Voice over IP (VoIP) or computer telephony integration (CTI) seems finally to be taking off. Enterprises, service providers, vendors and market analysts all agree. Perhaps it's not happening as quickly as early pioneers imagined it would, and certainly much later than pundits originally expected, but enterprises are now buying.

They're driven by the need to cut costs by merging disparate networks, the end of many PBX-based products' life-cycles and growth in the economy, resulting in more greenfield sites for which VoIP is almost a no-brainer.

That's all well and good but it leaves an orphan: mobile voice. Not only do mobile phone calls cost significantly more than landline-based calls, the user of a mobile phone owns at least two devices to manage and pay for. Whether or not that cost is allocated to the IT budget the money has to come from somewhere, so prudent IT managers will want to contain the costs of both hardware and voice calls. Naturally, thoughts turn to wireless infrastructures, the main contenders being Wi-Fi, Bluetooth and DECT.

Which to choose?

Technology choices
Complete concurrence in the IT industry is rare indeed but unanimity appears to be almost complete over the impending death of DECT. This digital cordless phone technology serves its purpose well, but is supported mainly by vendors such as Alcatel, Siemens and Nortel, whose main business comes from traditional circuit-switched carriers. It suggests meagre support for DECT from the mainstream of the communications and networking vendor community.

DECT isn't entirely moribund though. Earlier in 2003, Philips launched IP DECT, an extension of its DECT products to add IP capability and allow DECT calls to be routed over the enterprise network. According to Philips, this "opens the possibility to easily provide wireless connectivity across a multi-site company by just installing IP DECT access points at the necessary sites and connecting them to the company WAN." In its launch materials, Philips went on to point out that this saves on installed wired phones.

DECT is dead?
Philips occupies a rather lonely position though. As Tim Stone, Cisco's European voice group manager, pointed out, the technology has an honourable and useful past: "DECT has been very popular in Europe, and is complementary to PBX. It exists in many organisations such as hospitals and retail outlets." Keen to bury that past, Stone went on: "Cisco sees only one converged network so there's no future for DECT. However, our strategy isn't rip and replace - we do have our call manager which can interoperate with legacy systems."

Stone commented that voice operators are keen to move into the mobile environment by selling wireless products and services. While their enterprise customers need to justify expenditure on new voice equipment in terms of ROI, most find the prospect compelling. However, they possess legacy telephony investments - which by implication includes DECT. Much of this is due to the heavy spend preceding the Y2K panic but, when that hardware life-expires in 2005/6, Stone contended, there will be a bulge in the buying patterns. At that point he sees enterprises moving toward voice over wireless infrastructures.

Bluetooth and Wi-Fi slug it out
If DECT isn't dead already, many proponents of Wi-Fi and Bluetooth would wish it to be. Widely used data standards such as Wi-Fi and Bluetooth offer a better opportunity of reducing the number of wireless networks while enabling calls to cross the data network rather than the PSTN.

A good example of this is BT's latest wheeze, the Bluephone. Under development as a joint venture with T-Mobile and based on the SonyEricsson P900, Bluephone prevents users from making GSM calls using their mobiles when within range of a Bluetooth network. Instead, calls are routed over the data network. At the system's recent launch, BT's John Lee said "for a reasonable initial outlay, [the phones] will be improving the quality of mobile calls, be easy to use and ultimately saving you money. We will be offering significant cost savings against today's mobile spend." More specifically, BT predicts savings of up to 20p per minute, not least because internal calls traversing GSM will be sent over internal networks instead. BT said Bluephone would be launched in 2004.

On the other hand, voice over Wi-Fi vendor Airespace believes that Wi-Fi will take the place of DECT, largely because of Wi-Fi's close integration with data networks. Philips IP DECT system notwithstanding, Airespace's Alan Cohen said, "I think a lot of people in EMEA will move on and forget [DECT]. Wi-Fi is a horizontal enabling technology, like Windows or Ethernet. You have to bet that Wi-Fi is going to win, because it will be cheaper and has momentum." Well, he would say that, wouldn’t he...

Not everyone is pinning their hopes on a single technology though. Inventel sells products based on all three wireless systems and continues to research in all three areas. What's also clear is that Inventel, claiming a 17 per cent share of that market, still sees a future for DECT, as it believes the market will double over the next two to three years.

Technology challenges
But it's not just a matter of picking a technology. There remain technical challenges to be overcome. Neither Wi-Fi nor Bluetooth is ideal since they can't guarantee quality of service (QoS). If there's a spike in data traffic, voice packets will be dropped. SpectraLink, partner to Airespace (and just about every other Wi-fi player) is also working with wireless market leader Symbol to put proprietary solutions in Symbol’s 802.11x handsets that go some way towards solving this problem, according to researcher InStar/MDR.

Security too may be an issue. Although you may be able to encrypt voice traffic in a VPN tunnel, this only highlights the lack of QoS since no system can at the moment distinguish encrypted voice packets inside a VPN. That said, given the short range of the technology, there's little chance of a user not noticing someone shadowing them in order to hoover up unencrypted WLAN packets.

Handover creates problems too as the caller moves between access point cells. At handover time, the user must be authenticated and the data stream re-encrypted, all within 50ms, or the call could be dropped or become noticeably degraded. Few if any existing systems can do this.

On the positive side, indications are though that vendors can surmount at least some of these hurdles with upcoming standard IEEE802.11e, which adds QoS and multimedia support to wireless networks. The standard is currently undergoing revisions, with final approval expected early in 2004.

Issues such as battery life are also less of a problem than they were, according to Stone, pointing out that Cisco's latest wireless IP phone offers 21 hours standby and 3.5 hours talktime.

Conclusions
So passing voice over wireless data networks is not trivial and there's a lot of work still to be done. Despite the reservations of observers who suggest that IEEE802.11e won't scale well for enterprises, indications are that many of the hurdles will be surmounted over the next year.

And our prediction is that DECT will probably be around longer than its detractors expect, since experience tells us that technologies that fill a need don't simply disappear just because something else comes along.

Whatever happens, enterprises will select mobile wireless solutions according to a huge and complex range of criteria, including usage patterns, the upfront and running costs for their particular business, and the relationships they hold with suppliers and integrators.

Just like any other technology really.