service in rural regions between metropolitan areas.
     The coming of PCS will only expand the demand for more detailed land cover maps. PCS is a digital system that operates at a higher frequency than cellular, which means it has a shorter wavelength signal that can more easily be deflected by surface features, thus necessitating more accurate clutter mapping.

Clutter is Critical
Wireless communications designers refer to land use/land cover as clutter, which in this case is best defined as any obstacle or surface characteristic that can interfere with or block the radio frequency (RF) signal between the antenna tower site and the handset. Mapping clutter, therefore, is critical to cellular and PCS site design because both rely on RF transmissions to carry phone conversations.
     The impact of clutter depends on the composition and size of the obstructions. Different types of clutter interfere with signals in different ways. For example, forested land attenuates an RF signal, while open land and water allow it to travel great distances with minimal degradation. Many designers even request their maps to include differentiation of certain vegetation types and distinction of building densities in urban areas.
     There are several different RF design software packages on the market which utilize clutter maps. Some may require five clutter classes while others may use 20. The most common clutter categories requested by designers are urban, industrial, water, open field, forested, and vegetated. Forests and urban areas sometimes are broken into subclasses of high, medium and low densities.
     Although RF design software differs in data input, they all function on the same concept of signal propagation modeling. Each pixel in the image map is assigned a value based on the kind of clutter it represents. These values are used in an equation to calculate signal strength as it passes from one pixel to the next.
     When a potential tower site is located at a given point within the clutter map, a signal value is introduced at that pixel. As the signal passes through neighboring pixels, its strength degrades until it drops below a minimum threshold.
     Based on this information, the designer determines exactly where additional towers should be placed and how they should be oriented to ensure sufficient signal strength throughout the service area. These results often are tested in the field by placing a temporary tower at the chosen site and measuring signals at various locations and distances around it.
     Before wireless communications companies discovered the computer-based clutter interpretation and mapping services, they used pencils and rulers to interpret U.S. Geological Survey quad sheets manually.

Mapping Clutter with Satellites
More than half of all wireless site designers utilize ground clutter maps, and they prefer satellite imagery as their source of map data because imagery can be acquired quickly, economically and accurately anywhere in the world. And satellites offer the multispectral imagery necessary for generating most clutter maps.
     One telecommunications company (whose name is not used here to protect confidentiality agreements) tried using the traditional manual delineation method but eventually turned to EISYS to map its clutter data. Responsible for network planning in several major Texas cities, this wireless company needed detail in the dense urban areas. Designers at the wireless company quickly realized they were not obtaining the level of detail and accuracy from the manual ruler and pencil method required to adequately predict signal strength in the densely populated downtown area.
     The need for accurate mapping is crucial in urban terrain because that is where the population of wireless users is the most dense. Transmission towers in such areas must be located precisely and with just the right orientation and power level to provide proper coverage without interference.
     "Urban areas are the most difficult to classify in terms of detail," said Wade. "The company needed an accurate model of how the signal would travel within the urban canyon, down city streets and under freeway overpasses."
     For the Texas project, EISYS chose multispectral Landsat Thematic Mapper (TM) data to create its clutter maps. The value-added firm prefers TM data for clutter mapping because of its ready availability in the United States and its vast archive, which makes it a cost-effective alternative to other satellite data.
     After acquiring the appropriate scenes, EISYS ran an unsupervised classification on the data using a commercial image processing package. A series of supervised classification routines was then applied to the resulting land cover classes until all pixels had been classified into clutter categories.
     The classification routines were applied to different combinations of TM data depending on the topologic characteristics of a given area. For instance, visible green and near- and mid-infrared bands were used for identifying vegetation in the suburban areas. In clutter classifications of the urban centers where most features are not natural, visible red was substituted for visible green to differentiate buildings, parking lots and housing developments. EISYS technicians also interpreted imagery of the urban centers manually to ensure they would provide the spatial detail requested by the client.
     As a result of using the satellite-derived clutter maps, the wireless firm ultimately saved time and money. Construction of the cellular network was less costly than originally expected because the firm was able to install fewer antenna towers than the manual mapping method had predicted. In addition, time was saved in optimizing the system because there were fewer glitches to eliminate once the system was operational.

Panchromatic Helps in Urban Areas
Although other data sources were not needed for clutter mapping in the Texas project, EISYS often uses higher resolution, black-and-white panchromatic imagery for mapping in urban areas where surface features are smaller and more densely packed than outside a city. For example, a low-density suburb comprised of houses with yards will have much less effect on an RF signal than inner-city blocks of apartment buildings or central business districts with high-rise office towers.
     "High-resolution panchromatic imagery enables us to perform detailed clutter mapping visually in dense urban areas," said Wade.
     He explained that manual classification is preferred in mapping urban areas because the human eye is better at distinguishing many synthetic features. For instance, the tar roof of an office building has the same spectral reflectance as a tar parking lot, which would confuse a computerized classification.
     In extremely dense urban areas, EISYS employs either 10-meter resolution SPOT panchromatic imagery or black-and-white aerial photographs. Due to economic considerations, the firm prefers to use satellite images whenever possible. For these applications, EISYS is also evaluating imagery from the Indian IRS-1C satellite. Its panchromatic sensor acquires images at 5.8-meter spatial resolution.
     "IRS-1C panchromatic imagery gives us another level of mapping ability," said Wade. "We're really excited to get down to that level of detail."
     He added that panchromatic imagery often must be used during the refinement of a wireless site design, because once the site is up and running, unexpected interference and call drops occur. Higher-resolution imagery, such as panchromatic, is usually the solution to determining how the existing antenna can be reconfigured to eliminate the problem.

DEMs Filled in Other Variables
Clutter maps provided only some of the data required to fill in the RF signal propagation equation for the Texas cellular network design project. Elevation data is equally important because just as trees and buildings can interfere with a signal, so too can other landscape features such as hills, valleys, outcrops, mountains, and flood plain escarpments. More often than not, these topographic features do not just interfere with a signal but block it entirely.
     In addition to clutter maps, EISYS provided its Texas telecommunication clients with digital elevation models (DEMs) of the proposed wireless coverage area. Nearly all propagation modeling software packages accept inputs of DEM and clutter data to model signal strength.
     "Off-the-shelf DEMs usually weren't accurate enough for wireless applications so we extracted the DEMs ourselves from panchromatic image stereo pairs," said Wade. "In some cases, we have also derived DEMs from contour maps."
     In dense urban areas, accurate measurement of building heights has become as important as knowing the elevation of hills and mountains for wireless site design. While examining an IRS-1C panchromatic image, EISYS technicians noticed that building shadows are readily visible in the 5.8-meter data. They are now experimenting with the idea of applying photogrammetric processing routines to these shadows to extract building height data for future cellular design projects.
     "Essentially, we are trying to make digital elevation maps of the urban landscape using high-resolution satellite imagery," said Wade.

Higher Quality Maps, Better Connections
Land clutter and terrain maps derived from satellite imagery are giving many wireless companies the competitive edge they need to provide their customers with better quality service in the form of fewer dropped calls and clearer connections. For these wireless service providers, higher quality maps mean better service and happier customers, which translates directly into greater market share in an extremely competitive industry.
     For this reason, the telecommunications industry is awaiting the coming launches of higher resolution satellites with multispectral and panchromatic sensors, as well as stereo imaging capabilities. In preparation for these new data sets, the industry is refining its modeling software to take full advantage of the enhanced clutter and terrain detail that will soon be available.

About the Author:
Kevin Corbley is a freelance writer specializing in remote sensing, GIS and GPS. He is located in Denver and may be reached at 303-987-3979 or by e-mail at [email protected]

Back