2006 predictions for wireless LANs
by Matthew Gast
The first one is easy. Municipal networks were a huge story in 2005, and will continue to be deployed in 2006. The difference is that it will be much messier in 2006. As with any emerging technology, the early deployments received a great deal of help from the vendors. (If you buy early, you can always count on the vendor to make you successful--that's the biggest reason to buy from a start-up.) With many cities in the country trying to build these networks, the industry will struggle to codify the knowledge that they have gained up until this point. Who will be building these networks, and how will they learn? How do you run a mesh network that automatically tunes itself and readjusts packet routing? What sort of node density do you need?
A related personal prediction: I will still have my DSL line at home in San Francisco. Despite all the ink about the city's municipal wireless network and RFP, I doubt the network will get built fast. There's already concern about the mayor cutting back-room deals. Even if the city picks a winner, the result will be litigation, not building a network.
After building a municipal wireless network, what can be done with it? Sure, there's the standard answer of providing access to the Internet, and possibly services from the city government. Both those answers are essentially an arbitrage, moving bits off existing wires to offer better versions of the existing services. Useful, but not revolutionary. What sorts of services can be offered over a municipal network that are hard to offer over the alternatives? The biggest advantage that a municipal network has over wires and cables is that it knows the general where stations are. (If receivers start to have GPS, you could get excellent location information outdoors, too.) Expect a great deal of talk about how to build a location framework on a metro-scale basis. Once you know location, you can get targeted advertising for the neighborhood (hello, Google!), or build directions from a current location.
A related prediction: I've written about the NextBus service before. They use GPS to find the location of transit vehicles and predict arrivals at stops. Right now, the service reports the vehicle location and sends predictions to signs through CDPD, an old (and expensive) packet data system for cellular networks. Reporting information over a municipal network would be much cheaper, since LANs don't charge by the bit.
Google is building municipal networks to develop services for them. They already have Google Local. Expect to see Google You-Are-Here, at least for any cities that they support the network in.
Location is an interesting avenue to offer potential services (see MIT's iSpots), but there are big engineering challenges. Any "interesting" location applications will be based on fairly coarse (say, 10 meter or greater) resolution.
Shifting slightly to security:
Many municipal networks are being built without any security. In 1999-2000, most corporate 802.11 networks were built without any security. As the awareness of the issues came to the fore, security became one of the dominant themes of wireless LANs. Municipal networks won't suffer the same fate because they can borrow all the security mechanisms that have been worked out by corporate wireless networks over the past few years. Issues like authentication, traffic encryption, account provisioning, and separation of network privileges may rear their heads, but the basic technology is pretty well understood.
As municipal networks adopt security protocols, 802.1X will become more commonplace. Given that the most scalable user account systems are based on reusable passwords, I expect that TTLS will emerge as a formidable authentication method for wireless LANs. (This flexibility has been obvious since 2002, when I fist wrote a comparison of TTLS and PEAP.
WPA Security: I was surprised that easy-to-use automatic attack tools for WPA's pre-shared key mode didn't hit the mainstream consciousness in 2005. Pre-shared key authentication is nowhere near as secure as authentication against a RADIUS database. Although I hope that 2006 brings a realization that pre-shared keys should be replaced with built-in small-scale RADIUS servers, I wouldn't bet on it unless an automatic easy-to-use pre-shared key cracker is released and covered widely in the press.
802.11 on public transit will grow, but it will not affect me. Many of the earliest adopters have a large number of high-tech commuters that are equipped to take advantage of it. The emergence of cellular-based data services provides another alternative, and according to BART's current wireless plans, my commute will be completely covered by wireless data.
2005 saw the adoption of voice on 802.11 get underway. Everything that's currently on the market either uses the "ostrich approach" to prioritizing voice (there isn't any contention for network capacity, so we don't need to worry about it) or some proprietary implementation. Now that 802.11e has been ratified, there are protocol features that should help enable smoother coordination between transmissions. Expect to see some handsets that use 802.11e's Automatic Power Save Delivery (APSD) sometime this year.
In other standards activity, there will be increasing standards activity around fast roaming. 802.11r went to its first letter ballot in November. I don't expect it will be approved in 2006, but watch for it to be talked about a lot more. Also, expect to hear more about 802.21, the inter-network handoff.
Buy the book 802.11 Wireless Networks the last week and I want write to the autor for a pair of doubts about cover studies.
Can you say to me the e-mail address to ask to him?