Nailing Down the Link

Engineering a fixed-wireless system is not as complex as engineering a mobile-wireless system. After all, “fixed” means that both ends of the link stay put. Beyond that, however, all RF engineering follows the same principles, according to Doug Carter, XO CTO.

“The trade-offs are generally distance and speed in any radio link,” he said.

In the millimeter-wave bands, those above 10GHz, which include the LMDS frequencies, rain fade and line of sight also limit distance. But if experienced engineers have the right tools and databases, engineering the links is not difficult, said Bryan Boyd, TeraGo president & CEO.

“Look at the topography by radar mapping and through your own eyes,” Boyd said. “Look at the environmental databases for rain, humidity and temperature.”

Then put the information into the tools that determine the distance between the radios and translate that into site selection.

Antenna Alignment

Installing the antennas is pretty much the same as installing a DirecTV receiver, said both Allan Evans, Netro chief scientist, and Carter.

“The pickiest part is making sure that the two antennas are pointed at one another because they are very narrow (2-degree) beams,” Carter said. “It doesn't get out of alignment unless you did a bad job of mounting it.”

Carter said there are “a couple of ways you can screw it up.” The first is installing a link to go farther than the calculations say it can go. The other is mounting an antenna on something that might move, instead of something like a brick wall.

“A typical mistake is putting it on an air-conditioning enclosure that someone else might work on,” Carter said.

Although building sway is more of an issue in free-space optics, Evans pointed out tall buildings can sway by as much as 10 feet in each direction, so if precise calibration of the customer-premises equipment (CPE) is required, it could be voided by the sway. Netro's solution is constant long-loop calibration that can change the calibration or power levels depending on wind-stress or rain-fade conditions.

David Ackerman, Ceragon Networks president of U.S. operations, said that his system has intelligence built in the form of an automatic power control so that it can identify attenuation.

“If there is a fade, the system will automatically increase power output to keep the link up and reduce it later,” he said.

Evans said all Netro equipment is plug and play, making it simple to install. The installer plugs a multimeter into a port on the back of the radio, which tracks the received signal strength. He then points the antenna until he finds a peak in the received-signal strength and locks the radio into place.

Minimal Outages

Availability is a crucial issue for carriers in the millimeter-wave bands as they generally are providing carrier-grade services.

Ackerman recommends engineering links to five 9s, which means less than five minutes of outage per year.

Boyd said TeraGo engineers to five 9s to guarantee the customer three 9s in a service-level agreement. It involves more than simply monitoring to see what might go wrong.

“You also monitor different sources of information that will reduce problems,” Boyd said, including congestion, network capacity or other factors.

Network management is important even before the equipment is installed, according to Andres Belloni, Alcatel director for product management in the network and service management business unit.

“In our management system, the installation and provisioning of the CPE devices are two separate steps,” Belloni said, because these generally are two separate departments for the carrier. The network-management system must be able to do pre-provisioning. This will pre-define the link in terms of the bandwidth requirements and frequency, associating the CPE with a base station.

“Once the CPE acquires sync with the base station, the management system will automatically notify the operator of the existence of the device in the network,” Belloni said.

Reliability is factored into the equation that includes speed, distance and power, Carter added.

“With a given radio, the power is set, but you can change the speed and the distance,” he said. Radio links are designed to be “reasonably comparable with fiber.”

“When we design a fiber link, we can only run the fiber so many miles before we have to stop and regenerate the signal,” he said. It's the same with wireless. You want to achieve a certain bit error rate and a certain availability.

Licensed-Spectrum Advantage

By using licensed spectrum, a carrier can plan for, and thus protect against, interference.

Because the beam is so narrow, seldom will cell sites be close enough to interfere with one another, Carter said.

“Design is more noise-limited than interference-limited,” Carter said. “You just have to keep the signal above the noise floor.”

Evans noted that his company has worked out numerous frequency reuse plans for its customers that will allow them to get essentially 200% of the spectrum use on each base station and be able to maintain enough interference isolation between the base stations to allow a contiguous build-out pattern.

Frequency reuse is important because if you have spectrum and cannot reuse as many frequencies as possible, you're wasting money, Belloni said. As you add frequencies, you also may end up moving customers from one frequency to another, and the management system should accommodate that so it can be done quickly.

The bottom line is that once the link is up, a variety of things can go wrong, but a carrier can overcome them, Boyd said.

“Properly engineered installation is key,” he said. “Then you should know what might occur and have the solutions in place that will mitigate them or make them transparent to the customer.”

Will the Vendor Survive?

As carriers select equipment, they look at their needs, what the equipment will do and what it will cost. But lately they also are looking closely at the long-term viability of the vendors.

“We perform a rigorous assessment of their financial state, not just their own viability but the viability of the category in which they are participating,” said Bryan Boyd, TeraGo president & CEO. “Will the market they serve be there? Will they be in that market?”

This is always a consideration, said Doug Carter, XO CTO. But he added that if a company goes into Chapter 11, it doesn't mean everything it has is useless. For instance, XO has radios from Triton Network Systems, a company that announced plans to liquidate in August. The radios still work. What tends to wear out is the cables, Carter said. So XO called Triton, and Triton told the carrier who makes the cables so XO could get in touch with that company directly.

“We never have a case when a manufacturer is totally vertically integrated,” Carter said.

“Components are generally made by someone (else).”

Otherwise, vendor selection is pretty straightforward, according to Boyd. TeraGo just went through the selection process on core routing equipment. It short-listed the suppliers and then obtained equipment to lab test and then field test.

Carter describes the selection process as “an uneasy compromise between cost and reliability and ease of use.”

Vendor advice doesn't vary much. David Ackerman, Ceragon Networks president of U.S. operations, suggests first knowing what you want the equipment to do: applications, capacity, current and future needs, availability. He emphasizes vendor reliability and cautions about new products.

“Make sure the product was well tested and installed in significant numbers, or you might end up with a product with a very nice data sheet and still a lot of problems,” he said.

Allan Evans, Netro chief scientist, said to look for plug-and-play installation.

“You shouldn't have to have a PhD to calibrate the equipment at the time of installation,” he said.

Evans also said high service availability is important — not just the link availability, but the rest of the network — so full redundancy at the base station is critical.

Reliability comes first to mind for Andres Belloni, Alcatel director for product management.

“Go with a company that has a reputation and a history in deploying carrier-grade systems,” he said, referring to both the hardware and the management system.

What about price?

“We need a return on investment, so price is really a function of what (the equipment) does for you,” Boyd said. You can't spend money on something too big for the opportunity in terms of revenue, he said.

You have to look at the total cost of ownership, Evans said, not just the cost of the wireless equipment. Belloni said much the same thing in pointing out that if you pay a premium for a management system, you will be compensated in operational cost savings.

Price is at the bottom of Ackerman's suggested criteria for selecting equipment. Even if the price is low, it doesn't make sense if the equipment doesn't meet your needs, he said.

Houston Benchmarks

Drive testing of six networks in the Houston area included 2,448 miles driven between Feb. 5 and Feb. 15, 2001. Calls totaled 9,910. The figures represent an average of all Houston networks.

Next Market: Montreal

*Access failure is determined by dividing the total call attempts by the total no-service and blocked calls.

**Voice-quality measurements are scored on a 1-to-5 scale with 5 being the best. The figure is calculated by taking an average of the electronic mean opinion score for both the forward and reverse links of each operator's network.

Statistics provided by Agilent Technologies (www.agilent.com/find/benchmarking)