The Flat World Theory

There are some direct benefits the user sees. By reducing the number of hops on the network, data travels faster between end points, greatly reducing the network latency to help support real-time applications such as voice over IP (VoIP), gaming and videoconferencing. But the biggest operational advantage of the flat IP network is that it eliminates the artificial bottleneck at the BSC. Those BSCs treat packet data sessions much the same way they do voice sessions, even though the difference between the two is pronounced. One voice session varies very little from another, while a data session could be anything from a BlackBerry pinging its location to an enterprise server to a high-bandwidth, real-time video call or a multimegabyte download. Some applications are continuously running with an open connection to the network. Others might access the network just once a day for a few minutes. The BSC, however, manages all of these sessions as it would voice sessions, regardless of the amount or type of data being transmitted. The result is a network that doesn't scale easily for data — a network that can support far fewer data subscribers at the core than its radio capacity would imply.

As an example, Nokia Siemens Networks estimates a traditional 3G high-speed packet access (HSPA) network with 3000 Node B base stations would support 2.25 million subscribers. A typical hierarchal UMTS architecture could allow only 12% of those subscribers to have broadband subscriptions due to capacity limitations in RNCs and the core. Implementing a flat IP architecture, which would offload all packet data traffic directly from the base station to GGSN, however, would allow an operator to offer 40% to 75% more broadband subscriptions — a difference of 100,000 to 200,000 subscribers. The only other way to achieve that number of subscriptions on a traditional HSPA network would be to increase the ratio of RNCs to base stations. And as data traffic grows, the problem is exacerbated. Adding more capacity on the radio access network necessitates equal investment in the core.

“It's a painful, stair-stepped growing method,” said Bryan Boudreaux, head of global LTE solutions for NSN. “Adding a single RNC can add another $700,000 to cost of your network.” Imagine an operator having to add 100 new HSPA carriers to support increased data demands. Depending on the load of the current RNCs and where the new base stations are clustered, multiple new RNCs might be needed, multiplying the investment required in the network, Boudreaux said. “We're getting so much data traffic we needed to find a way to deal with it in a completely different manner,” he said.

The first flat IP architectures have emerged with WiMAX, and future LTE networks will be flat by definition. But what about the cellular networks of today, over which the mobile data revolution has already begun? The 3GPP has made a flat IP core an option in its standard for evolved HSPA, which many operators have begun to test and some have deployed. Several vendors have designed products targeting the current 3G market. Airvana and Alcatel-Lucent have designed a base station router that collapses the functions of the RNC and core elements into a single box, an approach that several femtocell-makers also have adopted. Ericsson and NSN have developed macrocellular implementations that would allow mobile operators to deploy an entire HSPA network without RNCs or SGSNs. So far, though, established operators haven't taken up the technology because they already have traditional UMTS networks in place that they're using for both voice and data.

“When the UMTS standards first appeared, no one anticipated the demand for data as we see it today. The mind-set was voice, voice, voice,” Boudreaux said. “For voice there was no reason to bring in a flat IP architecture.”

There may be a few operators that have spare UMTS spectrum to launch a parallel data-only HSPA network or others that may opt to replace their 850 MHz and 900 MHz cellular networks with HSPA — for those operators a collapsed network is an option. But for the majority of operators, the opportunity to flatten the network comes with the opportunity to start from scratch. The 3G operators that have launched NSN's Internet-HSPA flat architecture have been operators with no legacy network to speak of, such as T-2 in Slovenia and Terrestar in the U.S. Instead of deploying 3G voice, they're jumping directly into VoIP. As a new 4G technology, WiMAX also isn't shackled by a legacy network architecture. Cellular operators will get the opportunity to start over, too. For them the turning point will be LTE.