Carrier WiFi Opportunities

Enabling offload, onload and roaming

A Disruptive Analysis thought-leadership paper August 2011

Commissioned by iPass Inc.

Author: Dean Bubley Contact: [email protected]

Disruptive Analysis Ltd, August 2011 Carrier WiFi opportunities and challenges 1

Disruptive Analysis
Contents

Dont Assume

Contents ....................................................................................................................................2 Abstract / Summary ..................................................................................................................3 Introduction ...............................................................................................................................3 WiFi scenarios for mobile operators ......................................................................................5 A quick history of service providers and WiFi Courtesy WiFi as a value-add to mobile broadband Mobile data offload International & in-country data roaming Indoor / not-spot coverage Mobile data onload Local services Other use cases 5 6 6 8 9 9 10 11

Key enablers, standards & innovations .............................................................................. 11 The role of the connection manager client Connectivity via the operator core network 13 14

National & operator-specific WiFi variations ...................................................................... 15 WiFi: The wider operator traffic / policy context ................................................................ 16 When is a seam worth keeping? ...................................................................................... 17 Conclusion ............................................................................................................................. 20 About Disruptive Analysis .................................................................................................... 23

Disruptive Analysis Ltd, August 2011

Carrier WiFi opportunities and challenges

Disruptive Analysis
Abstract / Summary

Dont Assume

Many mobile operators are re-examining their strategies around WiFi (officially Wi-Fi with a hyphen) technology, driven by the explosive growth in smartphone penetration and usage. Handsets are very different to PCs in terms of user expectation and behaviour, and providing them with efficient and intuitive WiFi connectivity demands a rethink. A variety of new technical and commercial models for carrier-based WiFi are emerging, which leverage cellular assets such as roaming and authentication/billing systems, as well as integrating with mobile broadband radio, policy and traffic management functions. These offer the potential for offloading data traffic from congested 3G/4G networks, providing easier access to foreign hotspots for travellers, and driving new revenue opportunities from operators own WiFi footprint. However, getting both the network interoperability and user experience perfect remains a major challenge. People acquire and use smartphones in many different ways, and the WiFi and connection software implementations can vary widely, with new uses for WiFi emerging from sectors like the consumer electronics industry. This means that the operators are dealing with a complex and moving target, and need to be aware of brittle or over-engineered solutions for adopting WiFi. At the same time, there are also numerous other parallel initiatives in operators to deal with the opportunities and challenges of growing data traffic, with which offload must fit. Although much of the current industry rhetoric suggests that connection management should always strive to be seamless and automated, the reality is more complex: sometimes seams are better left neat and elegant, and not entirely invisible. It will be important for all parts of the mobile data ecosystem to work together holistically, rather than risking point solutions creating undesirable side-effects. Successful adoption of WiFi by mobile operators demands careful consideration of technical architecture, a good appreciation of real-world user interaction, and partnering with suppliers that understand WiFis subtle nuances and likely future trends. Overzealous attempts by the mobile community to over-reach and own the entire WiFi value chain risk failure and customer dissatisfaction. Commissioned by WiFi aggregation and roaming specialist iPass Inc., this independent White Paper examines the trends and options for operator WiFi, highlighting both new business models and avoidable pitfalls. The opinions expressed in this document are solely Disruptive Analysis own. Please contact [email protected] for more details, custom advisory workshops or deeper consulting engagements.

Introduction
WiFi technology is now in its second decade of use. Originally driven by the needs of laptops and industrial computing, it is now being rapidly adopted in a range of other devices such as smartphones, tablets and consumer electronics. Some of these notably handsets are strongly coupled to the provision of mobile network services. This is leading to traditional public hotspots taking on a variety of new roles, such as
Disruptive Analysis Ltd, August 2011 Carrier WiFi opportunities and challenges 3

Disruptive Analysis

Dont Assume

offloading traffic from cellular data networks. As a consequence, WiFi is rising up the agenda of many cellular operators, many of which have historically had indifferent or even hostile views towards it. But MNOs (Mobile Network Operators) use of WiFi for smartphones presents a different set of challenges to that traditionally seen on PCs. Firstly, such devices tend to be linked to a specific mobile data plan, which comes with carefully-crafted pricing and usage policies. WiFi access needs to dove-tail with that, in some fashion. Secondly, the connection management experience on a handset tends to be very different, because of both the physical constraints of the device (eg smaller screen) and the need to integrate cellular and WiFi accesses more tightly. (On a PC, the WiFi is usually integral to the chassis and OS of the device, typically used completely independently of a 3G/4G modem, which is usually supplied separately). There are also issues around managing security and authentication in a way which is consistent with operators expectations and obligations. Lastly, smartphone WiFi represents a potential threat for mobile operators it can act as a conduit for third-party applications and content that bypass their control and even visibility. Some use of open or bring your own WiFi on smartphones is inevitable it cannot and should not be blocked. But it is questionable whether operators want to encourage it much further, especially if it risks de-valuing their core, expensive, 3G/4G network assets by making them appear less attractive than free WiFi. This is a very delicate balance to achieve. It is also a very different situation than that experienced by fixed operators. For a typical broadband provider, WiFi adds value to a DSL, fibre or cable connection, by allowing multiple users to access it around the home. Some providers such as BT even seek to differentiate from their rivals through improved WiFi implementation in their gateway boxes, as well as mobilising their users by providing them with out-of-home access at hotspots. Although this might generate more data traffic volume, it is offset by other considerations such as reduced support costs and performance degradation due to poor in-building wiring. WiFi has undoubtedly helped broadband providers increase penetration in laptop-only households which would not consider installing cables at all. Consequently, mobile operators are now seeking to catch up; many are re-assessing their approach to WiFi. A number of important standards are being developed to make WiFi more telco-friendly, especially on phones. Interoperability initiatives run by bodies such as the Wi-Fi Alliance1 and Wireless Broadband Alliance are helping reduce friction experienced by both users and operators in getting connected to WiFi. They are looking to enhance both operators own hotspot networks and their roaming peers, as well as considering methods to improve security, control and convenience, such as routing WiFi traffic back through their core network and policy functions. Even those initiatives are not sufficient, though. While their aims are laudable, they do not address all scenarios from operators perspective especially how to deal with existing hotspots based on infrastructure (and providers) that do not support all the new standards and commercial structures. For some operators, there is an urgent need to offload traffic in the short term they cannot all wait for deployment of new access points
1

Note: the official name for 802.11 WLAN is hyphenated, Wi-Fi, as defined and trademarked by the Wi-Fi Alliance. However, in common with many in the industry, Disruptive Analysis uses the unhyphenated form WiFi

Disruptive Analysis Ltd, August 2011

Carrier WiFi opportunities and challenges

Disruptive Analysis

Dont Assume

(and devices) supporting 802.11u and other such enhancements. For these situations, more pragmatism in needed reutilising existing capabilities such as roaming mechanisms makes a great deal of sense.

WiFi scenarios for mobile operators

A quick history of service providers and WiFi
Before looking at the specific use cases, challenges and solutions for mobile operators around WiFi, it is worth a recap of how WiFi has evolved more generally, and where it has intersected with the service provider domain in the past. Historically, WiFi has had three main use cases: Sharing broadband Internet connections in the home, typically between multiple PCs and more recently, smartphones and tablets. Initially, this was dominated by access points bought by end users themselves and attached to their modems. Nowadays, a large proportion of home WiFi is provided by APs integrated into gateway routers, supplied by the broadband operator as part of the subscription package. Google has over 300m WiFi networks in its location database, the vast majority of which are in homes.2 Public hotspots for nomadic users, operated either by dedicated WISPs (Wireless Internet Service Providers) or provided as a courtesy by a venue owner themselves. There is also a higher-level aggregation tier of service providers operating WiFi roaming or partner networks, spanning multiple WISPs and also sometimes fixed access as well. There are thought to be approximately 1 million commercial hotspots globally, with the number growing fast,3 plus many other informal open access points run by small businesses such as individual cafes. There are also millions of virtual hotspots using shared home broadband connections, which are available to other subscribers of that provider. Corporate usage, either in carpeted offices for visitors and laptop-using employees, or in industrial buildings for connecting computing or specialist equipment. There is usually no direct involvement of service providers.

As can be seen from this, there has been relatively little strategic overlap between mobile operators core business and WiFi in the past. Some operators have owned hotspot networks (eg T-Mobile US), while other fixed/mobile hybrid telcos have offered combined WiFi+3G data plans, especially targeted at enterprise users (eg Swisscom, Orange Business). But none of these offers has really been focused on massmarket mobile subscribers. But now that there are hundreds of millions of WiFi-enabled smartphones, many bought through operator channels, the picture is changing. WiFi is becoming much more critical to MNOs business models, network strategies and broader ambitions. Apart from
2 3

http://mobile.eweek.com/c/a/Mobile-and-Wireless/Google-Engineer-Says-Location-Data-Extremely-Valuable-Report-698405/ WiFi Alliance estimated 750k hotspots in March 2011 http://www.wi-fi.org/news_articles.php?f=media_news&news_id=1048 but since then China Telecom alone has announced adding 400k hotspots to its previous 300k in China this year alone http://www.chinadaily.com.cn/bizchina/2011-05/11/content_12486999.htm .

Disruptive Analysis Ltd, August 2011

Carrier WiFi opportunities and challenges

Disruptive Analysis

Dont Assume

anything relating to offload, end-users increasingly expect to use WiFi on their smartphones, as many have been essentially trained by their experience with laptops. A poor WiFi user experience on a smartphone can be a source of dissatisfaction and maybe even churn. Disruptive Analysis has identified a number of business models for WiFi among wireless carriers: Courtesy WiFi as a value-add Data offload International and in-country roaming Indoor / not-spot coverage Data onload Local services

The issue of whether these services are delivered via the operators own WiFi footprint, via partners networks, or via completely unaffiliated access points, is discussed later in this document. It is also important to recognise that these are WiFi services enabled by the operator. In most cases, the phone will also support the use or WiFi as a function outside of the operators control or influence. In that mode, the device just acts in a similar fashion to a laptop, with the user controlling access to their home WiFi or other privately-run access points. This is a critically important distinction: cellular only operates as a service while WiFi can be both service and an inherent private capability of the phone.

Courtesy WiFi as a value-add to mobile broadband

At the moment, some MNOs offer their customers a certain level of public WiFi access as a part of their data package as a courtesy, or hygiene factor. This is generally less about offloading traffic from congested network, and more about differentiation and the appearance of providing a full bundle of services. However, in many cases this type of capability is used only rarely, and often involves a poor user experience, with clunky log-on procedures and little control/visibility of the actions of the subscriber. Even where operators have partnered with local or international WISPs and aggregators, the footprint of access points has often been comparatively restricted and users have needed to navigate cumbersome screens or deal with lengthy encoded passwords sent via SMS.

Mobile data offload

Over the last three years, mobile broadband usage has rocketed. Most readers will be overly-familiar with the notorious scissor diagram showing exploding traffic growth on 3G/4G networks, only partly offset by improved revenue generation from data plans. Although the specifics of those diagrams and conclusions drawn from them can be debated, what is certainly true is that many operators are keen to manage the load on their main cellular networks and the need for extra capex or spectrum purchases.

Disruptive Analysis Ltd, August 2011

Carrier WiFi opportunities and challenges

Disruptive Analysis

Dont Assume

A key element of the mitigation strategy involves offloading a proportion of data traffic, typically to WiFi but also perhaps to femtocells or other emerging small cell solutions. This has the potential to reduce load in particularly congested cells, or across the network as a whole, and can contribute to improved performance and slowed capacity upgrade expenditure.

Figure: Mobile broadband congestion is a many-dimensioned problem

Source: Disruptive Analysis

Among the more aggressive operators for WiFi offload have been AT&T in the US, PCCW in Hong Kong, KT in South Korea and KDDI in Japan. AT&T acquired Wayport (a major WISP) and has been deploying WiFi in locations such as Times Square in New York, as well as working with Apple to automate connections via the iPhone. KT has been reported4 as offloading as much as 67% of data to WiFi, while KDDI in Japan has announced plans to deploy as many as 100,000 WiFi access points. Numerous offload architectures have been proposed, trialled and deployed here, which are typically optimised for particular use cases and scenarios operator-owned hotspots vs. 3rd parties, in-building vs. outdoors, public venues vs. private homes and offices, laptops vs. smartphones, direct-to-Internet vs. through the operators IP core network and so on. In terms of choosing an offload option, much depends on a given operators starting point its current mobile broadband traffic patterns, its spectrum rights, its existing or potential future WiFi footprint and numerous other variables. A mobile-only operator with 85% of mobile data from laptop dongles is in a very different position to one with many smartphone-centric users whom it also supplies with fixed broadband. An extremely important point to make is that offload it not necessarily just about managing tonnage of data MB and GB of video traffic, for example. For many operators, this is actually less of a problem than many think but what can really hurt
4

http://www.totaltele.com/view.aspx?ID=460384

Disruptive Analysis Ltd, August 2011

Carrier WiFi opportunities and challenges

Disruptive Analysis

Dont Assume

the network is signalling load. This is caused by multiple levels of interaction between devices and network at the radio for power-control from the phones chipset, at the IP level (setting up and tearing down TCP connections and so forth) and various application-layer actions such as SIP signalling for IMS. Each of these can cause problems with unexpected bottlenecks RNCs, GGSNs, security elements and so forth. [See footnote5 for definitions of acronyms] There have been reports of signalling storms caused by data applications actually sapping capacity for delivering voice traffic. WiFi access has the potential to offload some of this damaging signalling load, as well as bulk traffic. This use-case has not yet been fully explored or exploited, in Disruptive Analysis opinion.

International & in-country data roaming

One of the current areas of awkwardness for the mobile industry is the price of mobile data roaming services. While the actual roaming and billing process works smoothly, many prices remain egregiously high sometimes orders of magnitude higher than domestic data costs. It is not unusual for fees to be $10-20 per MB or even higher for international cellular roaming perversely, more expensive than some satellite modem connections, or even local purchase of a new modem and prepaid data plan. WiFi roaming provides a (partial) answer to some of the issues. In markets where in-country national roaming (usually those with regional spectrum licences) is permitted, costs can also be prohibitive. Various industry structural issues conspire to keep the costs high. In particular, many visited operators have little incentive to reduce the punitive wholesale rates charged to the home operators of inbound roamers, which then pass on these charges (with an extra mark-up) to their customers. Where both operators are part of the same group (eg Vodafone or SingTel) or regional roaming consortium (eg Conexus in Asia), there are various lower-cost packages, but this still remains the exception rather than the norm. International bodies such as the European Commission are also taking action, either placing caps on pricing or even insisting on the right of customers to choose separate domestic and roaming providers. Although many travellers already use WiFi regularly when roaming, their domestic mobile operator usually has no direct involvement the user typically sources their own access independently, perhaps from a hotel, airport authority, or a third-party WiFi aggregator. By linking the cellular roaming function to a WiFi exchange, it may be possible to both automate this process and provide a revenue stream for the domestic provider. The net spend may also be higher, as normal operator billing processes can be used in place of credit card sign-up, or charging to a hotel bill although obviously this will depend on whether WiFi access is free or charged for at specific venues. Willingness to enter card details on handsets is much lower than on PCs, while there are often browser/splash-page issues. Furthermore, most travellers will not wish to pay twice, to connect both devices to the network. In some circumstances, using WiFi roaming may also enable home operators to mitigate the wholesale cost issue, by preferentially selecting (cheaper) access providers via WiFi
5

TCP = Transmission Control Protocol; RNC = Radio Network Controller; GGSN = Gateway GPRS Support Node; IMS = IP Multimedia Subsystem; SIP = Session Initiation Protocol

Disruptive Analysis Ltd, August 2011

Carrier WiFi opportunities and challenges

Disruptive Analysis

Dont Assume

wherever possible, rather than defaulting to cellular data roaming on unfriendly networks. As a result, visited networks that also have WiFi assets will need to think carefully about possible net cannibalisation of their inbound roaming revenues. However, for the roaming consortia, adding in WiFi will benefit their customers with better speed/coverage and potentially lower their own operational costs and relieve capacity constraints on visited networks. WiFi roaming is also likely to be more important in future LTE scenarios there is a huge range of frequency bands for 4G, and it is probable that many phones will not be able to support all visited foreign networks, even if the data wholesale price is acceptable. Roaming onto WiFi could act as a common denominator.

Indoor / not-spot coverage

One of the most important rationales for operator WiFi deployment is for improved coverage, especially for data connectivity indoors. Certain locations such as basements are often outside the reach of the macro-cellular network, especially at some of the higher frequencies used for 3G and 4G networks (above 2GHz) which tend to have poor propagation through building materials. While there is a wide diversity of technical indoor-coverage solutions such as femtocells, distributed antennas and other options, many venues will already have WiFi, or else the operator could potentially install its own. WiFi also has applicability for some outdoor coverage scenarios where cellular data cannot reach, such as sports arenas, airport aprons, campus grounds or perhaps small villages. There is also a secondary benefit here, especially in international roaming scenarios as discussed above where a cellphone loses contact with its preferred roaming partner, it will typically auto-select a secondary provider, with attendant wholesale costs for the home operator. For example, if a Vodafone UK user goes to Germany, their phone will automatically prefer Vodafone Germany as a roaming provider. But if there is no coverage, then it will instead opt for whatever else is available T-Mobile, O2 Germany or Base may have coverage instead. In this scenario, it may be possible to force the device to connect to a Vodafone-partnered WiFi network if one is available although that would also raise some questions about how to deal with voice calls.

Mobile data onload

Although the majority of discussion is about using WiFi to offload traffic from the operators macro-cellular network, there is a separate strand emerging with some operators wanting to onload access from customers of other network operators. There are three main categories of WiFi onload that Disruptive Analysis expects to emerge: 1. Operators that have rolled out extensive WiFi hotspot networks will look to resell access on these to visitors from other networks, especially inbound roamers from other countries. This is the reciprocal approach to their own use of foreign hotspots, and is likely to involve collaboration and roamingtype agreements, perhaps via a hub/exchange provider.
Disruptive Analysis Ltd, August 2011 Carrier WiFi opportunities and challenges 9

Disruptive Analysis

Dont Assume

2. The other style of onload is non-collaborative, and is where the operator decides to offer WiFi access, perhaps via an app, to smartphone users in general, without the involvement or agreement of their normal cellular provider. So a customer of Vodafone might have a WiFi app and service provided by Orange or even AT&T or Telstra, on their phone. There are numerous reasons for doing this direct revenue for WiFi access, an advertising footprint directly on a competitors customers phone, or various enhanced service or content offerings. 3. A third form of onload occurs with non-cellular devices, used with public WiFi facilities. An increasing number of travellers carry tablets such as the Apple iPad, which in most cases are sold in WiFi-only form. In future, they may also have WiFi cameras or other gadgets. Offering the ability to hook these into the operators hotspot footprint or those of partners enables such devices to be monetised in addition to normal cellular products, again re-using the operators billing and authentication capabilities.

Local services
Linked to mobile onload, there is growing interest in the prospect of network operators collaborating with specific venues or companies to offer specialised vertical localised services. Service providers such as O2 in the UK have discussed plans to roll out WiFi in locations such as major stores and sports venues. They hope that as well as providing visitors with Internet connectivity for use primarily with smartphones, they can use their wireless footprint and other infrastructure for value-added propositions. For example, retailers are interested in complex omni-channel propositions, in order to reduce the threat from shoppers merely browsing in-store, and ordering online when they get home. Giving away (or even forcing) devices to in-store managed WiFi networks could allow them to do sophisticated couponing or other promotional activities, as well as displaying advertising or product catalogues. Sports venues could offer fans video closeups and replays on their handsets, enable merchandising sales and a whole range of social functions during a game or match. Operators are not the only ones pursuing this type of opportunity some venues may self-manage their WiFi infrastructure, while additional competition from the likes of Google or Apple is plausible as well. Nevertheless, a key component will be around ease of use, and potentially linkage to other communications services, the phone bill or perhaps enablers such as identity-management and customer-profiling which telcos are well-positioned to facilitate. One of the most interesting aspects of this style of approach is that it will likely be offered on an OTT-style (over-the-top) basis i.e. to any mobile user, not just those that are subscribers to that operator. A customer of MNO #1 may download a smartphone app from MNO #2, which may not even be a competitor in the same geographic market. In general, this fits into a major trend that Disruptive Analysis sees occurring the disaggregation of telcos primary access and services businesses.

Disruptive Analysis Ltd, August 2011

Carrier WiFi opportunities and challenges

10

Disruptive Analysis
Other use cases

Dont Assume

As well as the points above, it is also worth mentioning a few other mobile operatorcentric use cases for WiFi: Sale of personal hotspots (such as the Novatel MiFi range and its peers) which use 3G/4G networks as backhaul for a portable access point, connecting up to 5 or so devices via WiFi. Enablement of certain smartphones to act as tethers, again using WiFi to share a single cellular data connection. WiFi-based location tools can complement GPS and cell-based triangulation, especially indoors. These use a database of visible network identifiers to help the phone be pinpointed accurately. Managed services: it has long been an ambition of some enterprise-focused services providers to offer outsourced WiFi infrastructure for businesses. In the past, this has not gone much beyond the occasional provision of guest-access hotspots in office lobbies, but viable models for wider coverage may emerge in future. Various vendors such as Cisco and Ruckus Wireless are targeting sports stadia as having particular potential. In Disruptive Analysis view, such visitorintensive venues are much more conducive to managed WiFi than carpeted offices where it is primarily employee-driven, often with a need access to both wired and wireless LANs.

Key enablers, standards & innovations

WiFi technology has evolved continually throughout its existence, with innovation coming from IEEE 802 committees, the Wi-Fi Alliance and individual vendors of access points, silicon and software for devices and hotspots. More recently, operator-centric bodies such as 3GPP, GSMA and the Wireless Broadband Alliance have also become involved in specifying or influencing its development. Recently, there has been a huge upsurge in WiFi developments geared towards service providers and, especially, mobile operators. Many of these new technologies have been stimulated by the rise of smartphones and the desire to use WiFi for mobile broadband offload or the other use-cases outlined in this report. Smartphones also have specific differences to notebooks, in that they are often supplied by operators themselves, which gives the opportunity for customised software stacks to be included. There are also ways for service providers to push configurations and settings to handsets that are not present in the PC marketplace. It is worth noting that this contrasts with an early operator push towards WiFi around 2005-7, which was more about fixed mobile convergence for voice and other centralised, operator-run services. Previous efforts by operators to exploit WiFi met with limited success, especially using enhancements such as UMA (Universal Mobile Access), VCC (Voice Call Continuity) and I-WLAN (WLAN Interworking). The more recent crop of innovations is much more oriented around data services or plain mobile Internet access. While various operators have enabled WiFi authentication
Disruptive Analysis Ltd, August 2011 Carrier WiFi opportunities and challenges 11

Disruptive Analysis

Dont Assume

processes in the past, these have typically been based on proprietary and localised approaches, sometimes with security flaws. There is now an impetus to standardise although there is still quite a diverse range of organisations involved. It is important to recognise that most of these developments are still in their early stages. Although many or most of them will be important with a 3-5 year horizon, they will take a long time to permeate through the existing installed base of devices and hotspots. In the meantime, interim solutions will still be urgently demanded. Some of the key standards and market initiatives (which in some cases overlap) include 802.11u: At the moment, WiFi access points can only broadcast their names (SSIDs). In some locations, there may be dozens of networks, with no indication to the user as to which is an accessible (or the cheapest / fastest / most-secure) option. 802.11u permits a much richer set of capabilities to be advertised (service discovery), including various authentication mechanisms for easier logon, an indication of commercial terms, and details of affiliations with particular operators or roaming partners. From the handset perspective, 11u-enabled devices can search for particular types of access point. Hotspot 2.0: This is a Wi-Fi Alliance initiative (especially championed by Cisco), aimed at enabling seamless and secure authentication at hotspots. It incorporates a range of technical standards, including 802.11u, WPA2-Enterprise certification and EAP authentication mechanisms. As such, it is purely focused on convenience at the connectivity level, and does not extend to dealing with traffic backhaul, integration with operator core and policy nodes etc. It should be branded as WiFi Certified, probably from mid-2012 onwards, to help users identify suitablyenabled locations. NextGen Hotspot: Closely linked to Hotspot 2.0 and Wi-Fi Certified, NGH is an initiative run by the Wireless Broadband Alliance. Its remit goes somewhat further, and is ultimately being driven towards deeper integration with cellular networks, such as backhauling WiFi traffic through the mobile core. In June 2011, WBA and the Wi-Fi Alliance aligned the two projects more closely. ANDSF: Developed by 3GPP, the Access Network Discovery and Selection Function is part of the new Evolved Packet Core (EPC) specifications for cellular networks. It enables mobile devices to discover and connect with non-3GPP networks (notably WiFi but also WiMAX) and enforce network policy controls through the operators Policy Charging and Rules Function (PCRF). SIPTO: This is a 3GPP standard which is being developed to enable certain data traffic to/from a mobile device to be offloaded (to WiFi, via a femtocell, or in the RAN transport network), while the remainder is taken back via the operators core network. I-WLAN: Originally developed as a way to deliver operator-centric services from an IMS core/application network over WiFi, 3GPP I-WLAN is now a method intended to hook a WiFi-connected device back into an operators core network via a VPN tunnel. However, it does not by itself deal with authentication to the WiFi access point in the first place. OMA DM 1.2: As part of Hotspot 2.0, the latest iteration of handset devicemanagement standards should enable WiFi policy to be downloaded to a phone along with cellular policy. In practice, this should enable it to prefer certain Wi-Fi networks to others.
Carrier WiFi opportunities and challenges 12

Disruptive Analysis Ltd, August 2011

Disruptive Analysis

Dont Assume

Wi-Fi Direct: One of Wi-Fi Alliances ongoing projects, this is a new way for devices to inter-communicate amongst themselves, for example between a phone and a laptop. However, it is not being designed to be operator-controlled, and is seen as somewhat controversial by some as it could potentially allow tethering automatically. WISPr: Standing for Wireless Internet Service Provider roaming, this is a mechanism for WiFi authentication which uses embedded XML objects to allow a handsets browser to access the service without the user needing to enter credentials directly. However, WISPr was not officially standardised, and can be somewhat open to specific interpretation, which has led to variability and inconsistent implementations. A later version, WISPr 2.0 developed by the Wireless Broadband Alliance, has now morphed into the Nextgen Hotspot initiative discussed above. EAP (Extensible Authentication Protocol): This is a family of authentication methods that use additional approaches, beyond the traditional username/password for WiFi access. Two of the most-important are EAP-SIM and EAP-AKA using smartphones SIM cards, but most hotspots will also need to accommodate non-SIM devices such as tablets using certificates via EAP-TLS, plus passwords for walk-up users buying one-time access.

It is also worth noting that various proprietary implementations of WiFi authentication have allowed zero-touch or automated log-ons, for example by using a handsets MAC address as an identifier. AT&T has worked with Apple for its auto-authentication services for iPhones enabled with iOS 3.0 onwards, which allowed direct connection to the attwifi SSID arguably introducing a security vulnerability in the process. There is a strong likelihood that Apple will continue to exercise control over the WiFi connection management experience in future, as it appears to understand its strategic significance. On the network side, there has also been an array of proprietary WiFi management solutions, mostly aimed at carriers wanting to deploy their own hotspot footprint. Ruckus Wireless, NSN, Cisco and others have all developed various gateway and interworking boxes to facilitate security and connectivity, typically using elements of the standards listed above as enablers. Kineto Wireless has also used a revised version of its UMA technology to provide offload capability, especially on Android devices.

The role of the connection manager client

While all the technologies are important, it is critical not to underestimate the role of the main piece of user-facing software, the Connection Manager (CM). This is the application on the device which presents the information to the user about what networks are available, handles any automated logons and related messages, and potentially helps resolve complex scenarios with passwords or applications needing specific access points. The usability and friendliness of the CM will go a long way to determining the user satisfaction and utility from WiFi, irrespective of the cleverness of the underlying standards implemented in the network and phone OS/chipset. In many ways it is also a strategic choke-point, as it will be the arbiter of which sets of configurations are prioritised, and how any conflicts should be resolved. It will present any seams to the user if these are relevant [see later section on seams vs. seamlessness]. In some cases
Disruptive Analysis Ltd, August 2011 Carrier WiFi opportunities and challenges 13

Disruptive Analysis

Dont Assume

the CM is essentially locked-down by the manufacturer or retailer (especially on Apples iPhone), while in other cases it may be defined and remotely-managed by the operator. As well as the CM, there may also be other applications that work collaboratively with it for example a third-party aggregation and authentication client. Depending on the device and OS platform, this software may mask or partially replace the native CM. Some operators such as Telefonica O2 are also exploring the use of such downloadable apps, in order to provide WiFi onload services on an OTT-style basis across any operators phones. In future, other advanced connection management software may aggregate WiFi and 3G/4G connections, bonding them together to maximise throughput, or routing specific traffic through one or the other. ZTE and InterDigital are among vendors working on this style of approach.

Connectivity via the operator core network

At the moment, almost all WiFi data traffic goes directly to the Internet via the infrastructure of the fixed broadband operator providing the backhaul. However, various approaches are intending to drive MNO-based WiFi data back into the operators own network, either right through the core or at least via an intermediate gateway that can enforce policies. Technologies such as SIPTO also play a part here, trying to send only operator-managed application traffic into the core, while offloading other content (eg bulk video files) straight to the Internet instead. In general, Disruptive Analysis is doubtful of the value of most use cases where WiFi traffic is backhauled via the operators core network. It seems likely that the majority of smartphone data traffic will continue to be transmitted to/from the public Internet, and a growing proportion will likely be encrypted using HTTPS or various video-based DRM solutions. That being the case, there is little to gain by funnelling it all through the core (current or evolved) and back out to the Internet again. What traffic is actually destined for operator-controlled services can be selectively tunnelled to the right gateway/server, either from the WiFi access point, the aggregator or an on-device client itself. However, there may be some uses cases where this type of approach makes more sense for example where the operators own VoIP service is being provided over the WiFi connection. It may be possible to define end-to-end QoS through the WiFi infrastructure, so the calls do not have to touch the public Internet with its unpredictable transmission characteristics. A full analysis of this side of operator WiFi architectures and use cases is outside the scope of this document.

Disruptive Analysis Ltd, August 2011

Carrier WiFi opportunities and challenges

14

Disruptive Analysis

Dont Assume

National & operator-specific WiFi variations

It is worth noting that there will be no one size fits all strategy for mobile operators when it comes to developing their WiFi strategy. There are many variables to be considered before any given firm can develop a definitive plan. Among these are: Extent of their own WiFi footprint and their ability to extend this through buildout or acquisition. In developed markets, many venues have longstanding exclusive contracts for WiFi access provision, so operators are almost certainly going to be confined to a select set of venues for their own coverage. Cellular capacity / coverage breadth and depth. If operators currently feel that they have adequate data coverage and headroom for usage growth, they may be less concerned about data offload than others. There is also likely to be variation within operators radio groups and business units when it comes to preferences for adding capacity WiFi, femto/metro-cells, adding LTE in new spectrum, using cellular innovations such as beam-steering and so forth. Spectrum availability and strategy is clearly linked to network capacity which bands are currently licensed or are expected to be offered. Both the total amount made available for mobile broadband, and its quality (ie exactly which frequency ranges) shift the economic and operational considerations and may drive the need for greater exploitation of WiFi in either the short or long terms. Patterns of roaming users, both inbound and outbound, are important determinants. If outbound roamers tend to mostly visit friendly networks within the same group, there is less concern about mitigating excessive wholesale data prices. If they visit operators outside the group, especially in markets with little competition, there will be a greater desire to find local WiFi partners as an alternative. The converse is true when it comes to enabling inbound visitors. The prevailing regulatory regime both domestically and internationally, has a large impact on WiFi strategy. Although WiFi itself is generally unlicensed, there may be country-specific implementation details (eg Internet user registration), or externalities which drive usage (such as roaming price caps). Some locations even outlaw cellular roaming use entirely, for security reasons (eg some parts of India), so WiFi could offer a work-around. The device mix within the operators customer base is critical. Although most smartphones now have WiFi, the implementation of connection management and even on/off WiFi switching can vary widely. Going forward, we will likely see different rates of adoption of the various new standards such as 802.11u in handsets, as well as how any new form of WiFi roaming is presented to the user on-screen. The addition of new device categories such as tablets, M2M and consumer electronics will only further blur the decision-making process for operators WiFi options. Prepay / postpaid mix of data users is remarkably important, with ramifications that extend beyond the mechanics of payment. Often, the model drives the type of handset they use (low vs. high end), and their propensity to buy it through carrier channels if subsidised (which allows custom software and configurations to be loaded more easily). The structure of their data plans will also vary, which may make offload more or less attractive. The prevalence of WiFi in subscribers homes and offices is also highly relevant. An increasingly important element of offload strategies is managing the use of private WiFi as well as hotspots. But a number of home broadband
Carrier WiFi opportunities and challenges 15

Disruptive Analysis Ltd, August 2011

Disruptive Analysis

Dont Assume

operators such as SFR and BT now have a virtual public hotspots built-in to their gateway routers, allowing them to do more general offload in locations outside the footprint of many normal WISPs (eg in suburban streets). Clearly, the extent and type of WiFi implementation will drive both the technical and commercial approach by mobile operators.

WiFi: The wider operator traffic / policy context

Although there are numerous other potential WiFi use cases discussed above, it is the data offload issue that is getting the most attention and which appears to be driving the most deployment and focus. Relieving congested cells, or delaying capex spend on capacity expansion and spectrum/site acquisition is seen as a hard driver of return on investment. It is also a boon to a number of WiFi vendors that are seeing saturation in their core enterprise market expanding to the service provider domain offers new opportunities, especially as many of them do not have to worry about cannibalising a macro-cellular RAN business. But it is important to realise that WiFi offload is just one of numerous options for operators faced with congested 3G/4G networks. It is being assessed and deployed alongside alternatives such as: Capped and tiered data pricing: Many operators have now ditched flat-rates, and have moved to monthly or daily quotas. There is good evidence to suggest that this adjusts user behaviour significantly. Many operators offer a separate WiFi data allowance alongside the 3G/4G quota. Throttling and policy control: Even if customers retain notionally-unlimited data plans, some operators are restricting speeds of access, especially at certain times or beyond certain usage thresholds. Various more fine-grained policy use cases are also emerging, such as dynamic pricing based on busy-cell/busy-hour conditions. Content and application optimisation: A controversial but effective approach to controlling congestion is the compression, trans-rating, re-buffering or other modifications to downloaded data, especially video content. Low-cost capacity enhancements: Some operators are seeking to add in extra cellular capacity at lowered per-unit costs, rather than limit demand. There are various options here, such as shifting to LTE in hotspots, moving to 6-sector cellsites, adding in picocells or other small cells and so forth. There are also various optimisations that can be performed on networks to improve throughput significantly, with relatively minor adjustments of configuration. On-device software clients: There is a wide range of options for operators providing on-handset software and intelligence to mitigate both congestion and data consumption. Ranging from simple fuel gauge meters to sophisticated tools to feed back on network problems, this is likely to dovetail with work around WiFi connection management clients. Working with content and application providers: At present, many mobile services are not network-friendly, consuming more resources than is strictly necessary. Some operators are hoping to charge such third-party providers through two-sided business models, but that runs foul of Net Neutrality laws in many places. Instead, others are collaborating to help these developers optimise
Carrier WiFi opportunities and challenges 16

Disruptive Analysis Ltd, August 2011

Disruptive Analysis

Dont Assume

their software and content. This is a win-win, as it frequently improves application performance and device battery consumption, as well as network load.

Figure: WiFi offload fits into a wider traffic management context

Source: Disruptive Analysis

The net effect of all these options and various others is that offload is only part of the landscape when it comes to mitigating network congestion. The problem is that each element of this toolbox comes with its own policy and management systems, which often get linked together weakly, if at all. If the network spots an HD video stream starting in a busy cell, should it suggest a nearby WiFi venue for offload, alert the user to likely stalling, offer a discount or rebate if the user watches it later, compress the stream to a lower quality or block it all together? We are some considerable way from having a unified or holistic approach to traffic management for mobile data. Operator WiFi will need to fit into this additional world of complexity, both in the network and on the device.

When is a seam worth keeping?

When the cellular community discusses the use of WiFi, if often uses the term seamless to describe authentication and connection of devices without user intervention. Some standards bodies and commentators go further, to suggest that seamless WiFi ought to function almost identically to cellular radio and be strongly coupled to an operators infrastructure, with traffic transiting the same core network and policy infrastructure. Often, the benefits of seamlessness are presented as self-evident truths, without a deeper analysis of whether or not this presumption is warranted. It is simply assumed to be obvious that the end-user will always prefer connectivity to be taken care of
Disruptive Analysis Ltd, August 2011 Carrier WiFi opportunities and challenges 17

Disruptive Analysis

Dont Assume

transparently and invisibly, without needing manual intervention. After all, few people would wish to be manually prompted when a handset hands over from one cell tower to another a stereotypical example of a seamless transition so surely the same is true of WiFi? Disruptive Analysis, on the other hand, has harboured serious misgivings about the notion of seamless handover between cellular and WiFi, since the early days of fixedmobile convergence, when UMA was first proposed around 2004-5 for extension of voice coverage over IP. There are two core areas for concern here: Firstly, borders between technology domains are important. At the boundary between WiFi and 3G/4G, a lot changes: speed, latency, cost, security, mobility, ownership, backhaul, ownership and many other parameters. There are plenty of reasons why notifying users or applications about a transition are important, sometimes requiring confirmation/rejection of the proposed shift. Some applications are bearer-dependent, for example and as discussed elsewhere, there may be cost differentials involved that a user may need to approve. Secondly, WiFi is a special case of a wireless connection mechanism. Because it is based on unlicensed spectrum, WiFi can operate as both a service (eg via a hotspot), or can be owned and used without involvement of a service provider. Reconciling these two use cases requires significant levels of flexibility and control on the part of the user, which go well beyond the rhetoric of seamlessness.

Consider an example: a particular venue has three accessible WiFi access points or SSIDs: An offload-centric SSID from the users operator (or roaming partner), chargeable against the users data plan and subject to policies which restrict the user to specific applications and speeds. An direct Internet SSID, provided by the venue for free, when the user purchases a coffee. This is completely open and unthrottled the user may use any application at whatever speed the access point is able to deliver. An local SSID which enables access to a closed portal (eg for ordering from the menu), and also gives a connection to a printer, a video screen which can be used to display pictures transmitted from the phone, or a WiFi-enabled juke-box for playing music.

In such a circumstance, a device seamlessly connecting to the operator WiFi offload SSID would cause significant problems for the user, especially on the first occasion that venue was visited. Having some form of well-presented and displayed seam, giving the user a choice, would clearly be a much better experience than having to disconnect from the operator WiFi and try to hunt down the other options deep in the phones menu. This is just one scenario there are many others for which the supposed benefits of seamlessness could lead to undesirable outcomes. It is critical to recognise that WiFi was invented primarily and by design as a private networking technology, not the basis of services.
Disruptive Analysis Ltd, August 2011 Carrier WiFi opportunities and challenges 18

Disruptive Analysis

Dont Assume

Figure: Smartphone WiFi includes both private data & operator offload

Source: Disruptive Analysis

The word WiFi is just a brand name used to describe the most widely-used Wireless Local Area Network (WLAN) technology, based on a family of standards from the IEEE, called 802.11. This is important WiFi is, ultimately, a wireless LAN technology, intended for local ethernet-based connectivity. There is nothing inherently service based about this: most fixed-ethernet LANs are owned and run by companies, connecting computers, servers and printers. Much of the original commercial deployment of WiFi was in industrial facilities such as warehouses or large retail stores, where mobile workers used products such as barcode scanners, or movable checkout tills. Importantly, new non-operator use cases for WiFi are currently emerging in parallel to offload and onload, and these will need to be accommodated in the user interface and connection-management elements on various devices. It is common for businesses to manage their own corporate WiFi access infrastructure, home consumer electronics is being enhanced with universal plug and play DLNA capability, and new non-operator innovations such as Wi-Fi Direct and Wi-Fi Display are being standardised through Wi-Fi Alliance. Maintaining appropriate but elegant seams will be necessary to ensure that such capabilities remain accessible. Current initiatives around carrier WiFi often seem to ignore or down-play such use cases. Disruptive Analysis has seen critical references to bring your own WiFi, or pseudo-offload attempts to suggest that private WiFi use cases are somehow inferior to carrier/WISP-controlled offerings. Other efforts such as the use of OMA DM to push WiFi configuration settings to phones could even result in
Disruptive Analysis Ltd, August 2011 Carrier WiFi opportunities and challenges 19

Disruptive Analysis

Dont Assume

devices being effectively SIM-locked to specific SSIDs or just a given operators WiFi access and those of its partners. Disruptive Analysis believes that operators need to internalise the concept of WiFi Neutrality actively blocking or impeding the users choice of hotspot or private WiFi is likely to be as divisive and controversial as blocking particular Internet services.

Conclusion
It is becoming increasingly clear that WiFi access will be a strategic part of mobile operators future network plans. There are multiple use cases, ranging from offloading congested cells, through to reducing overseas roaming costs and innovative in-venue services. As such, it is good to see industry emphasis on ways of improving user convenience, security and business model innovation. The ability for operators to build and monetise their own WiFi footprint more easily is going to become more important, as is their ability to exploit other providers hotspot assets. Operators making substantial investments in their own WiFi networks will wish to offer some capabilities to inbound roamers, just as they do with their cellular networks. The constraints and usage models of smartphones mean that convenience and autoauthentication gain a premium. Conversely, separate WiFi subscriptions or ad-hoc credit card purchases of access work less well on phones than PCs alternatives such as bundling, operator-collected fees (or free access) are preferable. However, it is also critical to recognise three areas of complexity, which will help determine the eventual shape of cellular operators activities in the WiFi arena: Mobile operator-controlled WiFi access will only ever be a subset of a users overall WiFi usage, even on handsets. MNO requirements and architecture need to coexist with those of other stakeholders, such as private in-building WiFi owners, free hotspot providers, fixed operators and device/software firms exploiting WLAN technology directly. This juggling act represents significant challenges, especially in terms of the on-device user experience and ultimate control points in the connection manager software and OS. While many aspects of the new telco-grade WiFi standards offer theoretical benefits 802.11u, ANDSF, NextGen Hotspots and so on many of these will take time to get deployed into both devices and access points. In the meantime, existing WiFi deployment and related usage models will continue growing strongly. There is therefore a need for operators to have access to interim WiFi aggregation and connectivity mechanisms, for example by exploiting current approaches such as WISPr, proprietary platforms such as FON, or tighter integration with cellular roaming mechanisms such as those from iPass or others. Increasingly, users will own multiple WiFi devices, and will have access to multiple WiFi providers in any given location. Operators will need to pay careful attention to arbitrage opportunities and risks, to ensure customer satisfaction. For example, authenticating seamlessly to a charged hotspot, when a free connection or tether is also readily available will provoke anger.
Carrier WiFi opportunities and challenges 20

Disruptive Analysis Ltd, August 2011

Disruptive Analysis

Dont Assume

Over the last decade, Disruptive Analysis has watched multiple efforts by the cellular industry to offer new services and exert control by inventing complex architectures and interfaces in the network. IMS, UMA, mobile video, application ecosystems and so forth have all been designed network-first. Many of these have abdicated any upfront consideration of on-device software and user interface models, which are deemed to be mere implementation details to be handled by so-called terminal vendors and their software suppliers. This is the wrong way around. As Apple and others have illustrated, getting the device software and user-facing experience right is the hard bit, and where much of the overall value resides. Complex though it is, the networks control layers only play a supporting role. Current approaches to operator WiFi are still too network-centric it is the connection manager and the way the handset OSs exploit IP security and access that will take the lead role in smartphone WiFi use. Work around OMA DM for pushing configuration settings to handsets is a step in the right direction, but as noted above, it is essential that there is explicit recognition of multiple use cases coexisting only some of which will be operator-centric. Any attempts overt or covert by the cellular community to own the overall WiFi experience will not only fail, but likely backfire and hand more power to Apple, Google and their peers. In summary, mobile operators should successfully exploit new and easier ways to add WiFi connectivity and a variety of usage modes, whilst still maintaining users expectations of flexibility. The current focus on seamless connection should be recognised as improving convenience, but risks pandering to some market participants zeal for over-control. The same is true of the desire to funnel offloaded WiFi traffic back through cellular core networks and policy infrastructure. Instead, operators need to look for options that strike a balance. Implemented well, with elegant seams instead of no seams, third-party or operator-branded solutions can exploit the app ecosystems, HTML5 browsers, SIMs and roaming capabilities of devices. Coupled with back-end services such as roaming exchanges and hubs, linking both MNOs and traditional WISPs, it should be possible to add significant value through WiFi in the short-to-medium term.

Disruptive Analysis Ltd, August 2011

Carrier WiFi opportunities and challenges

21

Disruptive Analysis
Sponsors note:

Dont Assume

About iPass Inc. and OMX

www.ipass.com

Founded in 1996, iPass manages mobile connectivity for large enterprises and global service providers through mobility and cloud services. With thousands of enterprise customers, iPass is a leading provider of enterprise mobility services which simply, smartly and openly facilitate access from any device on any network, while providing IT with the visibility and control necessary to support the demands of the enterprise workforce. iPass also provides in-country and international WiFi off-load and roaming services robust enough to support both the business and mass market requirements of service providers. Additional information is available at www.iPass.com or on Smarter Connections, the iPass blog. The iPass Open Mobile Exchange (OMX) helps mobile service providers capture new revenues in the rapidly growing Wi-Fi market while dramatically improving the economics of service delivery - enabling them to incorporate unlicensed spectrum and Wi-Fi as an integral part of their mobile broadband strategy, all with speed to market and a reduced cost and risk profile in mind. iPass has uniquely combined a zero 3G-touch mobile data offload solution with a zeroclick Wi-Fi data roaming service, simplifying user connectivity when traversing 3G or 4G networks and Wi-Fi. The end result is that Wi-Fi access is as easy to use as cellular and functions like mobile data roaming, automatically connecting users to preferred networks to create an always best connected experience.

Disruptive Analysis Ltd, August 2011

Carrier WiFi opportunities and challenges

22

Disruptive Analysis
About Disruptive Analysis

Dont Assume

Disruptive Analysis is a technology-focused advisory firm focused on the mobile and wireless industry. Founded by experienced analyst Dean Bubley, it provides critical commentary and consulting support to telecoms/IT vendors, operators, regulators, users, investors and intermediaries. Disruptive Analysis focuses on communications and information technology industry trends, particularly in areas with complex value chains, rapid technical/market evolution, or labyrinthine business relationships. Currently, the company is focusing on mobile broadband, operator business models, smartphones, Internet/operator/vendor ecosystems and the role of governments in next-generation networks. Disruptive Analysis attempts to predict - and validate - the future direction and profit potential of technology markets - based on consideration of many more "angles" than is typical among industry analysts. It takes into account new products and technologies, changing distribution channels, customer trends, investor sentiment and macroeconomic status. Where appropriate, it takes a contrarian stance rather than support consensus or industry momentum. Disruptive Analysis' motto is "Don't Assume". For more detail on Disruptive Analysis publications and consulting / advisory services, please contact [email protected] Website: www.disruptive-analysis.com Twitter: @disruptivedean Blog: disruptivewireless.blogspot.com Quora: Dean-Bubley

Intellectual Property Rights / Disclaimer

Copyright Disruptive Analysis Ltd., 2011. All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior permission of the publisher, Disruptive Analysis Ltd. Every reasonable effort has been made to verify research undertaken during the work on this document. Findings, conclusions and recommendations are based on information gathered in good faith from both primary and secondary sources, whose accuracy it is not always possible to guarantee. Disruptive Analysis Ltd. disclaims all warranties as to the accuracy, completeness or adequacy of such information. As such no liability whatever can be accepted for actions taken based on any information that may subsequently prove to be incorrect. The opinions expressed here are subject to change without notice.

Disruptive Analysis Ltd, August 2011

Carrier WiFi opportunities and challenges

23