Voice-over-WLAN (VoWLAN) technology is supposed to help end-users stay in touch with customers and colleagues even when they're not at their desks. But it is getting started slowly - and one of the big problems is waiting for network gear makers to create quality-of-service (QoS) standards.
Ted Stevenson, executive editor of the Jupitermedia's Wireless Internet Resources Channel, an online wireless technology resource, said VoWLAN equipment, which transmits phone calls over Wi-Fi networks, is "taking hold very slowly," and the turtle-paced adoption rate might stem from a protocol void.
VoWLAN lacks QoS techniques to ensure voice traffic gets enough bandwidth for jitter-free transmissions. Without QoS, voice must compete against data traffic. If data gets in the way of voice, users could experience garbled conversations.
Standards bodies host the voice-data war
Standards bodies are devising ways to make peace between data and voice on wireless nets - the most significant move being the IEEE's proposed 802.11e ("11e") protocol, which aims to bring QoS to WLANs.
We have a description of the two flavours in our feature, Quality of Service over WLANS, but Thurson identifies the two strands as coming from different industry biases. The first one, dubbed "mandatory," is also known as Wi-Fi Multimedia Extensions (WME). Data equipment makers designed this version, which simply lets Wi-Fi access points discern between data traffic and voice calls, so the network can give priority to the vocal track, and it became part of 802.11e.
The second 11e iteration, Wi-Fi Scheduled Multimedia (WSM), is a more sophisticated prioritisation scheme created by voice technology companies, says Thurson. It checks network capacity before letting voice transmissions onto the WLAN. The whole 802.11e package, including both WME and WSM, should reach ratification by mid-2004, Thurson told a panel session at the Wi-Fi Planet conference in Toronto in March.
802.11e has been slow in coming because it's difficult to reconcile voice and data on the same network, says Thursonn, especially when voice and data experts come up with different notions of prioritisation (WSM and WME respectively).
"It's not so much a fight as they [data experts] don't think about it," Thurson said, explaining that perhaps people in the data camp didn't appreciate just how sensitive voice transmissions are to network traffic congestion.
Joel Vincent, director of product marketing at Wi-Fi system provider Meru Networks, (which launched in Europe this month, and has products aimed at voice and security support on WLANs) said the debate over what to do about prioritisation swung between different schemes like WSM and WME, and creating a whole new wireless substructure, something designed from the ground up to support voice traffic.
He also pointed out that once equipment makers adopt 11e, they will have to do it whole-heartedly, and it may be an issue for IT managers to keep 11e running. If 11e-enabled devices compete for bandwidth against non-11e devices, the 11e equipment will lose. "The guys without 11e will dominate the conversation," he told the same Wi-Fi Planet panel.
Roaming is great but give us QoS
Another problem for VoWLAN implementations is roaming - the speed at which voice transmissions hop from one access point to another. When users go out of one access point's range and into the purview of another access point, the handoff between the two wireless nodes must be below 50 milliseconds, otherwise users would notice a dip in transmission quality. But right now WLAN handoffs take closer to 500 milliseconds.
The IEEE is investigating a "fast roaming" protocol to address this problem, but there are plenty of vendors keen to say they have a (proprietary) solution that already works.
Meru claims to offer fast roaming because its access points are all combined into one "virtual access point" by the central switch. Alain Mouttham, chief executive of SIPQuest, said his company's "collaboration agent" provides handoffs in the 10-millisecond range.
"It's handled at the application layer, which gives much more flexibility," says Mouttham, however he does not see it as a replacement for the eventual IEEE standard: "It's proprietary," he said. "You always prefer a standards-based solution."
So which is most important: traffic prioritisation or quick handoffs? Stevenson reckons that prioritization is the more basic need, pointing out that fast roaming (the ability to use the phone while on the hoof) means nothing if users can't get a dial tone in the first place.