Foliage Spoilage

If a call drops in the woods, does anyone hear it? The answer is more technical than philosophical. Foliage can attenuate signals as much as the metallic window tinting that's common in modern office buildings.

"It's a concern any place where there are fairly dense trees of significant height: greater than about 10 feet to 15 feet," said Harry Anderson, EDX Engineering president & CEO.

Take pine and fir trees, which can be a particular nuisance.

"The needles can create special resonance problems when they are close to an even fractional wavelength (such as) 1/4 or 1/2 of the system's signal," Anderson said.

How great a problem can vary.

"Based on our experience, it's pretty much the same as regular foliage," said Munira Jaffar, LCC senior manager of engineering. "If it's different, the losses probably are higher by 1dB to 2dB. For sites serving areas with a lot of pine trees, maybe you have to increase the transmit power slightly."

In some cases, nudging up the power might not be enough.

"I've been through some areas where pine trees shadowed a roadway," said Roger Reeves, GTE Wireless manager of radio and network quality, Midwest. "We lost a great deal of signal when we drove behind that type of tree: probably 20dB to 30dB."

One solution is to deploy a repeater, particularly if the area is large and the attenuation is significant. If the trees shadow a road, and if the siting allows it, the solution might be to reorient the antennas so that the signal runs straight down the road. If the attenuation is minor, the best option might be simply to leave it alone.

"As you grow, you're going to add sites for coverage and capacity," Reeves said. "A lot of times, you'll address that concern with the natural growth of the network."

If that foliage is lost each fall, the sites serving that area might need to be optimized or even powered down slightly to ensure that the bare trees don't allow the signal to travel farther than it should and create interference elsewhere.

"The difference is somewhere between 5dB to 6dB better than in summer," Jaffar said. "But operators normally design for the worst case, which is summer. I've heard operators fine-tune the network to incorporate the change from summer to winter, but I'd say that they do rarely because the frequency reuse in rural areas isn't so tight. So even if there's a big change, 5dB to 6dB, it doesn't cause problems."

One caveat: Rural areas don't necessarily stay rural, especially if they're on a city's fringe, where land is relatively cheap, and subdivisions and shopping centers can replace forests in a matter of months.

"Rarely do you want to take this for granted," Jaffar said. "You still have keep your cell updated with the development of the area if, say, a forest changes into a development."

PLAN AHEAD Perhaps the best way to avoid being short-changed by foliage is to identify potential problem areas during the design stage. Most design tools base their propagation predictions on a terrain database, which has the land-usage type, such as rural or dense-urban. Looking at the clutter, whether it's buildings or a forest, helps identify areas where coverage might be spotty.

The clutter data's accuracy directly affects the design. In a typical scenario, a database house turns topographical information, such as aerial photos and satellite images, into a database of different clutter regions.

The clutter database's accuracy depends heavily on both the images' age and resolution. One source is the U.S. Geological Survey (USGS).

"That's one most often used, and that's a classic example of something that's out-of-date and low-resolution," Anderson said. "The clutter data that most people use from the USGS has a 200-meter grid size. That's everybody's inexpensive starting point."

The higher the data's resolution, the easier it is to understand whether an area is urbanized. Urban areas generally are less prone to foliage-related attenuation, but New York City's Central Park is a good example of an exception to that rule.

"If you went down to 50 meters, you might be able to pick up whether there are really tall buildings or whether there's a park there," Anderson said. "At 50 meters or less, potentially you can pick up freeways because they're broad spans of concrete. You can start to pick up details like that rather than (assuming that) it's all urbanized because what's on both sides of it is urbanized. Your propagation can take that into account and be a little more accurate."

"Tuning" the design tool adds another safety margin. Drive testing parts of the market to determine how signals propagate and then feeding that data into the tool helps ensure that its predictions match the real world as closely as possible.

One option is to use test transmitters at planned site locations, but if there's another carrier in the same band already serving that market, they might be unnecessary.

"You could do some testing off of their signals and get that same kind of information (about) what impact the topography and trees would have," said Howard Bower, GTE Wireless director of network, Midwest area.

Another factor is the time of year when the drive-test data is collected. A network designed in the winter and turned on in the spring might be plagued by spotty coverage because the drive testing was done when most trees were bare. If the drive testing were done in the spring, and the foliage is particularly dense, the network might suffer from interference in the fall and winter because both the power and antenna heights were increased to accommodate the signal losses.

"Seasonal variations for the foliage on deciduous trees adds a further issue that's difficult to model," Anderson said.

At the very least, understanding that foliage can affect propagation helps avoid as many seasonal surprises as possible. That understanding appears to vary depending on the latitude where the wireless-service provider's markets are located.

"Much more effort is spent on predicting coverage in dense-urban areas, where there is big business," said Maria Thiessen, Ericsson senior specialist, RF-propagation models. "I have heard about a case in Malaysia where they were discussing how to take wet leaves into account. But in Europe and even in the United States and Canada, the effect of foliage is not taken into account -- even if there are four seasons."

That attitude could change, thanks in part to sophisticated tools that have replaced crude assumptions and show that foliage can be a significant factor.

"As tools and computers have gotten better, people have been willing to take into account things that they waved their hands at before," Anderson said. "For example, 10 years ago, rather than trying to use a clutter database, people would just say, 'I know there's a bunch of trees around here, so I'll say the signals are 10dB weaker than my prediction tool or hand calculations show it being.'"