Wednesday, August 22, 2007

VOE: Fixing What is Wrong with Muni WiFi

Voice of Experience Department.

[Lafayette's decision two years ago in voting to build a Fiber To The Home system rather than a cheaper, less capable wireless system is being validated by current events and the emerging pattern suggests that local citizens might end up owning the nation's most impressive model of a real, inexpensive, municipal network with modern bandwidth and workable mobility. Read on...]

Business Week picks up on current net buzz on the difficulties encountered by municipal wifi networks and the story does a good job in laying out the current unhappy state of such projects. It's a sad story for a lot of people in a lot of places.

The static crackling around municipal wireless networks is getting worse.

San Francisco Wi-Fi, perhaps the highest-profile project among the hundreds announced over the past few years, is in limbo. Milwaukee is delaying its plan to offer citywide wireless Internet access. The network build-out in Philadelphia, the trailblazer among major cities embracing wireless as a vital new form of municipal infrastructure, is progressing slower than expected.

My friends in Philly say the network is pretty near useless where it is up—service is beyond spotty and it comes and goes unpredictably. The boards tell a similar story in Corpus Christi where Earthlink, a private provider, had bought the municipal network with a promise of upgrades. Google's hometown Mountain View network isn't anything to brag on either. The problem isn't with public networks; difficulties seems to be hitting private and public muni wifi WANs (Wide Area Networks) pretty much equally.

There has been a lot of doom and gloom about the problems muni wifi networks are encountering (the Business Week article among them) and there has been the inevitable reaction to that on the part of advocates pointing out the immaturity—and naivete—of the original business plans. Business Week does, at the end of the story, note that a more mature business plan relies on the local city government being involved:
To make the business more profitable, Wi-Fi service providers are trying to pass more of the cost to the cities. "There's no one that I am aware of right now who'd build a network without the city as a paying customer," says Lou Pelosi, vice-president for marketing at MetroFi, which six months ago stopped bidding for projects unless the city agreed to become the network's anchor tenant.
Advocates imply that a naive business plan is all that is wrong with the current crop of wide area wifi networks. Would that it were so.

The doom and gloom is overstated. But the truth is the version of muni wireless that emphasized cheap (or free) residential service using a wireless mesh to minimize costs was always a castle built on shaky technical grounds. From the beginning the fundamental concept was that you'd take a single expensive connection to the net and divide it up like the loaves and fishes between many users and still end up with sufficient connectivity to feed the masses. Thinking that way was hoping for the sort of miracle that doesn't occur in our daily world. An analogy might be taking your home connection and "sharing" it with most of your neighborhood. That might work at times. But service could never be very fast or reliable. (Yes, it's more complicated; I know--but that's a fair analogy.) Additional problems having to do with the nature of the spectrum allocated to wifi (short range power and a frequency that has trouble cutting through vegetation or walls) added the limitations of physics to the questionable network design decisions.

Those problems can be overcome. It's not even a twelve step program. Two will do

Step One is to abandon the idea that a wifi network will ever work well as a person's primary, reliable, home connection to the full richness of the network.

Rock solid reliability is not in the cards for wifi--and affordable access to a reliable always-on connection is a prerequisite for full participation in the emerging digital culture.

You will need a hardwired, preferably Fiber To The Home connection if you plan to make full, reliable, consistent use of downloadable video, cable TV, Voice over IP, security alarms, medical monitoring and the like. With a fiber connection every individual can easily and cheaply provision their own in-home wifi network if wireless suits their style.

Any community that takes that stand abandons at one blow all the unrealistic demands that wifi technology simply cannot fulfill. Concentrate on ubiquitous local coverage, emphasize mobility and help people understand that cell phone levels of reliability is the best that can be hoped for. (That level of service would be a huge boon even without the unrealistic expectation, with ubiquitous coverage I could get a connection anywhere while on the go. I might not be able to do everything with it I could do at home--but I could do almost anything I can imagine that I would want to do on the run. Including in the best case, which I'll get to below, mobile, albeit cell quality, VOIP.)

I do understand, and deeply sympathize with, the hope that cheap wifi could help close the digital divide. There still may be a role for it there if the bandwidth issues can be overcome (again see below). —But the reliability issues, arguably, are fundamental and the hacked-up solutions necessary unstable and too technically exacting to expect large populations to manage on their own. Pretending that wireless connectivity is the same as wired connectivity is profoundly misleading—and is a recipe for creating a second-class version of net usage where poorer users simply can't rely on the net being there and so aren't able to trust it fully enough to make it as central as their better-off brethren. Imagine what would have happened to telephone usage in our culture if the well-off got good, reliable, always on wired phone service. But "other people" got cheap, spotty, poor "radio" service on "garbage" bandwidth that might or might not work on any given day or location. That is the sort of divided service model was avoided in our phone history and if we try it today it will cause trouble downstream that I, for one, would rather avoid. The real solution to too expensive wired network connections is cheap wired network connections. And that is the solution that any conscientious community should seek. [I am grateful that that is the solution Lafayette has sought—LUS proposes to narrow the digital divide by making service significantly cheaper.]

With cheap, wired, reliable, big broadband available in every home the threshold moves to making some form of connection available on every corner (ubiquity) and making it available while you are on the move (mobility). That's what wireless networks are good for—and why cell phones, as unreliable as they are, remain useful and hence popular.

Step Two is to abandon the the belief that wireless mesh networks can be used to turn an expensive wired connection into many cheap wireless ones.

It can't; only Christ could manage the miracle of the Sermon on the Mount.

Build, instead, on the real virtues of wireless networks: ubiquity and mobility. Do your absolute best to minimize its weaknesses by making it as fast and reliable as possible within the confines set by physics and federal regulation.

Abandoning the idea that one connection the wired broadband internet can serve many users over a broad area well is the key to succeeding. Instead of designing the wireless network so that each wired connection feeds five, six, or more wifi access points, limit the ratio of access points to internet connections to 1:1. This makes for much less sharing of limited bandwidth among users, greater reliability, and dramatically reduced "latency" (the lag caused by mulitple jumps that makes VOIP phones impractical on most muni networks).

Better yet, attach your wifi network directly to a full throttle fiber network. Fund the entire capacity of wireless protocols. (Outside of a few University or corporate campuses very few of us have ever used a wireless network that worked the internet as fast as they could. The usual limiting factor is the wired network that supplies bandwidth to the wifi. If Cox or AT&T only gives me 5 megs of wired bandwidth to my access point then the theoretical 54 Mbit/s that is theoretically possible is limited to at most 5 Mbit/s. You'll never see the other 49 Mbit/s no matter what it says on the side of the box.) A fiber network can easily supply a minimum of 100 Mbit/s supplied to the wifi access point; split that 100 once to a second wifi access point and something close to the full 50 megs of bandwidth that wifi is capable of could actually be seen on the street. Even split among a sizeable group of users on two nodes that would be plenty fast enough to support excellent quality VOIP with no discernable lag, great data connections, and many, many extras. Even if turned out to be less reliable and a bit slower in use than its wired counterparts the virtues of ubiquity and mobility would be there and our willingness to use cell phones proves that we find this trade-off acceptable.

A wifi network built this way would be as much superior to its wireless competitors as the fiber network would be to its wireline competitors.

But getting to that dream requires abandoning unrealistic expectations...and starting with a fiber network running down every street.

Lafayette is positioned to realize the ultimate dream: a cheap, blindingly fast, reliable, fiber-optic connection made available to every home and, based on that, a solidly architected, cheap, uniquely fast municipal wireless network that is demonstrably better than any muni wifi network in the nation.

Living large in Lafayette.

(Thanks go out to reader Scott who forwarded the story.)


Eric Dynamic said...

We're fighting corporate wifi in SF. Some of us are still focused on wifi and want SF to do it. Some of us (myself included) want to bypass wifi (as a terrible technology) and do FIBER first - THEN see what wireless technology would best overlay the fiber network.

A GSM cellular system such as they have (for 10 years had) in Tokyo, that delivers 26mbps streaming video to handhelds, would have met the SF wireless bid spec - but GSM providers were not invited. We don't want to pay a telco's rates for a GSM network, perhaps by the time the fiber networks are running, municipalities will be in a better position to bargain for coverage.

The true path to freedom is reform at the FCC. Whichever carrier has leased GSM frequencies in the SF Bay Area, we tell them to deploy within a year or give up the license.

"wifi", meaning 802.11-anything, is a junk short-range technology intended for casual civilian use. It can be operated in a mesh for improved service - citizens in SF tired of the Earthlink snafu are wiring themselves up using Meraki nodes. But wifi, running on unlicensed frequencies, is junk for "carrier-grade" applications. It is not reliable, always-on or fast.

My recommendation to SF is to deploy wifi only to underserved community locations. However tempting to use wifi to "fill in in the interim", don't spend much time and attention on it - please. Not at a city level. Let users in an area connect their neighbors (and take up the admin overhead to do so.) Leave the City free to finish fiber deployment and only then see what remains to do wirelessly.

Wifi will work better without congestion - it is not hopeless. It is just not adequate to wire whole regions without being hideously time-consuming, expensive, labor-intensive and problematic. Try as best you can to distribute the costs and labor to the USERS - then they'll understand why YOU (the city itself) don't want to be stuck with it.

See for the model I'm trying to push in SF, Berkeley and Oakland.

John said...

Hi Eric,

Good to hear from the Golden State. I've been following the mess in San Francisco and have to agree to you that the appropriate order is fiber first and then wifi--and that wifi isn't likely to be much good for a primary connection to the net.

What is the state of the fiber push in SF? Is it likely to win through? I understand they are talking about running it in the underground utility corridors?

Eric Dynamic said...

They aren't doing much with it yet, BECAUSE they're hassling over "wifi or no wifi". (SF Mayor) Gavin Newsom made ridiculous claims for what wifi could and would deliver, and he's doing everything he possibly can to push his corporate giveaway deal through, DESPITE being pounded on about fiber and having been told in volumes that wifi will suck and that the corporate deal sucks. He's pretending deafness.

Newsom is trying to get the Earthlink deal through by trying to have SFcans believe that if they don't get the Earthlink deal, they'll never get anything. He's tried to make people believe that wifi is some golden solution and that the "bad" city supervisors are trying to "take a wonderful thing away from you the citizens." He and his yes-man DTIS (dept of telecom and IS) ignore all challenges to their assertions - that is, they get to lie to the public repeatedly, and the local media do nothing about it - except for the SF Bay Guardian, a well-known free paper here. Then the lead on the board of supes (Aaron Peskin) instead of taking responsibility to kill the deal in a vote, punted and either put, or let the mayor put, the issue of "do you want free wifi?" on the ballot to be voted on as an initiative. But of course neither Peskin nor Newsom will make sure the voters hear what a swindle the deal is before they vote on it. Sasha Magee compared the initiative to one that reads "would SFcans like free ice-cream?", meaning that of course who would say NO. So ...

I find it very tiresome to watch politicians manipulate the public through the public's lack of information. They (the pols) have an oath of office, to serve their citizens. Refusing to inform the voters is dereliction of duty, yet again (still), the local media pay no attention and themselves do nothing to inform the voters - about the issues, or about their politicians' abuses of their ignorance.

No accountability => you get shit and a note explaining that you should like it.

If I can find a way to force a debate out of any of these jokers, I should. Televised or on the radio at least.