Up for Grabs

Leap Wireless took another significant step toward wireless data services in November with its selection of Cambia's mobile data server solution. A crop of entrepreneurial start-ups similar to Cambia Networks is racing to ink similar carrier deals. While applications have been the focus of the wireless data industry for about the last two years, recent focus — and venture capital — has shifted to the network.

According to Rutberg & Company research, funding for startup wireless infrastructure and semiconductor vendors accelerated in the past 15 months while investment in application developers, middleware makers and carriers has slowed.

The realization that serious wireless data applications won't run well on existing network management solutions spawned companies such as Cambia, WaterCove Networks, Tahoe Networks and Megisto Systems.

While recognizing that the first necessary step toward always-on connections is deployment of 1X and GPRS radio access networks, these mobile IP startups contend that true profitability will elude carriers until they install the intelligent wireless data management solutions they're building and testing. Cambia was first to market with its announcement of the Leap deal, which expects to fully deploy Cambia's mobile data server along with 1X early next year.

All of these new solutions involve a combination of servers, serving nodes, home agents and software. Configured in different ways, their goal is the same: efficient, scalable provisioning, management and billing of wireless data services.

Shotgun Wedding

To successfully place their management solutions within carrier networks, most admit they'll have to partner with incumbent vendors such as Ericsson, Lucent, Motorola, Nokia or Nortel.

“Very likely, for a large-scale deployment, we would be shotgun-wed with one of those guys,” said Mark Tubinis, WaterCove Networks vice president of development. Despite the colorful terminology he uses, Tubinis would welcome such a deal. And off the record, most of these startups admit they're likely acquisition targets for larger network vendors. The exchange of financial and intellectual capital between startups and incumbents is familiar territory for many of these entrepreneurs. Take Carol Politi, for example.

Last May, Politi and Gordon Saussy founded Megisto Systems after working for a short time at Ericsson, after it had purchased their former employer, Torrent Network Technologies.

“We saw there was a field of challenges in the core space that weren't being addressed by any of the incumbent vendors,” said Politi, Megisto's vice president of marketing. “This is not Ericsson-specific; Ericsson is a fantastic company. But just like all of the other incumbent vendors, they make most of their money from radio gear.”

Politi contends that incumbent vendors are narrowly focused on basic cell-site equipment, which makes up 90% of the overall wireless network market. The remaining 10% of the market, she said, is up for grabs because incumbents haven't the expertise or financial incentive to pursue it.

Their attempts so far have fallen short, in her opinion: “In essence, most incumbents are modifying gear that was designed for the fixed network to work in the mobile core,” Politi said. “But this is a totally different plane.”

Instead of working within Ericsson, Politi and Saussy started Megisto. Others have followed similar paths. Chris Rittler was hired as Cambia's senior vice president of product development after 13 years with Motorola, having worked most recently on its CDMA wireless infrastructure team. Tahoe co-founders Anthony Alles and Arthur Lin sold Shasta Networks to Lucent prior to embarking on their latest entrepreneurial adventure.

Ready, Set, Grow

Cambia's mobile data server architecture includes a packet data serving node (PDSN), home agent (HA), control module and server. These basic elements are common among new wireless data management solutions, said Rittler.

“What's novel is our approach,” he said. “We're leveraging open architectures — IBM servers and open operating systems like Linux. By doing that, we're leveraging the best-in-class density and scalability.”

Early WAP gateways, for example, only addressed part of the wireless-data problem. They're circuit-based, often drop sessions and cannot easily measure sessions — a requirement for moving beyond flat-rate billing.

As networks become packet-based, and new wireless data management solutions are deployed, carriers will have the ability to charge by the byte, or download. These more detailed billing systems will be required as data use grows, and spectrum gets stretched, according to WaterCove's Mark Tubinis.

WaterCove's solution allows carriers to use real-time metering in the core network to tell applications, such as a streaming video service, when a user's pre-set limit has been reached.

Load ‘em Up

U.S. wireless data subscribers are still measured in the millions, not tens of millions, per carrier. But the industry is beginning to show signs of significant growth.

“The reality is, you can manage mobile subscriber services on a PC or a Sun platform if you don't have any real subscribers,” Politi said.

Current management systems using a half-dozen servers won't be able to serve growing data networks efficiently.

“We're focused on simplifying the service that gets provided by the operator so they can provision the service and manage the subscriber at the same point in the network,” she said. “If an operator's got to manipulate that much equipment … the operational costs alone are unmanageable.”

Martin Lord, Tahoe senior architect, said his company's solution provides data services, management and control from the mobile Internet edge, which can be compared to the MSC on the voice side of a carrier's network. (See Figure 1.) But repurposing voice networks for data won't do the job.

“A lot of people took for granted that it wouldn't be that hard to build nodes for the data plane,” Lord said. “Some managed to scale a data box on one of two fronts. They scaled the data plane well, they're not scaling the control plane.”

To succeed in providing wireless data services to tens of millions of subscribers, intelligence must be equally distributed among all network elements, according to Alan Cohen, Tahoe spokesperson.

“The interesting transition when you look at the old world of the mobile network, is the handset was smart, the base station was kind of smart, the network wasn't very smart, but the switch was smart,” Cohen said. “If you look at the Internet, most clients were smart but used dumbly — the browser was not a very smart client — the network was dumb, and the (application) servers were very smart. I think what you'll begin to see is a distribution of intelligence across the servers, switches and network elements.”

In a similar vein, WaterCove's Tubinis said wireless data management solutions must touch all layers of a network's architecture.

“Our solutions transcend not only the transport, or switching, dimension, but we also feed up through the control layer and the service delivery layer,” he said. “We felt that the operators trying to overcome the daunting challenges of their early (wireless data) services roll-out needed some additional tools in their arsenal, and our solution basically filled in some of those gaps.”

By giving the switching layer access to the application layer, WaterCove's management solution would allow a carrier to add virtual processing power to the application on a user's handset.

“Applications have a hard time doing packet counting, because of the sheer processing power required to do that at the application level,” Tubinis said. “So we have engines in our network switching layer that tell the applications when triggers are hit. They exist in the voice networks of today's mobile operators, but weren't incorporated in the first designs of data networks because it was just too hard.”

These network-based triggers could be used to rate shape data service, bill for streaming video or perform other data service management functions — without overtaxing handset processing power.

“What the operators have said is they want to preserve an open Internet construct,” he said. “But they need information on where their customers are going so they can build billing relationships with content providers. We've provided them with the tools to do that.”