Don’t look now, but your cell tower may be following you. AT&T’s networks are getting an upgrade that will transform them from static cellular grids into a kind of network organism whose cells will grow and shrink as customers move through them. Ultimately these networks — what we call self-optimizing networks (SONs) — will be a critical component in providing cheap and ubiquitous mobile data.
In the near term, SONs can help make better-performing networks that drop fewer calls, boost data capacity and smooth out the bandwidth peaks and valleys that still persist as we move from cell to cell. In the long term SONs will be a key component in implementing the high-density, multitiered heterogeneous networks of the future.
The immediate benefactors of this technology will be the carriers that will be able to build networks that pack enormous data capacity into small areas without buying new spectrum. But ultimately everyone stands to benefit from cheap and plentiful bandwidth. Consumers will get much bigger buckets of gigabytes for lower prices. Developers will be able to make bandwidth-intensive applications without worrying that spotty connectivity or costly data plans will prevent customers from using them. And device makers like Apple and Samsung can continue to innovate on mobile computing devices unimpeded by scarce mobile data capacity.
Let’s take a closer look at two different examples of SON techniques being implemented today.
Building a network one customer at a time
Networks are evolving from big macro cellular honeycombs that provide a blanket of coverage to heterogeneous networks. The latter, often called hetnets, contain all different kinds of nodes, from home femtocells and public small cells to Wi-Fi access points. That layering of access technologies means a network that was previously built on thousands of towers becomes a network with hundreds of thousands of transmitters.
U.K.-based Arieso is trying to solve the problem of where to put all of those new cells by building intelligence that automatically identifies where capacity is most needed based on the actual use of a carrier’s customers, said Arieso CTO Michael Flanagan. That intelligence can be used to plan large-scale capacity upgrades by deploying huge clusters of small cells or Wi-Fi access points at the exact locations where congestion is a problem. By “exact location,” Flanagan means not in the general vicinity of a tower but in the exact households, coffee shops or sidewalk intersections where that capacity is needed.
“For instance, if I have a customer who lives right at the edge of a cell consuming enormous amounts of data, he’s messing up the network for everyone else using that cell,” Flanagan said. But if a carrier were to put a small cell right on top of that customer, the network resources he was draining suddenly become free. “For less than a $1000 investment, I clear up a capacity problem, and assuming it’s a good customer paying a $100 monthly bill, I get a return on that investment after a year.”
That kind of precision turns network planning from an exercise in general guesswork to one of pinpoint planning. And once you scale into thousands or hundreds of thousands of pinpoints, you get a network truly optimized to deliver capacity where it is most needed — and for a fraction of the cost of adding capacity to the overall macro network.
There’s not much “self” in Arieso’s SON today — technicians still need to install the individual cells — but it’s getting there. Flanagan said the company’s tools will eventually be able to feed their awareness into network management systems, allowing carrier engineers to remotely reconfigure the network as capacity demands change and congestion shifts to different parts of the network.
But where SONs get really space-age is when that reconfiguration takes place both automatically and dynamically.
Breathing life into static cells
The thing about cellular networks is that they aren’t flat, evenly distributed pools of capacity. The bandwidth any device can receive depends on how close it is to the tower it’s connecting to. The closer you get to that cell edge, the smaller your broadband pipe becomes and the harder it is for the network to maintain your phone call.
What’s more, cellular networks don’t offer dedicated pipes to your cell phone like a DSL or cable connection. Instead capacity is shared among all the users within a cell. That means if you’re far off on the cell edge, the network is forced to devote more of its resources to maintaining your connection. That leaves fewer resources it can allocate to other customers sharing the same cell. Bottom line: One user traversing the edge of the network can wreck the party for all the customers closer in.
Intucell is challenging the notion that any customer ever needs to be on the edge of a cell: Rather than force the customer to move to get a better connection, why not move the cell? Its SON technology is dynamic, meaning the network changes shape as customers move through it. If a single customer is close to a cell edge while there are dozens of other customers well inside it, then the network morphs, telling a neighboring uncongested tower to grab the outlier. Meanwhile the original cell shrinks its radius to encompass the remaining customers, providing them faster and more-consistent data speeds.
“Until now optimizing a radio access network has been sort of a type of witchcraft,” Intucell CEO Rani Wellingstein said. “Engineers would typically retune networks on a weekly or a monthly basis. That kind of model was good for when traffic was not so dynamic, but today congestion is moving dynamically through the network.”
Currently operators like AT&T are using Intucell’s technology to retune and reconfigure their networks every few minutes, which is ideal for addressing the changing traffic patterns in a general way. But the future promise of dynamic SONs is that they will start controlling networks in real time.
Take the example of city bus with a Wi-Fi access point fed by an LTE connection. By anticipating where that hotspot on wheels is going, the network could grow the cell it is driving through, keeping its edge well ahead of the bus and thus maintaining its high-bandwidth connection. Meanwhile neighboring cells running on a different frequency would also expand, so that they overlap with the bus’ cell. When it’s finally time for the first cell to relinquish its connection to the second, the bus is already well within the receiving tower’s halo, giving it immediate access to a much better connection. The bus technically never gets near a cell edge, and the usual huge fluctuations in bandwidth it would normally experience moving through a static network never materialize.
Where SONs fit
AT&T is an early adopter of SONs, but it certainly won’t be the last. In fact, SONs are a critical element of future wireless networking technologies like LTE-Advanced. LTE-Advanced will not only incorporate hetnet principles, creating networks of every sort and size of cell, but will also introduce new techniques that will require carriers to precisely control their network configuration in real time, allowing them to mitigate interference and direct capacity exactly where it’s most needed in any given moment.
Given how critical SONs are to future networks, it’s actually rather surprising that smaller companies like Intucell, Arieso, Actix and Optimi are taking an early lead in developing the technology, said Peter Jarich, the service director of service provider infrastructure at research firm Current Analysis. All the major wireless infrastructure vendors have developed their own SON technology, and when it comes to buying gear that is essential to the evolution of the network, carriers like to stick with Ericsson, Alcatel-Lucent, Nokia Siemens Networks, Huawei and a handful of other Tier I suppliers, Jarich said.
“Would AT&T have gone to Intucell if SON was something they could have gotten from Alcatel-Lucent or Ericsson?” Jarich asked. “All of them have built self-organizing technology into their OSSes [operations support systems]. Organizing your network is a well-established principle. The optimization side, however, is obviously a very new space.”
In one sense, it’s logical that independent vendors would take an early lead on SONs. Hetnets by definition will be multivendor networks, with different suppliers providing the different components. But the SON technologies the Tier 1 vendors have developed are designed to work well only with their own base stations and OSS platforms. In other words, being vendor agnostic is critical to making SONs work.
That entrenchment, however, is changing as the big vendors delve into managed services, Jarich said. NSN, Alcatel-Lucent and Ericsson are no longer content to simply sell gear; they’re using their expertise to build, maintain and run carriers’ networks. Often that involves installing, fixing and operating a competitor’s equipment. For instance, Sprint has essentially handed to Ericsson the operational reins of its current network and the systems integration of its future CDMA and LTE networks. So the Swedish telecom giant is not only managing its own gear but that of Alcatel-Lucent, Samsung and innumerable other smaller suppliers.
Since SONs will be such a key piece of future networks, those big vendors will need to develop SON architectures that work across the board, if they haven’t done so already. Ericsson acquired Optimi in late 2010. “The question is how long before Intucell is a part of Ericsson or Alcatel-Lucent,” Jarich said.
Why SONs matter
Whoever supplies them, SONs are going to be critical to meet the ever-increasing bandwidth and capacity demands of the mobile data revolution. SONs are a key component to hetnet, and hetnet will provide the means for operators to add gobs of cheap capacity to their networks.
AT&T and Verizon Wireless both claim the capacity apocalypse is coming. While they may be overstating the direness of the situation to further their spectrum consolidation and acquisition goals, they are correct that adding new capacity the old-school way is becoming a much more expensive proposition. Airwaves suitable for mobile broadband are increasingly scarce, and licenses come with hefty price tags. Building macro networks over what new spectrum is available is extremely expensive and produces limited capacity gains.
If we’re going to arrive in a world where mobile broadband is not only ubiquitous but also cheap, we’re going have to stop thinking in terms of doubling or tripling mobile network capacity. Rather, we have to start thinking in terms of increases of 50 and 100 times. And to get there we’re going to need much more complex and nimble networks.