Wi-Fi Offload: Build Your Own Wi-Fi Network or Share It?
By Monica Paolini, Senza Fili Consulting
For mobile operators fighting network congestion, Wi-Fi offloading has been an unqualified success. In some high-traffic locations in Hong Kong, up to 80 percent of the cellular data traffic is offloaded. In South Korea, SK Telecom has 62,000 hotspots which are actively used for Wi-Fi offload of cellular traffic. The operator plans to add 45,000 more locations during 2011.
In the U.S., the Wi-Fi offload strongest proponent is AT&T (NYSE:T), with 25,000 Wi-Fi locations. AT&T was the first operator in the U.S. to see the sudden increase in data traffic in urban areas where adoption of the iPhone was the highest (most notably New York and San Francisco) and, as a result, was the first that had to address systemic network congestion--not just the occasional congestion that may be driven by a specific event.
An increasing number of mobile operators are looking at Wi-Fi offload to relieve network congestion, keep costs under control, and improve the subscriber experience. As in many public locations Wi-Fi access is free and most households and office locations with smartphone users have Wi-Fi, offloading cellular traffic to these networks is a low-hanging fruit. The marginal cost to both operator and subscriber is zero (there is a cost to the operator that backhauls the Wi-Fi traffic, but so far Wi-Fi offload represents a small fraction of the entire wireline network traffic, so this has not yet surfaced as a major issue). All the operator needs to do is to nudge the subscriber towards Wi-Fi--not a difficult task, since most subscribers know performance over Wi-Fi is likely to be better than over 3G.
The other type of Wi-Fi offload--the one championed by SK Telecom and AT&T--is fundamentally an extension of the operators radio access network (RAN) to include hotspots directly managed by the operator. Wi-Fi hotspots are not designed to capture most of the network-wide traffic, but rather to reduce the traffic load in areas where the cellular infrastructure does not have enough capacity. As such, operators can choose to deploy hotspots where (and when) they see congestion in their network.
While congested areas typically represent a small percentage of a nation-wide cellular network, they are where many of the subscribers concentrate and where the need (and value) of mobile data access is highest. At these locations, the Wi-Fi infrastructure can carry the "low-margin bits," as Jeff Thompson, CEO of Towerstream, calls them--the traffic from bandwidth-intensive applications like YouTube that generates limited revenues to operators, but that has the potential to disrupt higher-margin services such as voice over the macro cellular network.
Mobile operators that decide to offload to a Wi-Fi hotspot network specifically built for this purpose have so far preferred to build and control their own networks. But is this the most effective--in terms of cost, performance, and time-to-market--route to Wi-Fi offload? Or should operators have some other company build them and operate the Wi-Fi infrastructure on their behalf? And if doing so, would it make sense to share the network with competing mobile operators--not on a roaming basis, but sharing the infrastructure with them?
The obvious advantage in building and operating a network is that the operator has complete control of it. As Wi-Fi uses license-exempt spectrum that is inherently shared with whoever decides to use it at any given location, mobile operators cannot exert the same level of control they have over the cellular RAN. Furthermore, the Wi-Fi infrastructure is typically not integrated within the broader cellular network, and operators have little visibility into what the subscribers experience is when they use Wi-Fi or to manage the Wi-Fi traffic as they do (or can) with the cellular traffic.
The depth of expertise among mobile operators resides on large and complex macro networks, not in Wi-Fi hotspots. Macro base stations are expensive and require lots of expertise to deploy and, in most cases, dedicated real-estate assets (i.e., towers or roof-top locations). Wi-Fi equipment is cheaper, faster, and easier to install, but the number of installations is higher and getting access rights and backhaul to them may require substantial effort and money.
The inherent lack of control over Wi-Fi networks and the fact that they are not developed or operated as an integral part of the cellular networks makes Wi-Fi offloading an ideal candidate for a wholesale or infrastructure-sharing model, in which one mobile operator or third-party operator builds and operates a Wi-Fi network that can be shared among operators, or in which one Wi-Fi operator builds and runs the networks for a mobile operator.
Although mobile operators cannot use the Wi-Fi network as a differentiator from other operators if they share the same network, they gain access to a much wider network with higher capacity than they could if they built one on their own at a comparable investment level, and do not have to risk to get distracted by the Wi-Fi network as they try to upgrade 3G or deploy LTE.
To date, the success of infrastructure-sharing initiatives has been very limited in the U.S., where operators do not seem to be under sufficient pressure to relinquish total control over their networks (although Sprint Nextel's (NYSE:S) plans to develop 4G network in partnership with LightSquared and Clearwire (NASDAQ:CLWR) seem to indicate a more welcoming attitude towards infrastructure sharing). This makes a third-party network for Wi-Fi offload a risky proposition, but nevertheless one that may gain acceptance because the need of operators to act quickly to increase network capacity in high traffic locations as their resources are more effectively aimed at their 4G networks.
Towerstream has launched an ambitious project to build a near-carrier-grade Wi-Fi network (to the extent that this is possible using license-exempt spectrum) that mobile operators can use for traffic offload. They decided to build the first network in Manhattan (San Francisco is next), where they have 1,000 Wi-Fi locations, with a plan to grow to 1,500. The next 500 locations will be chosen on the basis of "heat maps" generated by social networking traffic, which is expected to be correlated with mobile data traffic. The Manhattan network is currently in a trial stage, but we should expect tenants in a few months' time.
It will be interesting to see how this project develops and what type of acceptance it will gain among operators. There have been attempts at building Wi-Fi wholesale networks in the past. In fact, AT&T's Wi-Fi network is mostly the result of the acquisition of 20,000 hotspots from Wayport and the Starbucks network (previously operated by T-Mobile USA) in 2008. Before 2008, Wayport was the wholesale provider to AT&T. However the Wayport network was not built for Wi-Fi offload, but rather around partnership with companies like McDonald's which facilitated access to a nationwide network of well-recognized locations.
Towerstream approach and core advantages are distinctively different. The Towerstream network is optimized for offload and covers mostly outdoor locations, which pose a very different set of challenges than indoor locations such as Starbucks. Not only equipment has to be protected differently if installed outdoors, but power and backhaul connectivity may not be available or too expensive. Here Towerstream advantage is that it has access to backhaul connectivity (Towerstream is an enterprise wireless service provider in 12 cities in the U.S.) and has the experience in gaining access to a wide-range of outdoor mounting assets.
As real estate access and backhaul are crucial to build the networks that mobile operators need for offload, cable operators are also well positioned to address this market as they can use strand-mounted equipment installed on assets they control (no need to get deals with Starbucks) and use their own network for backhaul.
The success of a third-party business model can greatly expand the potential for Wi-Fi offload to address congestion in cellular networks. But intriguingly it can also offer a very good model for small cell deployments, where very similar real estate and backhaul challenges exist and are tempering mobile operators' enthusiasm for emerging heterogeneous topologies.