10 February 2009

Voice-Over-Wi-Fi Options for the All-Wireless Workplace

Pervasive WLANs can and will be leveraged to remove dependence on wiring to desktop phones, and to enable mobility inside the enterprise for employees who are frequently away from their desks. Just a there exist different classes of enterprises and users, so too are there different voice over Wi-Fi solutions - no one solution will be dominant across the enterprise.

A recent article by Michael Finneran in the No Jitter blog provides the catalyst to explore these solutions in greater depth. Finneran suggests that UMA, a carrier-based FMC architecture, is not suited for the enterprise, in part because it has no integration with the PBX or UC network. We agree that UMA is not suitable for all employees, but there is certainly a segment of users in most enterprises that would benefit from UMA, and indeed we believe certain types of organization could profitably convert to UMA en masse.

The “right” form of FMC depends on how the telephony service is used. If the objective is to support users who roam within a building, single-mode Wi-Fi phones can do the job. Phones from Polycom (formerly Spectralink), Cisco, and others that were been developed for distinct vertical markets (retail, manufacturing, healthcare) are increasingly finding their way into schools, universities and even mainstream enterprises. Most of these phones support WPA2/802.1X security, and because they are designed for an enterprise environment their inter-access point handover performance and PBX integration is good.

If the objective is to support users on the move within and outside buildings, the right solution depends on the current PBX infrastructure, the user’s needs, and the budget. One option is to simply use a cellular phone whenever there is a strong enough signal. While not usually cost-effective, and lacking any form of PBX integration, it may be sufficient for some users. The other options involve some form of FMC.

The simplest, most comprehensive and most widely-used form of FMC today is UMA. There are perhaps 2 million UMA phones in use, compared to perhaps 2,000 – 20,000 for alternative, PBX-anchored FMC solutions. The UMA phone automatically switches to Wi-Fi when it detects good reception from a suitable access point, returning to cellular when it loses a usable Wi-Fi signal. When on Wi-Fi, the phone sets up an IPSec tunnel over the Internet to a gateway at the cellular carrier’s site: all signalling and media traffic is carried through this tunnel, so it never interacts with the enterprise WLAN or PBX.

UMA is simple, it works well, a reasonable range of handsets includes Nokia, Windows Mobile from various vendors and BlackBerries, and it meets the basic needs of the outside-the-building employee. Build-it-yourself Wi-Fi coverage, a single phone number, good cost savings are available with no behavioral changes on the part of the user. UMA does not integrate with the PBX, but that’s the reverse side of simplicity. At just the incremental cost of the phone, plus an optional $10/month, UMA is certainly the low-cost FMC solution. For the many employees who today use their cellphone for nearly every call, UMA is a good answer.

The second indoor-outdoor solution is a dual-mode phone operating in ‘two-number’ mode, i.e., a Nokia E-series such as an E71 or a Windows Mobile smartphone. The VoIP capabilities of these phones (particularly Nokia) are extensive. The E71 includes a full SIP stack on top of a Wi-Fi interface with the best inter-access point handover performance available today. It can be integrated with any SIP-capable PBX over an enterprise WLAN using the most stringent Wi-Fi security, WPA2/802.1X.

Very little behavioral adaptation is required. When someone calls your cellular number, the phone rings. When someone calls your PBX number, your phone rings when you are in WLAN coverage, otherwise it goes to enterprise voicemail. Most modern voicemail systems with an email notification feature will then push the voice message to the phone over the cellular network. Outgoing calls can be directed via the PBX when Wi-Fi is connected, or revert to cellular. The user gets one device that works as a cellphone + PBX phone inside the enterprise, and as a smartphone with e-mail when roaming away from the enterprise.

The shortcomings of the ‘two-number’ solution are that there is no single-number option (is that too obvious?), it requires a SIP-capable PBX, and it often requires a per-extension SIP license fee from the PBX vendor. And because there’s no handover between Wi-Fi and cellular, if you walk out of the building during a call, the connection is going to drop. But assuming that the PBX and WLAN are ready, this solution can be yours for the price of the phone, with no additional monthly charges. If users are prepared to give up their desk sets and switch to an E71 or equivalent, this option can be very cost effective.

As we have rolled-out an all-wireless workplace within Aruba, we have found that some employees prefer to use a desk phone while working in the office, but a cell phone-like device while roaming. A desk phone provides a large, easily-read display of calling name/number, one-finger dialing won’t push the phone across the desk, the handset can be cradled on the shoulder, and there’s typically a speakerphone function with good acoustics. All of these are problematic with a cell phone form factor…unless manufacturers release more docking stations with a corded handset or headset. While we introduced a number of Aruba employees to softphone applications on their PCs, most still prefer a separate device, either in deskphone or mobile phone form. Even so, we would acknowledge that softphones in general appear to be enjoying some popularity in enterprise settings.

All Aruba employees are given a Remote Access Point (RAP) device for use at home, providing direct access to the corporate servers and PBX without requiring any client software. Effectively this provides inside-the-firewall access when WPA2/802.1X is used. The RAP can be used with a dual-mode phone, or with a wired deskphone on the second Ethernet port to get PBX dial-tone while at home. This solution has proven exceptionally useful to employees and addresses 99% of the needs of remote workers for voice and data communications. It is possible to combine this approach with other FMC solutions such as extension-to-cellular to access PBX features when cellular-only coverage is available, and with full-fledged single-number solutions.

Full FMC solutions Agito, DiVitas, and others provide comprehensive single-number reach with cellular/Wi-Fi handover. But their penetration of enterprises may be gradual. Firstly, their solutions require a software client on the phone. They also require a data centre server that communicates with the PBX, to allow simultaneous call legs over Wi-Fi and cellular, with coordinated switching for handover. Further, these solutions are expensive, ranging from $100 - $400 a seat, and CIOs will have to consider how much the soft-dollar productivity benefits worth in hard cash, and if this level of functionality is considered business-critical when less expensive, slightly less capable alternatives are available?

These issues are typical of early-stage technology. While they will be addressed over time as customers field systems and report their experiences, we see them as significant obstacles to market acceptance in the near future.

When FMC in any or all its forms can deliver benefits that exceed its costs, it will succeed in penetrating the enterprise. We believe that no one FMC architecture is suitable for all applications: rather the right architecture must be selected from a palette of choices, and it is incumbent upon the wireless LAN vendor to offer a canvas on which the different colors of FMC can be rendered. This is a daunting technical challenge which only Aruba, Cisco, and possibly Motorola could hope to undertake.

In this note we have covered the range of voice over Wi-Fi and FMC solutions available in conjunction with enterprise WLANs. These range from wireless deskphones and softphones through single-mode and dual-mode cellular/Wi-Fi devices. The latter can bridge the cellular and PBX worlds in various modes, ranging from single-number cellular-only (UMA) to the converged device (but separate network) two-number solution to full seamless single-number handover with a comprehensive FMC solution.

One of the advantages of UMA in particular its simplicity. It might be a 70% solution, but the remaining 30% is so fraught with complexity that it sometimes looks ideal by comparison. We believe many enterprise users and CIOs are ready to embrace an inexpensive single-number solution even if it is not integrated with the PBX.

Pervasive wireless enterprise phone systems are inevitable: the business case for rightsizing out wired in favor or wireless solutions is too compelling. From the environmental aspect alone, eliminating wired phone cable drops, plastic wall plates, and wired desk sets represents a major opportunity to reduce an enterprise’s carbon footprint while reaping considerable monetary savings.

The means to that end is less clear. We believe there will be no “big-bang” solution to FMC, a single creation writ whole in one fell swoop. Enterprises will eventually reach the goal by taking small steps and achieving incremental improvements. UMA looks like a promising next step on the stairway - the riser is short and manageable because of the availability of robust adaptive wireless LANs, the tread deep and wide because of the prevalence of suitable handsets. The open question is how big a stretch is necessary to reach the next tread.