SF Politics and BroadbandPosted by sasha at 12 Feb 2007 11:03 am
The good and bad about the Wi-Fi deal - quick reference
I was stunned to hear that the normally very well informed Rita at SFist said the other day, “We must confess we’re still very vague on the pros and cons of the [Earthlink wi-fi] proposal”. Either she’s not been reading Left in SF, or I’ve not been clear about the pros and cons of the deal. I can’t correct the former, so I’ll address the latter with this quick reference to the good and bad points of the deal (note that I am not dealing with the philosophical or political issue of whether it’s a good idea for the city to trust a corporation to run such a vital function as communication):
Good points:
- Free (as in beer) to the city, including free lower-speed service to anyone who can get a signal.
- Comes with a little money. The city will receive a couple hundred grand a year.
- City-scale wi-fi networks are EarthLink’s new area of business. Their dial-up business is dwindling rapidly, so they need a new way to make money. They’ve settled on city-wide wi-fi systems, so they likely either have or are developing expertise in this area.
- The network is designed to have symmetric up and down speed. This is an isue which we’ve shamefully neglected to discuss, but is one of the few forward-thinking elements of the city’s original RFP. I’ll try to address this in a future post, but it means that however good the network is, it’s as good for producing content as it is for consuming it.
- For those who have decent network coverage, it can be a good way to begin to transcend the Digital Divide.
- Google’s involved. Everybody loves Google! Except Chinese dissidents, anyway.
Bad points:
- Service will be slow. Free service is supposed to be 300 kbps. That’s less than half the speed of DSL, and likely to drastically slow down as more people on the same block try to down- or upload stuff. For $22/month, you can get speed that’s comparable to what people in other cities are getting for free.
- The amount of money the city is getting is trivial. For reference, the $300,000 the city hopes to receive every year would be about .005% of the city budget.
- There’s no enforceable guarantee that the network will cover every part of the city. Earthlink will have an incentive to concentrate their work in the areas where they’ll make the most money, which are likely to be the areas that already have the best broadband coverage.
- The contract is long. It looks to be very difficult to not renew the contract, so we’re likely stuck with this network for 17 years.
- The network will be exclusive. Although the network is not an explicit monopoly, it will essentially take up all the bandwidth at the frequency wi-fi uses, so it would be difficult or impossible to have a competing network without using a completely different (and likely more expensive) technology.
- Even if the network covers all of the city, it won’t get to interior rooms or above the second floor. If you don’t live in a room with a window facing the street, or you live above the second floor, or you don’t live near one of Earthlink’s wi-fi nodes, you’ll need a repeater box (also sometimes called Customer Premises Equipment, or CPEs), which can cost anywhere from $50-150. You’ll also need somewhere to put the repeater where it gets a signal to begin with. So prepare to ask favors of your neighbors!
- Although getting on the network will not cost money, it will involve giving up your browsing habits to companies that may use your history for marketing, sales, or even more nefarious purposes.

February 12th, 2007 at 3:51 pm
Summary very helpful. Sounds like a lousy deal to me. My dsl experience with Earthlink was bait-and-switch.
Can’t we get a better deal?
February 12th, 2007 at 5:00 pm
So, I live in a six-unit, three-floor building with an additional two units in a back building. My unit’s on the first floor, but in the rear of the front building. Does this WiFi plan mean my first-floor neighbor — in the only first-floor unit facing the street and also the smallest unit — will be putting CPEs in her unit for all seven of us? I can’t say I see her wanting all that equipment, especially in a small efficiency unit. How do we get that equipment wired into our apartments, and into the back building?
Living in the Western Addition, I don’t think my landlord’s setup is at all unique — most properties on our block are like the one I live in. Has there been any analysis of how many CPEs will be required?
Thanks for your coverage of this issue — I wish I felt the BoS was paying attention to this….
February 12th, 2007 at 5:14 pm
Teddy,
the theory is that she’d need only one repeater/CPE box. It would then allow the rest of you to get a signal.
The caveats are that a) she needs to get a decent signal to repeat (so there needs to be a wi-fi node close by, without a big building in between), and b) that depending on your building’s construction, you may need another repeater on the third floor, or for the units in the back.
Just a point of clarification, though, is that the experts foresee problems above the second floor, not on it. So putting a CPE on the second floor’s likely to work, assuming the first floor does.
February 12th, 2007 at 5:18 pm
Willie,
That’s the $15 million question.
February 13th, 2007 at 12:30 pm
You say:
“The network will be exclusive. Although the network is not an explicit monopoly, it will essentially take up all the bandwidth at the frequency wi-fi uses, so it would be difficult or impossible to have a competing network without using a completely different (and likely more expensive) technology.”
Do your homework, man. This makes absolutely no sense from a technological point of view. The network itself will have limited bandwidth, but it can’t eat up the bandwidth from other competing 802.11b or g networks. Right now, in my apartment, I can pick up signals from at least half-a-dozen networks, and that’s because many wifi networks can co-exist in the same geographical space. They do it by hopping between “channels” on the 2.4 GHz segment of the spectrum that wifi uses.
Rob Flickenger, author of “Building Wireless Community Networks,” explains it like this:
The original 802.11 specification defines two RF modulation schemes, FHSS and DSSS. Both operate at 2.4GHz, but use the spectrum differently. Frequency Hopping Spread Spectrum (FHSS) breaks the available spectrum into 77 channels, each 1MHz wide. It uses a time-based, pseudo-random algorithm to quickly skip between all of the available channels in an attempt to avoid noise from other 2.4GHz devices. Direct Sequence Spread Spectrum (DSSS) breaks the same frequency range into 11 overlapping channels, each 5MHz apart (but 22MHz wide). It uses one channel at a time and employs more sophisticated encoding techniques to avoid noise and increase the data rate. (see more at: http://www.oreilly.com/catalog/wirelesscommnet2/chapter/ch03.pdf)
Technically, we will all be able to run our own private or open wifi networks in San Francisco if the EarthGoog deal goes through. The real question is whether the contract the city makes with EarthGoog will allow other commercial providers to create city-wide networks. Given that the deal is exclusive (not unlike SF’s cable franchise deals over the years), I think not. That’s the real thing to worry about, not your pseudo-science.
February 13th, 2007 at 1:30 pm
Thanks, Annale, for providing facts not gibberish…It is appreciated.
February 13th, 2007 at 5:00 pm
Well, that was my thought initially. However, after some conversations with some of the folks already employing wireless networks in San Francisco, I am convinced that two networks–even speaking strictly technically–will be problematic. I may have oversimplified, and possibly overstated the case, and IANARFE (I am not an RF Engineer), but here’s my understanding of what happens:
The more radios there are, the more signal is being emitted. Now, obviously, that signal is spread across the spectrum at hand differently, by the techniques you describe. However, as more and more radios are added, the number of collisions grows. For example, two devices using FHSS can land on the same 1MHz slice, or two devices using the same DSSS channel can broadcast at the same time. Now, as you mention, Wi-Fi has a large number of encoding and error-correction techniques, precisely to handle this kind of stuff. Unfortunately, these techniques do not come without a cost. For instance, “When an 802.11 device senses an interference burst occurring before it has started its own transmission, it will hold off transmission until the interference burst is finished. If the interference burst starts in the middle of an ongoing 802.11 transmission (and results in the packet not being received properly) then the lack of an acknowledgment packet will cause the transmitter to resend the packet.”
That kind of error correction doesn’t come without cost. It can really slow down the network, and only gets worse the more Wi-Fi nodes are in use. This is what I meant when I used the (admittedly incorrect) “saturate”.
As I said, I am not an RF engineer, but the folks I know who are are convinced that the EarthLink network will preclude another one of similar scale. Maybe not by making it impossible to have a similar (i.e. 2.4GHz 802.11b/g) network, but by making it impossible to put up a competing network that wouldn’t cause crappy performance by both of them.
Finally, I am interested where you found in the proposed contract that made you state “the deal is exclusive (not unlike SF’s cable franchise deals over the years)”. As far as I can tell, the deal is explicitly non-exclusive, in the legal sense (unless I am missing something). I refer you to Section 2.9: “Non-Exclusive. The authorizations granted in this Agreement shall be non-exclusive and nothing contained in this Agreement shall be construed to limit, alter, or waive the right of the City to enter into agreements authorizing persons or entities other than EarthLink to access and use Municipal Facilities to install any form of communications equipment or for any other purpose, except as specifically set forth herein.” and Sections 9.1: “Open Access” and 9.2: “Network Non-Discrimination.”
Thanks for your input. I’ll try to be more clear in this area next time.
February 13th, 2007 at 10:53 pm
Couple of other points on the questions in the comments:
Ever hear of First Mover Advantage? Earthlink/Google solution will mop all the low hanging fruit business. That will really raise the bar for a profitable business case for a competitor..
Also from my tour of the Tempe WAZMetro solution in 2005 with their network architect it was clear that only certain lightpoles can be used - ones that have 24/7 power (and not gang-switched by a solar detector for a string of lights), that are not blocked by a building, tree, hill, don’t have decorative fixtures, not in a historical location.
Why does this matter? Because about 30-40 well placed nodes per sq mile have to be deployed. When the next company comes to town or the city tries to rollout their own wifi network, the pole agreement with Earthlink understandably says hey you can’t interfere with what we already have up there - electrical power or radio interference. What’s the definition of interference here? Is it: uses the some overlapping WiFi channels? - i bet that’s the read.
And guess what? - there are even poles that only can support one device because of weight, wind stresses or limited available electrical power.
Unencumbered Vertical Assets of the city (lightpoles and building tops) are absolutely required.
COMPETITION
On the exclusive issue - there are two exclusive items in the contract.
1st: Daily Pass and Occasional Use- business travelers and tourists - (you know like hotel, t-mobile @ starbucks and jwire, paid hotspot cafes)
This is a high profit item that Earthlink has carved out for itself- you can bet t-mobile, jwire , zrnet and other paid wifi cafe options in the city would like to resale this access citywide - and that is good for us if they can.
Same for the free wifi from Google - they get that exclusively for everyone that signs-in (for which ACLU has now sent five letters - EFF and EPIC are concerned too).
You can bet that MSN, Yahoo, MetroFi, NetZero, and Anchorfree would love that eyeball traffic and would do deals for it and revenue sharing agreements on the ad revenue and offer various value added services.
Right now, there is no ad revenue sharing agreement in the contract - HELLO? Have you seen how much money Google is making? Bill Gates recently said in the near future people will be paid to surf the web - indicating how much revenue there is in online ads…
Forget location based services - if you live hear you know the nearest deli - but if you need something new nearby where do you look? Yellow pages. Basically Google is lining up an electronic Yellow Pages franchise - they are trying to get into local ads also with their click-to-call service on maps.google.com.
If the city ran this network they can still do a deal with Google and other providers and say how much anonymous eyeball traffic do you want (ie portal page that shows up when you open your browser) and what will you pay us and how are you going to help us promote locally run and owned nonchain businesses on your maps.google.com or whatever - after all this is affecting our newspaper media which is dying due to the online ad revolution and which pays for reporters that cover our local democracy.
And by the way, why isn’t the city getting any free wifi accounts in the deal for city business? Tempe got 3,200, and why isn’t their required unwiring of all the city library branches indoors and city hall and schools and SRO’s this should be an absolute requirement - that these building gets indoor coverage absolutely - that’s much more important than getting wifi in a sunny outcoors park that you use one a year and get a headache squinting to read the screen…
February 14th, 2007 at 10:02 am
Here’s another explaination of the wireless protocols and interference as relates to plans for 802.11N the next wifi standard.
http://blogs.zdnet.com/Ou/?p=411
February 20th, 2007 at 9:13 am
http://tinyurl.com/3dozfv
Ten myths of Wi-Fi interference
Myth #1: The only interference problems are from other 802.11 networks.
Myth #2: My network seems to be working, so interference must not be a problem.
Myth #5: I can overcome interference by having a high density of APs.
Myth #8: “There is no interference at 5 GHz.
February 22nd, 2007 at 8:01 am
more on interference (click link to get fully formatted and linked references):
http://www.unstrung.com/blog.asp?blog_sectionid=290
Thinking About Interference
FEBRUARY 21, 2007
discuss >
4:45 PM — If you examine the key barriers to the adoption of WiFi in historical perspective, it started with doubts about the reliability of radio in this application and proceeded to lack of a standard, low throughput, concerns about security, issues with total cost of ownership (TCO), network management challenges, rapid technological evolution and its impact on TCO, and now interference. All of these have been addressed — but few know about solutions for the last one.
I initially set out to discover if interference really is an issue or not in enterprise settings. I previously published a general White Paper on this topic and a methodology for evaluating the above. (See Thinking About Interference.)
I’ve now applied this methodology to real-world settings and completed testing on the effects of interference on general WiFi traffic, VoFi, and VidFi (that’s video over IP over WiFi). And the answer is yes, interference is already, or is going to be, a problem for many WLANs. Real-world testing revealed that some common forms of interference (like microwave ovens and cordless phones) can in fact be devastating to WiFi — and can regardless be very difficult to identify. You can read the results in Farpoint Group ’s three Tech Notes, published here, here , and here. But — cheer up. Good solutions are already available.
The key is to use Spectrum Assurance (SA) tools from companies like AirMagnet Inc. , Cognio Inc. , Fluke Networks , and WildPackets Inc. Farpoint Group owns a copy of Cognio’s Spectrum Expert and we use it frequently. Ultimately, we see SA tools being integrated with the RF Spectrum Management (RFSM) tools used for (among other mundane chores) setting transmit power levels and channel assignments already present in enterprise-class WLAN systems. So, over the next few years, the impact of interference will be minimized to the greatest degree possible. And the next barrier? Whatever it is, we’ll fix that one, too.
— Craig Mathias is Principal Analyst at the Farpoint Group , an advisory firm specializing in wireless communications and mobile computing. Special to Unstrung