Will femtocells evolve into the small cells of the future?

Globally, there are 2.3 million femtocells deployed, according to Informa, but almost every single one of them is of the private variety—home and business access points used to supply select customers with coverage in their living rooms and office corridors (Unfiltered: A dubious achievements: femtos now surpass macros). The femto industry, however, has long been prophesying a transformation of the femtocell from a personal base station into a public access point populating dense urban environments, where its high-capacity and small footprint can be put to best use.

When will this transformation take place? Continuous Computing vice president of product line management and incoming RadiSys CTO Manish Singh believes it will take place simultaneously with the transition to long-term evolution (LTE) networks. And he isn’t just guessing. Continuous is seeing plenty of evidence of that transformation in its changing customer base.

Today Continuous has more than 30 customers licensing its Trillium software and reference designs to build 3G femtocells, but those customers are primarily consumer electronics makers—the same vendors that build the set-top boxes and gateways for home and enterprise networks, Singh said. But for LTE, Continuous is attracting a distinctly different type of customer. Telecom equipment and outdoor coverage solution providers make up the bulk of its 22 LTE licensing deals, and those customers have a much different product in mind than the home femtocell, Singh said. They’re building public access femtocells that will be used to supplement mobile broadband network capacity. In short, they’ll be some of the first commercial network nodes that live up to the small cells and heterogeneous network concepts that have taken hold of the industry of late, Singh said.

When that new breed of femtocell emerges, Singh said he believes that not only will deployed femtocells scale from the millions into the hundreds of millions, but that the business case for femtocells will become crystal clear. Today operators are trying to sell the femtos to their customers as a coverage solution, which is the primary reason why femtos have failed to gain traction, he said.

“How do you sell a coverage proposition if coverage is what you’ve been selling forever?” Singh said. “Operators haven’t yet crossed that bridge from coverage to capacity. When they do, the value of the femtocell will change.”

Rather than sell or give a private base station to a customer in order to prevent him or her from churning—making it a customer retention expense—femtocells will be used to add gobs of bandwidth to overtaxed networks at a fraction of the cost it takes to deploy a macrocell—thus turning the femto from an expense into capital savings. Each femto offers the same capacity as its macro equivalent, but they can be deployed in dense clusters, exponentially increasing overall network capacity without exponentially increasing costs, Singh said.

But if femtos are so efficient at relieving network congestion, why is LTE a trigger for the mass deployment? Why are they not being used to supplement 3G and high-speed packet access (HSPA) networks, which are today being strained by data traffic demands? According to Singh, the reasons are a combination of LTE’s sophistication as a technology and missteps vendors made in designing the first generation of femtocells.

The biggest reason operators cite for their hesitancy is interference. While some are deploying femtos on completely separate frequencies than their macro networks, most don’t have that luxury. They need their femtos and their macro networks to share the airwaves, and when femtos are placed within a macrocell at the same frequency, interference is bound to occur. Self-organizing network (SON) and radio resource management technology was supposed to sort all this out. The femtocell is supposed to determine its location through GPS, scan the local radio environment and then adjust its transmission power to avoid interfering with the overarching macro network and other nearby femtocells. But judging from the interference complaints of the operators, femtos still aren’t doing what they’re supposed to do.

Singh plainly acknowledged that the initial femtocells haven’t quite lived up to their interference mitigation claims. A big part of the problem is where femtocells were first introduced on the wireless technology evolution path. When the first femtos emerged, the 3G standards had all been finalized and CDMA and UMTS/HSPA had been around for some time. While femtocell SON technologies were included in later iterations of the 3GPP standards, they haven’t been implemented in current 3G networks. Although femtocells could optimize themselves relative the macro network and organize around one another, the femto and macro networks couldn’t collaborate directly with one another.

Also, the femto SON technologies that were available were implemented haphazardly. SON was supposed to be a technology with which femto vendors differentiated themselves, but the result was a bunch of proprietary SON solutions that didn’t fully address operators’ interference concerns, Singh said.
Femtocells, however, arrived on the scene just in time for LTE. Provisions for small cells have been fully adopted into the standard. “In the case of LTE, early on it was incorporated into the architecture that you would networks running with multiple layers of cells,” Singh said. Rather than shoehorn femtos into a network that was never intended to support them, the LTE standards assume that every cell is an intelligent node that “senses its environment and changes its power to mitigate its impact on that environment,” Singh said. “In 3G, the macro network simply can’t do that.”

Furthermore, LTE’s orthogonal frequency division multiplexing access (OFDMA) radio interface allows for much more sophisticated SON implementations, Singh said. The sub-channels—or tones—of OFDMA give operators much more flexibility in shaping their cells, which will be critical to interlocking the different shapes and sizes of cells in heterogeneous networks. “Much more complex SON algorithms can be created for OFDMA,” Singh said. “That gives you much more freedom in managing your resources.”

The opportunity LTE presents to create a new type of dense network topology hasn’t been lost on other vendors. Alcatel-Lucent and Nokia Siemens Networks both recently unveiled their future network architectures, called lightRadio (CP: ALU’s new building block architecture) and Liquid Radio (CP: NSN pours out Liquid Radio) respectively, which both use small radios as the building blocks for future networks. Ericsson has talked extensively about how future networks will evolve into heterogeneous topologies (what it calls het-nets), with small and large cells working in unison (CP: Shrinking cells and narrowing beams). Huawei has launched something akin to outdoor public femtocell, calling it the Microcell, to target these kinds of multi-layered network deployments (CP: Huawei doing small cells the old-fashioned way). Even carrier Wi-Fi vendors are getting into the act. BelAir Networks is working with at least one operator to use its strand picocell to augment its 3G capacity (CP: BelAir stakes its claim in small cells).

The difference between public access femtocells and the big vendors’ small cells is debatable. The big network equipment providers tend to call their products metrocells or microcells to differentiate them from femtos, but both accomplish the same goal: adding a dense layer of network capacity under the macro network’s coverage umbrella. The biggest differentiator between them is their conceptual approach to the network. Alcatel-Lucent and NSN believe there should be a single network built using the same basic elements but with those elements used in different configurations, while femto vendors are proposing a duality. The macro network and small cell network would be deployed and managed separately, though they would collaborate through standard interfaces. Although the simplicity of the former approach is much more elegant, femtocells have the advantage of being a lot cheaper.

Singh said he believes different operators will adopt different approaches to the network, and some may even combine the two. Either way, Continuous Computing and its future owner RadiSys probably aren’t too worried, since Continuous licenses its small cell technology to the big telecom vendors as well as the specialty femto manufacturers.

So what’s the verdict? Do we have to wait for broad-scale adoption of LTE before we’ll see femtocells or their metro and micro equivalents take off? Continuous isn’t sitting on its hands waiting around for that day. Singh still thinks femtos have a shot in 3G as a small cell capacity technology. CDMA and HSPA femtos will never be able to negotiate interference the way an LTE femto could, but 3G femtos can do a much better job than they are today, Singh said.

Continuous is addressing the problem of poor SON technology by introducing its own more sophisticated software in its next release of its Trillium architecture. Vendors that have SON down pat may not need to tap into Trillium’s new capabilities, but those with a less-than-desirable SON solution can take advantage of it. Operators with unused time division duplexing (TDD) spectrum are also looking at using those bands solely for femtos, creating what amounts to a mobile data offload network completely separate from the macro network, Singh said.

He also pointed out that 3G networks won’t simply go away when LTE networks go live. LTE will provide tremendous relief to those networks, but operators will have billions of 3G customers consuming tremendous amounts of data. If carriers and vendors don’t deal with the interference issues surrounding 3G femtos today, they’ll have to do it tomorrow.

“You can never completely take away interference,” Singh said. “It will always be a problem, but interference mitigation is not an impossible goal. The industry has come a long way with SON interference management.”