Looking at analysts' projections, we see that Wi-Fi based Wireless Local Area Networks (WLANs) are becoming widespread and that Voice over IP (VoIP) is destined to displace conventional circuit switched technology. Combining these predictions, we might conclude that Voice over Wi-Fi (also termed VoWi-Fi, VoWiFi, VoWLAN or VoFi) is inevitable. Another argument from general principles is that cell phone designers are on a feature inflation treadmill; a Wi-Fi radio is just another soon-to-be mandatory checklist feature, like a camera and Bluetooth. And when every cell phone is dual-mode, (Wi-Fi + Cellular), why not use the Wi-Fi for voice? Looking at vendors’ plans, ABI Research in September 2006 predicted that dual-mode phone shipments will reach 300 million units worldwide in 2011. On the other hand, how inevitable can VoWi-Fi be unless the technology is ripe? And even then, might it not be just another technological solution looking for a problem? Might the main issues that it aims to resolve relative to cellular service, high cost and poor coverage, be solved by alternative means, such as flat-rate voice over 3G, possibly aided by femtocell technology?
There are several flavors of VoWi-Fi, so the answers are contextual. Let’s make a rough division of the market to define four segments in a two-by-two matrix, Consumer vs. Business customers on one axis, and Dual-mode vs. Wi-Fi Only on the other axis. For this discussion, let’s also assume that the call path is VoIP end-to-end, not only on the WLAN but also on the access link.
Several factors bear on whether a VoWi-Fi implementation will perform adequately: the bandwidth of the access point (802.11a/b/g/n) the number of phones per Wi-Fi access point, whether or not the access point is configured for QoS (802.1e and proprietary mechanisms), the need to hand off from access point to access point and from network to network (802.11r, 802.11u, 802.21), the amount of interference in the radio environment, the battery life of the phones, the usability of the phones, and of course manageability and security requirements. Each of the four quadrants of our market matrix demands a different mix of these.
The following paragraphs go through the four quadrants of the matrix: Wi-Fi only phones for Consumers, Wi-Fi only phones for business customers, dual-mode phones for consumers, dual-mode phones for business customers.
Wi-Fi Only/Consumer
A Wi-Fi phone call requires two Wi-Fi devices: the Wi-Fi access point and the Wi-Fi telephone handset. The phone must associate with the access point, and they must interoperate according to the relevant parts of the 802.11 standard, notably 11e for QoS and 11i for security. If the home requires more than one access point for complete coverage, the phone must be able to hand off between them without losing a call in progress.
In a residential setting with a consumer broadband router access point and one or two Skype or SIP (Session Initiation Protocol) phones, call quality can be surprisingly good, considering the potential for impairments. With any Wi-Fi connection there is a possibility of interference from microwave ovens and other ill-behaved RF (Radio Frequency) emitters. If you have a large house or an old house with lots of metal in the walls (like steel mesh under plaster) you may have areas with weak or no coverage. If you happen to live in a densely populated space like an apartment building, your neighbors' Wi-Fi networks can interfere with yours. If your network is 802.11b/g, a single b-device with a weak signal can slow the entire wireless network to a crawl. Supposing none of these impairments affects you, you can still suffer from bandwidth contention from your own web browsing and video streaming on other devices around the home. This bandwidth contention applies to both the residential bandwidth bottlenecks: the wireless LAN and the broadband access link.
The good news is that every one of these problems can be mitigated.
Unfortunately most Wi-Fi phones and access points still depend on their users to configure their technical details. It is absurd to expect consumers to gain these skills; they rightly expect their electronic equipment to be self configuring and easy to use. Some service providers view this as an opportunity, but it isn’t clear that consumers are prepared to pay for a home network management service no matter how much they need it, and it isn’t clear that service providers can provide this level of support cost effectively. So to make this market happen, equipment vendors must strive to make their equipment self-configuring, or at least easier to configure.
Phone replacement is only one of the consumer uses for Voice over Wi-Fi. Consumers use VoIP on their PCs and game consoles, and on new types of device like the Sony Mylo. This means that only a small proportion of current consumer voice traffic over Wi-Fi traffic goes through Wi-Fi phones.
Wi-Fi Only/Corporate
Business phones must offer PBX features, and present a user interface with full PBX functionality. This is a major difference between consumer phones and business phones, and one of the reasons that corporate IT departments are likely to prefer a SIP (Session Initiation Protocol) based phone to a Skype one. Spectralink and Symbol pioneered the corporate Wi-Fi phone category. Symbol abandoned it, but now several other manufacturers have offerings.
Corporations are always concerned about manageability and security, and VoWi-Fi presents challenges in both these areas. Although 802.11i has gone a long way to fixing authentication and encryption issues, ABI Research maintains that corporations may still view VoWi-Fi as vulnerable.
Corporations have very different Wi-Fi infrastructures than consumers. They are usually based on WLAN switch/thin access point architectures. These are professionally installed with attention to RF engineering. The result is that while consumer grade systems max out at around half a dozen or so VoIP sessions per AP, companies like Meru, Trapeze and Aruba claim as many as 30 concurrent voice sessions per access point while maintaining QoS. Beyond this, some vendors, notably Cisco (with CCX) offer client-side optimizations to Wi-Fi to further improve VoIP performance.
Dual-mode phones
Both corporate and consumer SIP phones require effort to get working, though interoperability is improving; from this point of view they can't yet be described as mainstream devices. But even if setup and roaming were effortless, Wi-Fi-only phones suffer from a fundamental problem: they don't work everywhere, so you end up still needing two phones, your Wi-Fi phone and your cell phone. And cellular service is so cheap and ubiquitous these days that people may not be motivated to carry a Wi-Fi phone in addition to the cell phone already in their pocket. These factors indicate that Wi-Fi only phones may remain niche products. So what about the other column in the matrix, the dual-mode phone column?
The appeal of the dual-mode phone is that you only need one phone for both cellular and Wi-Fi voice service, and it is always with you.
Adding a Wi-Fi radio to a cell phone brings the benefit of high-speed, low cost data connectivity; that sounds ideal for VoIP, but even assuming that the technical challenges have been addressed, it isn’t so easy for the user. The first issue is whether you get your VoIP service from your cellular provider or a third party. The main benefit of getting it from your cellular provider is ease of deployment; the main benefits of getting it independently are first that you don’t have to wait for your cellular provider to offer the service, and second it may be cheaper. This is the model suggested by the UK Wi-Fi VoIP service provider Mobiboo for the dual-mode phone that they brand.
There are plenty of smartphones already available with built-in Wi-Fi, where the VoIP functionality is fully partitioned from the cellular functionality, each having its own service provider, phone number and user interface. The user decides which one to use on a call-by-call basis and calls can’t be moved between them. This is how the Nokia Eseries (business oriented) and Nseries (consumer oriented) phones work.
If you want to get one of these dual-mode phones and run something like Skype on it, there are issues. First, the phone must be open enough for you to install a VoIP soft phone. Even if the phone’s operating system is intrinsically open (like Symbian, Windows Pocket PC or OSX) the openness may be disabled by the manufacturer of the phone, or by the mobile operator that distributes it. And even if you succeed in loading a soft phone onto the handset, it may work poorly for VoIP if the Wi-Fi implementation is tuned for data and battery life rather than voice traffic. D-Link has optimized their V-Click phones for this usage model.
The next level of sophistication is to have a single phone number for Wi-Fi and cellular calls. For this an additional piece of network equipment is needed to “anchor” the call as you move from network to network. This device is generically called a mobility controller, or in IMS (IP Multimedia Subsystem) parlance, a Call Continuity Control Function (CCCF). The mobility controller can be located in the service provider’s network or in the customer’s LAN. Most of the implementations with the mobility controller in the service provider’s network are for the consumer market. The business market presents the additional challenge that the phones must have PBX functionality, so business-oriented implementations generally locate the mobility controller on the customer premises. An exception is a company called Sotto which combines VoIP Centrex technology with cellular service. Sotto is currently trialing their service in Seattle and Charlotte.
Dual-mode/Consumer
The single phone number version of dual-mode service is often called Fixed/Mobile Convergence (FMC), because the Wi-Fi call connects to the public telephone network over fixed broadband access. In-Stat in 2006 said “Interest in FMC services is high across all broadband household groups.”
Cellular service providers like FMC because Voice over Wi-Fi does three things for them. First, Wi-Fi provides coverage in residential environments where cellular coverage is often weak. Second, the cellular provider gets this improved coverage relatively cheaply, since they don’t have to build out their wireless network. Third, this additional coverage offloads the cellular network, freeing capacity for more subscribers. Cellular service providers can get the same benefits from femtocells, which are like Wi-Fi access points except they run cellular protocols at cellular frequencies. If these become as cheap as Wi-Fi access points, some cellular providers may prefer to sell them instead of dual-mode phones.
Wireline phone service providers (telcos) like FMC because they can use it to mitigate Fixed Mobile Substitution (FMS), the increasingly common phenomenon of consumers abandoning their home phone lines and using their cell-phone as their only phone. Their play is to retain the billing relationship with the customer by offering cellular service in competition with the mobile operators. Since they generally don’t have wireless networks, they must resell wireless minutes from mobile operators (becoming Mobile Virtual Network Operators, or MVNOs). Obviously the margins on this must be razor-thin, but a large proportion, (over 50% by some estimates), of cellular calls is made from home or work. Because FMC runs these calls over the customer’s existing all-you-can-eat broadband network, the traffic is effectively free to the FMC service provider. This dilutes the amount that a dual-mode MVNO has to pay to their mobile wholesaler, and potentially greatly improves their margins.
FMS isn’t the only trend that is causing traditional wireline phone service to decline. Telcos are also losing customers to VoIP service over broadband data access networks, offered both by pure-play VoIP service providers like Vonage and Skype, and by Internet Service Providers (ISPs) like the cable companies (sometimes called Multi-Service Operators, or MSOs). FMC appeals to these MSOs for the same reasons it appeals to telcos: it lowers their costs when they become MVNOs in order to offer quad-play services: voice, internet, TV and cellular on a single bill.
It is just becoming possible to get FMC service in some parts of the world. In England, BT has announced a Wi-Fi flavor of its Fusion service. The USA is close behind, with T-Mobile currently running a trial in the Seattle area, though they haven’t said when it will be more widely deployed. Both these deployments are based on a technology called UMA, or Unlicensed Mobile Access, pioneered by a company called Kineto. UMA goes the whole way, offering cellular to Wi-Fi handoffs (and vice versa) in mid-call. UMA has been adopted by the 3GPP under the name Generic Access to Networks (GAN), and it might better be described as "GSM over IP" than Voice over IP, and it lacks many of the advantages of VoIP, which is more versatile, more extensible and allows a wide variety of codecs for superior call quality.
UMA is one of several alternative technologies for controlling the cellular-to-Wi-Fi handoff; the other main contender is IMS/VCC (Voice Call Continuity). Early service provider deployments use UMA, but because IMS is SIP-based it is expected ultimately to become more widely deployed.
Regardless of how dual-mode is done, FMC-style seamless handoff between networks may be overkill for now. There are other usage models intermediate between the fully partitioned model and the perfectly integrated FMC one, for example a unifying application on the phone that wraps cellular and VoIP service from two different providers into a consistent user experience.
The major benefit of dual-mode phones is to let you use your cell phone as your only phone. FMS claims the same benefit without the need for Wi-Fi. Dual-mode additionally claims to reduce costs and improve sound quality on calls made within range of a Wi-Fi network. FMS may be able to offer these benefits through network upgrades, possibly aided by femtocells. But since no cellular provider offers network-based PBX functionality (IP Centrex), FMS is not currently a challenge to dual-mode for the business market.
Dual-mode/Corporate
The concept of the corporate dual-mode phone predates network FMC. Motorola, Proxim and Avaya founded the SCCAN forum in 2004, and several vendors have since come out with offerings along the same lines. In this model the mobility controller is located by the PBX rather than in the service provider network. This approach is parallel to using a corporate PBX rather than Centrex service, something that corporate IT departments continue to prefer. This preference is reflected in the distribution channel for telecom equipment to businesses, which is substantially independent of the service providers. Nokia has recognized this as an opportunity with their ESeries dual-mode phones, which are targeted at the traditional PBX distribution channels rather than the mobile service providers, and has announced an agreement with OnRelay to integrate cell phones into the corporate PBX. Other vendors to watch here include RIM, which recently purchased a mobility controller startup called Ascendent, and Microsoft, which in June of 2006 announced an alliance with Motorola to combine Office Communications Server with Motorola’s Mobility Suite.
Wi-Fi Hotspots
Once we have accepted the idea of a Wi-Fi handset, it is natural to consider how to use Wi-Fi instead of a cellular service when we are out and about. This is far more challenging that using it for fixed service for several reasons.
VoWi-Fi at hotspots shares all the impediments of home and work VoWi-Fi. For example, in November of 2006 a company called Airmagnet checked 15 hotspots in central London, and found only 4 of them to be VoIP-ready. In addition to sharing all these difficulties, hotspots add some new obstacles. First, most hotspots are not free, so you need an account with the hotspot operator, or with an aggregator like Boingo or iPass, or with a provider that has a roaming agreement with the hotspot operator (for example, the Wireless Broadband Alliance is an international group of telcos with mutual roaming agreements that cover almost 60,000 hotspots). Since none of the aggregators has complete (or even 50%) coverage of all the hotspots in the world, you will be unable to connect at a good proportion of the hotspots you find. Of course you can get around this by tracking down a location that you know your provider supports, like Starbucks if you have T-Mobile or McDonalds if you subscribe to Wayport.
Supposing you find yourself at a hotspot that you are authorized to access, you then have to log on. If your phone supports web browsing, this is possible, though not always easy. The Wireless Broadband Alliance website has an amusing video clip showing their six-step web-based login process. If your phone doesn’t have a browser, you will be out of luck at most hotspots. Fortunately this is changing. Skype is working with phone manufacturers and hotspot providers to offer zero-touch authentication. But it’s patchy: Skype authenticates onto Boingo hotspots, but currently only with a phone from Belkin; they authenticate onto FON hotspots, but currently only with a phone from SMC; Skype on SMC phones will also authenticate onto TheCloud hotspots; TheCloud also has an agreement with another VoIP provider, Truphone, which can currently authenticate only some Nokia phone models. Devicescape has a promising light at the end of this tunnel in the form of a universal client-side application that handles web-based log-ins in the background, even if your phone doesn’t have a browser. So far the only browser-less phone that they claim support for is the Linksys WIP300.
So coverage for Wi-Fi outside your home and office is likely to be extremely spotty, unless you happen to live in a city with metro Wi-Fi. Metro Wi-Fi adds an interesting dimension. Conceivably it could be implemented so that Wi-Fi service within the coverage area competed with cellular service, even while driving around – not yet, but when 802.11r bears fruit.
Conclusion
The market is ripe for VoWi-Fi, but there remain innumerable factors that could cause it to stall, including substantial technical challenges. The equipment vendors, system builders and service providers who solve these challenges will reap substantial rewards.