Flarion, which has just completed its first round of financing to the tune of $12.5 million, differs from other companies in the next-generation cellular race by starting with data, said Rajiv Laroia, the company's founder and chief technical officer.

"The problem is that the cellular networks to date have been voice-centric, with data laid on top," Laroia said. "We have taken the opposite approach by focusing on data transport first and then laying on voice as the networks expand in time to support voice-over-IP Internet Protocol."

The use of the proprietary Flash OFDM technology — "flash" refers to the fast-hopping technology employed — gives the new cellular proposal a robustness not yet seen in this field. That's important in an application plagued by the vagaries of signals bouncing off many surfaces — a phenomenon known as multipath interference — or off moving objects. Where conventional single-carrier transmission schemes send only one signal at a time using one radio frequency, OFDM sends multiple high-speed signals concurrently on different frequencies. This results in very efficient use of bandwidth and robust communications in the presence of noise and interference.

Flarion also expects its scheme to tame latencies — which will determine its effectiveness in an interactive environment — to 20 milliseconds while cranking up data rates scalable from 384 kbits/second to 1 Mbit/s downstream, and up to 1 Mbit/s upstream. (The upstream limitation has more to do with the battery power required to deliver more bits per symbol as the data rates go up.)

The company is also keeping an eye on carrier costs. "Most people design wireless systems for maximum bits per second per hertz," said Laroia. "We designed ours for bits per second per hertz per dollar spent on infrastructure."

So armed, Flarion will be vying for its share of a wireless communications market that the Yankee Group, a Boston-based analyst firm, says will top $13.2 billion per year by 2003.

Competitive field

Of course, the first question is whether a new modulation scheme can make its way in a market dominated by various flavors of CDMA, TDMA and FDMA. All of these are before the International Telecommunication Union, in one form or another, as candidates for standardization under the elusive term of third-generation broadband cellular networks.

But according to Craig Mathias, principal at the Farpoint Group, a research and consulting firm based in Ashland Mass., Flash OFDM and any other viable technology that can answer the demand for high-speed wireless data certainly stand a chance.

"Keep in mind that 3G is a concept, and not an implementation of technology," Mathias said. "There's no reason to expect that we'll see a single 3G standard, or that it will be based on wideband CDMA or any of the current proposals. I really don't believe that the market is closed. There's still lots of room for innovation, as we're still years away from deploying this stuff in volume."

"It's essentially a footrace for dollars invested," concurred Ray Dolan, president and chief executive officer at Flarion. "The size of the market will easily justify investments in alternate technologies." The incumbents, he said, will be chasing "more efficiency for their networks." And for new players entering the market "with more of a data approach" — like top-branded Internet service providers and competitive local-exchange carriers that are already in the data space — "this will be what they're looking for," Dolan said.

Asked why no one has implemented a mobile scheme based on OFDM to date, despite its inherent robustness in the presence of multipath interference, Mathias responded, "Part of the issue is the complexity and the fact that OFDM has its genesis in the military, thereby limiting the exposure engineers have had to it. However, the pending arrival of 802.11a and HiperLAN2 means exposure will increase, so the cost will come down."

Mathias sees a battle emerging over the next few years between wideband and ultrawideband, CDMA and OFDM for the heart and soul of the mobile communications market.

Charging into that battle, Flarion has armored itself with physical-layer (PHY) and media-access control (MAC) layer devices. The pair were closely tied from the initial design stages — essential, said Laroia, if the requirements of the data-centric model were to be met, considering the vagaries of the wireless channel.

Double design

"Many people take the PHY and then design the MAC," he said. "We did both together, as that allows us to address many of the problems with the wireless channel, in terms of reliability, low latency and high bandwidth efficiency."

Laroia said it does no good to take a MAC designed for other markets — for example, the Docsis data-over-cable spec — and add a new PHY and expect it to work for wireless. "In my opinion, this is a very wrong way to design a system with very scarce spectrum," he said. "You want to design these two layers for optimal use of spectrum, with maximum data quality."

For the PHY layer, Flarion opted for its own, patented version of OFDM. In Flash OFDM, a signal hops from tone to tone at the rate of roughly 10,000 times per second. Every user ends up signaling across the band on all tones, effectively turning OFDM into a spread-spectrum technology.

"Wideband CDMA is also spread-spectrum, but only in the time domain," said Laroia. "Flash OFDM is spread in the frequency domain, so you end up getting all the frequency-diversity advantages and interference-averaging benefits that spread-spectrum offers in the cellular environment."

Flash OFDM can be likened to putting time-division multiple access on top of OFDM, yielding all the benefits of both technologies in terms of robustness under mobility and channel-delay spreads, interference averaging from other cells and orthogonality inside the cell.

The MAC itself is key to the system's potential success over the very difficult air interface. According to Laroia, "When it comes to error correction, end-to-end retransmission is very expensive. The most unreliable part of the link is the air link, so we have link protocols built into the MAC that offer a very reliable link layer to the IP traffic, along with proprietary error-control mechanisms that are not end-to-end."

The link layer thus has feedback built in that does not cause end-to-end retransmission — only across the wireless link. This is another benefit of the PHY/MAC linkage, which in this case lowers the retransmissions and, therefore, latencies.

As for scalability, the current cellular infrastructure is capable of handling voice at roughly 8 kbits/s. But the inevitable rise of high-speed wireless data soon will take its toll and require a much denser deployment of cellular basestations if the demand is to be met.

Basestation buildup

Unfortunately, that isn't so easy. The public isn't open to the idea of a cellular station every 50 miles. Nor is it cost-effective, since each added basestation may require frequency reengineering of the entire network. This is acceptable if, as happens now, a scant few basestations are added each year. But some say the burgeoning wireless data market will boost the deployment rate to one a day.

Eliminating growing pains is key to Flarion's system. To add in a basestation, it simply has to be connected to the backhaul, followed by a punch of the reset button. It needn't know about other basestations in the neighborhood, nor must any existing basestations undergo changes once the addition is made.

"We believe this will be critical as data catches on," said Laroia. "The trade-off you make in a system to aid deployment costs is really what will determine if the architecture scales to handle capacity requirements." Laroia says flatly that "the current proposed architectures are not scalable cost-effectively for rapid deployment of data."