Growing Pains

Imagine turning to the business section of the local newspaper and seeing an article about your network's debut. A proud moment, right? Now imagine that the articles next to it are about a new, 50-acre business park and a 250-unit upscale subdivision, both slated to open in a year's time. It's the first you've heard of them, and now it's a scramble to figure out how to accommodate more traffic.

You didn't have a crystal ball when you did the frequency plan, but that won't soothe the boss when you explain that there isn't room for any more cable runs and antennas on the co-located sites in the area. New sites of your own? You'll pay dearly for them now: The announcements have sent real-estate prices in those areas through the roof. Now what?

One attribute of a good frequency plan is its ability to accommodate growth. The more information about the market that goes into the network design, the better.

"Growth trending is fundamental to the frequency-planning process," said Andrew Groome, Safco Technologies manager of technology transfer & training. "As a general rule of thumb, if I'm assigning channels for what I have today, I am way behind."

There's no shortage of forecasts of how a market will change. State and federal transportation agencies, regional economic-development agencies and municipalities' master plans can help predict when and where growth will occur. Even geocoded income and education data can help predict how subscribers in a particular area will use their phones.

"Morphology is starting to play a bigger and bigger role in the things RF engineers look at," said Javier Pardo, TeleworX senior RF engineer. "They're expensive, but people are realizing the importance they have. Carriers are putting more focus on getting more precise forecasts."

If a market appears poised for rapid growth, a frequency plan that favors smaller sites and tighter frequency reuse might save money in the long run if land and tower space have grown scarce when it's time to add or move sites. Reserving large-coverage-area sites mainly for highways also limits interference and improves frequency reuse.

"A rule of thumb would be that if you have a site, and you can see it from all over the city, don't build it," said John Arpee, a 16-year veteran of network design and now CTO of Scoreboard Wireless, an optimization-software company. "It's wonderful at the beginning, but you know that you will be taking that site down."

Ultimately, a frequency plan is shaped as much by business decisions as engineering decisions.

"If you expect to run out of capacity in two or three years, then you're probably better off putting in smaller cells," Arpee said. "You really need to pay attention to where you think your traffic is going to go and how much spectrum you have. That will tell you when you need to move those cells. That kind of analysis should be done from day one. You shouldn't wait until after the network is built."

One temptation is to put ample capacity everywhere so that unexpected growth won't cause problems.

"Allocating three or four frequencies per cell at build-out time might be a good strategy in order to minimize future frequency-plan tuning, but it might prove a poor and inefficient spectrum allocation from the launch day until the network reaches that level of maturity," said TeleworX's Pardo. "It is a tradeoff that the frequency planner needs to make between retuning as the network growth requires it or allocating and not using precious resources -- channels -- ahead of time."

The alternative is to tweak the frequency plan as the network grows. Automatic frequency-planning tools can streamline that chore and shrink the time to produce a full plan to a few hours. Even so, their ability to create a design that matches the market's coverage and capacity needs hinges largely on the user.

"The typical propagation tool has a standard deviation of around 8dB," said Vikrant Tannan, LCC senior principal engineer. "That means, on the average, you're going to be 8dB off when you do propagation prediction. For frequency-planning purposes, when you're doing a C/I (carrier-to-interference) calculation, that's essentially a difference of between two signal strengths. If you have a standard deviation of 8dB, the difference of signal strength will have a standard deviation between 8dB and 16dB. That makes it a little worse than the propagation prediction."

One way to ensure that the tool's design will work is to use test transmitters and do drive testing to see how signals propagate. Feeding the collected data into the tool refines the design in a process known as measurement integration, where real-world conditions replace models.

Although automatic data-collection tools can streamline the process even further, drive testing usually is limited by factors such as budget, sheer network size and time to market. One short cut is to identify a handful of areas that typify the rest of the market and then drive test only those areas.

"You get a cluster of 10 or 12 sites, and you put all your effort into that," Pardo said. "Then you would interpolate those results into the other areas that you think share the same characteristics. But ideally, if you can do the whole network, why not?"

Optimize to Maximize

Even the best frequency plan can't accommodate every vagary of a market's growth. A steadily increasing number of blocked and dropped calls can indicate that capacity is stretched thin. As the geographic distance shrinks between cells using the same frequencies, the resulting interference can trigger another indicator: customer complaints.

"That information gets fed to the young frequency-planning engineer, who will begin to pull his hair out because he just can't find any more channels," said Safco's Groome.

The budget is another consideration: adding a channel easily can be costly, depending on whether backhaul upgrades and other modifications are required. But adding channels is a solution for only so long.

"Interference definitely goes up as traffic goes up because there are more channels being squeezed into the cell sites," Arpee said. "The engineers should be tracking what the traffic trends are. If you're staying on top of that, ideally, you never have a capacity issue in the network."

If real estate is available and zoning ordinances aren't overly draconian, cell splitting is one option.

"Cell splitting affects frequency planning dramatically," Pardo said. "The level of impact basically depends on how the previous frequency plan was designed. If it was designed with the future cell splitting in mind, the frequency planner might not even have to change anything. That would be the case when the frequency planner reserved frequencies for future use.

"On the other hand, if the frequency plan previous to cell splitting was making use of all the available frequencies, then the frequency planner will basically have to put a new frequency plan in place from scratch."

Another alternative is RF optimization. Lowering the height or power of a site and downtilting antennas are two ways to reduce interference with other sites and thus improve frequency reuse.

"Make sure you've taken full advantage of the capacity of your network," Tannan said. "That means optimizing your network. There's hidden capacity in your network. You just need to know how to find it."