Hotspot Road Trip, Part 8
January 03, 2005
Our traveler looks beyond just cellular to Wi-Fi hand off at the potential for voice over WLANs, both in public hotspots and at the enterprise level.
Mario's Bohemian Cigar Store is a San Francisco throwback, as is much of this section of North Beach. There's tangible nostalgia for the beats, and Mario's bridges the past with the present. Inside, it could easily be 1960. Except for the fact that I have a laptop open: Mario's is now a hotspot.
Despite its Cigar Store billing, no one is smoking, and Mario's feels more like an Italian café than the home of cigar aficionados. In fact, it's reputed to have the best cappuccino in the city. It was good, I admit.
As the day gave way to evening, I was to meet a friend up the street at Vesuvio, another North Beach establishment with a beat-era feel. In case you're tone deaf to this sort of thing, its location on Jack Kerouac Alley removes any doubt.
In my last post, I touched on WLAN-cellular integration, as well as voice over WLANs. The latter development, while being driven by the enterprise, could provide a spark to the hotspot market. While traveling, I eat up my fair share of cell minutes, and I'd certainly rather conserve my peak minutes by offloading the traffic onto a WLAN. In the interim between Mario's and Vesuvio, where my friends and I had not yet arranged a definitive meeting time, I burned up plenty of minutes with stupid two minute conversations about logistics. I was in a hotspot, big pipe, excess bandwidth, and it was a little annoying to be on a crowded cell network during the premium hours.
A workable WLAN-cellular integration strategy, though, necessitates better bundles from the carriers, as well as more far-reaching roaming agreements. But before any of this, it's important to look at WLANs themselves. Are they voice capable?
It depends on whom you talk to.
"When you add voice to a WLAN, you must invariably expect a heterogeneous device environment," said Scott Lindsay, vice president of marketing for Engim, a developer of multi-channel WLAN chipset technologies. In other words, when you're in that Starbucks with your laptop, you'll be sharing the network with voice-ready PDAs and data-capable mobile phones.
"The problem with this is that constrained devices can act as a drag on your network," he added.
An oft-overlooked issue of WLANs is that they're only as strong as their weakest link, and by weakest link, I don't mean the weakest link in the network architecture, but rather the weakest device. Constrained devices associate with APs at lower data rates, and these low data rates, due to 802.11 protection mechanisms, bring down the overall capacity of the network. "When you have mixed-device WLANs, it's the more powerful devices, like laptops, that suffer," Lindsay said.
I would much rather burn my anytime minutes than put up with a slow WLAN connection rate due to someone's poorly associating phone.The forthcoming 802.11e standard, which introduces Quality of Service (QoS) mechanisms to better support applications like voice and video, may alleviate voice issues in WLANs; however, there's already debate about whether 802.11e will scale to the enterprise level. For hotspots, it may help, but we'll just have to wait and see.
Lindsay argues that in large deployments what's needed is a dual-network approach, with a voice overlay network on top of a data-only network. This could even be an 802.11a voice network, say, over and 802.11b/g data network—although it could simply be different channel utilization within the same protocol. This approach protects performance for data devices, while also shielding the entire network from slow-associating devices.
The problem with enterprise WLANs today, from a voice perspective, is that the majority of the installed base is 802.11b/g, but these standards offer only three non-overlapping channels. (Remember, 802.11g offers better throughput than .11b, but it still operates in the crowded 2.4 GHz band and still only provides three channels.) 802.11a, which many analysts have dismissed due to its lack of legacy support, makes much more sense when you introduce voice into the mix.
Based on OFDM modulation, 802.11a offers the same theoretical throughput as 802.11g (54 Mbps), but it operates in the 5 GHz band where it doesn't face interference from things like microwaves ovens and cordless phones. Today's 802.11a gear also provides up to 12 non-overlapping channels, although that number is going up since the FCC recently opened up more spectrum in the 5 GHz band. Future 802.11a deployments could utilize more than 20 non-overlapping channels, a boon when you start running voice applications.
"I agree with you in regard to the evolution of 802.11a," said Richard Watson, director of product management for voice application developer Longboard. "One of the major problems with 802.11b/g is the limited number of channels, which makes it very difficult to achieve the high overlap coverage required by robust VoIP applications. I think 802.11a will soon achieve prominence in the enterprise, since it enables capacity and congestion to be better managed."
Watson noted that vendors focusing purely on hotspot access would see congestion problems arise if VoIP usage ever happens. "Imagine a meeting of 20 people down at the local Starbucks. What happens when they are all attempting to surf the net and use wireless VoIP at the same time? It's not going to happen in today's networks."
Another issue involves the subscriber-service provider relationship. "Who owns the Wi-Fi network?" Watson asked. "Is it the carrier, the enterprise, a third party? If it's anyone other than the carrier, it creates the logistical problem of negotiating services agreements."
Some carriers, including T-Mobile and Verizon Wireless, have begun investing in hotspots, but with so many hotspots in so many obscure locations, this is only a start. And how do you integrate the enterprise, which is critical to the success of VoWLAN efforts? From past conversations with WLAN switch vendors, including Trapeze, Airespace, and Aruba, one point becomes clear: for broad VoWLAN penetration, at some point the carriers have to be involved.
Ultimately, the success of VoWLANs will depend on the handoff between cellular and Wi-Fi networks—which brings me back to where I started this discussion in my last post. Today, WLANs and cellular networks have vastly different attributes, making integration a chore. "Our perspective is that we're not going to be able to get carriers to change the cellular system, and Wi-Fi is too decentralized to modify in any coherent way, so our goal at Longboard is to create software that mediates between the two," Watson said.
And make no mistake, that mediation will be needed soon. Earlier this month, the FCC freed VoIP from state regulations. Execs from VoIP vendors like Vonage, Packet8, and Free World Dial Up have already been making noise, claiming that this ruling ensures the viability of VoIP offerings and hinting that it essentially marks the end of the POTS (plain old telephone system) reign and the beginning of the VoIP era.
That may be overstating matters, since simple issues like enhanced 911 services present problems, but there's no doubt that the FCC ruling certainly makes the business case for VoIP more compelling. For large enterprises, this is a no-brainer. VoIP is going to save them a ton of money, grant them more control and manageability, and it'll be a natural evolution to extend this over WLANs.
The question while on this hotspot road trip, though, is: "Will voice ever really work in hotspots?" Many hotspots will retain legacy APs for a long, long time—meaning they won't exactly be voice paradises. On the other hand, many of these hotspots are also under-populated from a subscriber standpoint, so congestion issues will vary greatly.
While Watson's point about a crowded Starbucks is a good one, won't people who are using dual-device phones just simply switch back to the cellular channel if the WLAN is clogged? Rather, won't the devices have the smarts to do this for them? Well, if I've learned anything in covering this industry it's that just because it makes sense doesn't mean it will happen. We'll have to wait and see if these devices work as promised.
Jeff Vance is a technology writer and the president of Sandstorm Media. He focuses on trends in wireless communications, next-generation networking, security, and Internet infrastructure. If you have ideas about hotspots he should visit or questions he should investigate while on the road, you can contact him at firstname.lastname@example.org.