VOIP on WLANs
There's been a lot of talk about voice over IP (VOIP) on wireless LANs (WLANs), but it seems that the term "Wi-Fi" is really taking over lately, despite the fact that it's someone's trademark. For those of you unfamiliar with the term, VOIP involves implementing digital voice telephony over an Internet Protocol (IP) connection, typically a LAN (sometimes with Internet connectivity as well, such as the services offered by Vonage and Net2Phone), or the interconnection of LANs carrying voice traffic via WANs. Connections can be point-to-point and all-IP through a network, or gateways can be used to convert between VOIP signals and traditional telephone connections. Many believe that VOIP will eventually constitute all telephony, but if it happens, it is many years away. There are a lot of copper loops in the world, and these are by no means obsolete just because we can do voice over IP.
But with so much WLAN capacity going into the enterprise and public spaces, why not use WLANs for VOIP? And, indeed, the designers of 802.11 included in the original standard a couple of mechanisms designed to support the time-bounded traffic typical of voice (and, for that matter, video); these are known as the "Point Coordination Function" and "RTS/CTS". The Point Coordination Function, or PCF, is now being improved by the IEEE 802.11e Working Group to further enhance the performance of time-bounded packets within a wireless LAN. And there are already a number of VIOP WLAN products available, for example from Cisco, Spectralink, and Telesym. One would expect that VOIP on WLANs is really going to take off over the next couple of years; some analysts and investors believe that voice is going to be the major driver of future enterprise WLAN installations. I've pointed out that VOIP on WLANs will be a major incentive for cellular operators to make investments in public-access WLANs. It should even be possible to hand off a call, in real time, between a cellular network and a VOIPOWi-Fi network. Home cordless phones? Sure, another good application, facilitated by the upcoming availability of combined cellular/Wi-Fi handsets. And using your cell phone as your office phone, with cordless service as a bonus, has undeniable appeal - one handset, anytime, anywhere.
In short, there are no real network-related barriers to the VOIPOWi-Fi scenario. But there are a few radio-related issues. First, just as a reminder, wireless LANs operate in the unlicensed bands available in many parts of the world. The unlicensed bands are potentially (if not fundamentally) noisy, often loaded with interference, and thus a fairly hostile environment. Data communications typically aren't too badly affected in these bands because (a) the products use spread-spectrum communications and a variety of error-checking techniques to improve throughput, (b) interference tends to be bursty rather than continuous, and (c) while more data throughput is always desirable, it's not usually essential. Voice has essentially opposite requirements, with relatively small amounts of data transferred but very tight timing requirements. This means any interference can be deadly - the equivalent of a bad cellular or cordless-phone connection. Such can be unpleasant and in some cases fatal to a given connection.
The other big issue has to do with the 802.11 protocols themselves. As we noted above, the designers of 802.11 have undertaken considerable effort to consider what's required to support time-bounded communications, and improvements are under development. But the solutions in 802.11 essentially involve the prioritization of time-bounded traffic. This means that VOIPOWi-Fi will work mostly via the goodwill of all involved on a given network. If someone cheats and claims their traffic is higher priority even when it isn't, then the whole network will be adversely affected. Can this problem be avoided? In many cases, yes - but not always.
We also need to add that the lower throughput requirement for voice comes with a corresponding need for more network "headroom", to allow the bursty traffic typical of VOIP to gain access to the airwaves without unacceptable delay (latency). A data network can be fairly heavily loaded before adversity becomes obvious; not so with voice.
Nonetheless, I expect VOIPOWi-Fi to be very popular, especially on the 802.11a frequencies above 5 GHz. There's a whole lot of underutilized spectrum here, and even more will be allocated in the near future. Right now, 802.11a is viewed by many as an exotic technology - but it's a natural homeland for many future WLAN applications, including voice.