Your browser is not supported. Please update it.

30 September 2005

Standards process shows its failings in QoS

The 802.11e standard for Wi-Fi quality of service, essential for VoIP because it prioritizes voice (and video) packets, is almost complete, but uptake of an interim specification published by the Wi-Fi Alliance, WMM, has been disappointing. This indicates that standards processes are lagging behind real market demands, particularly in critical areas such as QoS. In the data-only world, prioritization of packets was a side issue for Wi-Fi. In a world where the remit of that technology has been expanded to support voice services and citywide meshes, QoS is one of the biggest areas of risk.

WMM (Wi-Fi Multimedia), is a subset of 802.11e and designed to be a stopgap since the demand for services such as wireless VoIP was seen to be moving more quickly than the IEEE processes. However, Cisco and other key vendors have released products that do not incorporate WMM but stick with the vendors’ proprietary QoS mechanisms. Some suppliers, notably Meru, which makes voice-optimized Wi-Fi switches, claim their QoS is superior to 802.11e too, although they will have to support ‘e’ in order to retain Wi-Fi Alliance certification once this support is made mandatory, probably in late 2006. Cisco did not include WMM support in the new firmware for its enterprise 7920 VoWLAN handset and, although Spectralink does support WMM, it will also provide its own SVP (SpectraLink Voice Protocol) in the enterprise.

The issue highlights, again, the problem of standards bodies moving too slowly for real market demands, and now it seems that even the vendor-driven Alliance is failing to keep up with proprietary developments. Increasingly, we may see a ‘fast track’ approach adopted in preference to the broad ranging but slow methods of the IEEE and other bodies. This has already happened by default in the world of UltraWideBand short range networks, where the WiMedia Alliance’s platform, having been stranded in a stalemated IEEE process, has instead been submitted to a fast track standards body, ECMA. Unlike the IEEE – which allows for proposals, even once they are adopted as the preferred route for a standard, to be challenged and debated further ‘ ECMA allows no such complexity once a proposal is adopted (except on grounds of technical malfunction). Therefore it provides a process that is a couple of steps from rubber stamping. Though this carries dangers of proposals being adopted without sufficient discussion, or because a major vendor backs them, it certainly looks more appropriate to the timescales of the wireless industry than the fully consultative method, and is likely to be emulated in many areas of technology.

The Alliance started its WMM work in early 2004, with Meru chief software architect Joe Epstein in the lead, and with two projects, one for the home and the other for the enterprise. The home case is simpler because only one access point is required, and so little emphasis is placed on roaming. The average number of traditional handsets in the home case ranges from two to four, so security provided via manually entered WPA pre-shared keys should be enough. The MRD (marketing requirements document) for the home case has been completed, and the technical task group is now generating the methodology and producing test cases.

In the enterprise, issues such as roaming and battery life are more significant and centralized authentication schemes will have to be devised.

The Wi-Fi Alliance is concurrently working on the Simple Configuration Security specification, based on one-step security set-up products such as Broadcom’s SecureEasySetup and Buffalo’s AOSS.

Also available are admission control to prevent too many calls on one AP bringing it down, and APSD (automatic power save delivery), which allows devices to sleep when not in use. These two features will be certified, as WMM-SA, in the first half of 2006.

The Alliance is also working on its WCC (Wi-Fi/Cellular Convergence) project, and cooperating with the cellular body, the CTIA, to decide how it should approach dual-mode devices with its certification programs in future. Certification should start early next year with a simple SOHO case and then move on to the more complex enterprise case later.