paul callahan…mostly common sense

At the heart of the war of words between wireline carriers like AT&T et al. and Web businesses like Google is the carriers’ saber rattling about “tiered services.”  The carriers are threatening to offer services that sense different traffic types, charging more for zippy video and good quality voice than pokey email and Web traffic.

This one of the oldest recurring debates in networking.  I can hear the annoying “beep beep” of roadrunner as wylie e. coyote fires up another defective Acme rocket sled.  An unnatural act — but highly amusing, nonetheless.

Here’s how it goes.  If bandwidth is so dear, wouldn’t it be better to prioritize traffic with different classes of service (CoS) to ensure quality of service (QoS)?  In other words, wouldn’t everybody be happier if the current pipes were carved up into high-speed, mid-speed, and low-speed lanes?  Wouldn’t users be guaranteed better service and couldn’t the service providers make more money?  Even the terminology is so old it gives me a headache.

Guess what?  The answer is always, “no.”  And the answer will be “no” again this time, too.  Why?  Because:

1. Simpler has always been cheaper.  The complexity of inspecting and classifying every data packet always leads to more expensive and potentially quirky equipment.  Equipment that provides a faster, dumber pipe has always been easier and — therefore — cheaper to build.  When that rule of thumb changes, send me an alert.  Until then, QoS with all of its angels dancing on the head of a pin will continue to be a cruel joke repeatedly played on unwitting engineers.
2. Competition rules.  If collusion was legal and competition did not exist, the big boys would get their way.  That’s not happening.  Someone in the pack will always break rank and offer the consumers of bandwidth what they want — throttle-wide-open pipes for less.
3. A filled pipe is a happy pipe.  Carving out reserved bandwidth is usually a dream that leads to less money.  By definition, carving out a high speed lane means the highway overall will not be filled to capacity — especially when you commit to delivering it end-to-end.  Wide open bandwidth is much easier to run at capacity and therefore more cost efficient.

Equipment vendors will definitely make money selling a new class of gear with QoS features for tiered carrier services.  Why?  Because the customer (the carrier) is always right.  But in the end it will be like selling picks and shovels to hopeful gold miners — a good thing until the miners discover there ain’t no gold in them thar hills.

Directory Tags: | Internet,| Wireline, | QoS, |Google, | AT&T |

Conflict makes great news, great reading.  People love drama, even when it involves the geekdom of network technology.  In the wireless networking industry, one of the dramas on stage at the moment is the unlicensed spectrum versus licensed spectrum soap opera.  The emerging demand for muni-wireless has created two camps which are developing as direct competitors.  Here is how they would probably describe their positions.

1. Unlicensed is open.  The world should be open and free.  All people should have access to the Internet and to community networks for free — or for advertising.  Wireless networks should be some combination of kumbaya, my Lord, kumbaya, power to the people, and Madison Ave.  The push for muniWi-Fi is a movement — and like the 60s movement, if you are in city government you’d better pay attention.  If you want to keep your job, you’d better be putting together an RFP for a municipal wireless network for your town or city.  Key players include, Nortel’s mesh for the non-US market (let’s say the city of Prague), Tropos-Earthlink-Google for San Francisco, and BelAir-Wildfire for Scottdale, Arizona.
2. Licensed actually works.  If you’re dying of a heart attack and you make a call on your free GoogleNet Wi-Fi phone as your lying on the sidewalk in Haight-Ashbury, you might not get through because some dodo on the same block is downloading a porn video, or stepping on your frequency with his death-ray continuous wave X-cam wireless camera.  If you really want wireless to work, you gotta pay.  And you gotta pay the wireless operators, Sprint, AT&T, Orange, blah, blah, blah.  So if you want to keep your city government job, you had better buddy up with the incumbent carriers and operators who know how to run a network 24×7.  Key suppliers include Lucent, Nortel, Airvana, Siemens, Alcatel, Huawei.

But a few players have actually figured out something that the others haven’t.  This not to say that others don’t have products under development, but for the moment these are the only two I can find.  What is the magic pill they have swallowed?

If you stop grousing and put the two technologies together, you end up with a much better product.

Who are they?  For the moment:

• Motorola’s MotoMesh.  Moto has recently started commercial rollout of its public safety mesh product.  Though it uses a multi-radio Atheros-based configuration which would normally be unlicensed-only Wi-Fi, the company tunes some of the Atheros radios to the 4.9 Ghz licensed band with boosted transmit power for a reliable backhaul.  Downlink, or subscriber links are predominately unlicensed Wi-Fi.  Pretty good answer.
• BelAir + ip-access.  By stuffing a licensed band miniature GSM mobile base station into its unlicensed mesh metro product, the two companies have enable a new class of product.  Base stations and APs on a pole that can serve up mobile voice traffic over wireless mesh backhaul and deliver Wi-Fi to the subscriber.  Also a pretty good invention.

Party on with those mash-ups and mix-ins.

Directory Tags: | Wireless, | MuniWireless, |Mobile, | GPRS, | HSDPA, | GSM, | 3G, | TV, | EVDO | Wi-Fi | WiFi | DMOZ Wireless | Yahoo Wi-Fi

I’m not going to argue about whether TV on your cell phone — mobile broadband — is going to be big or going to be little. Fact is, TV on your cell phone will happen. Whether it happens quickly and is a smash hit, or whether it dribbles out at the very top end of the market, is a different discussion.

Comcast, Cox, Sprint, and Time Warner are doing a joint venture, and Verizon has already rolled out Vcast.  Those activities are enough to cause a lot of money to move, if only to build out significant portions of the infrastructure to support spot markets.  What happens as a result?

1. Rapid build-out gets more critical.  Odd, isn’t it, using the word critical.  TV on your phone is about as far from critical as you can get.  So is PVR playback.  But no one will pay for that candy — or even use it — if they can’t see not-freezing, not-disconnecting images on their tiny screens.  Heck, they won’t even try it.  Those factors argue for a fairly substantial deployment.  But the biggest impact will be the need to address hotspots.  Teenagers are a critical segment and they want what they want and they want it NOW.  If coverage at your average high school is where the action is, the operators will need to jump on it quickly.  That argues for relatively unencumbered, rapid deployment.
2. Little cells take hold.  If you believe the argument that TV in your phone drives the need for better spot coverage and flexible deployment, then it’s logical to think that mobile operators are going to need an alternative to towers.  With all those cable partners in hand, Verizon and Sprint could ride on that infrastructure with oodles of on-the-pole micro and pico cells versus big honking towers.  Baby, what a new product opportunity.
3. Different pricing model is needed.  For gadgets, the retail technology merchandising scheme is really simple.  When Linksys or D-Link introduce a new product at CompUSA, the new doodad takes the top spot away from the previous doodad — let’s say at $129.99 — and pushes the previous doodad down to $99.99, which pushes the old doodad in that slot down to $69.99, and so on.  In other words, new features are introduced to support the price point, not to raise it.  TV on your phone should follow the same model.  If all I’m willing to pay is $80 per month, and that’s my story and I’m sticking to it, then the price of my non-TV phone service should go down to $60.  Customers are not going to buy an increase for TV on their phones.

Directory Tags: | Wireless, | Mobile, | GPRS, | HSDPA, | GSM, | 3G, | TV, | EVDO | Wi-Fi | WiFi | DMOZ Wireless | Yahoo Wi-Fi

More evidence of multi-radio, multi-service devices coming to market with interesting and — hey, look at that! — useful applications.  WebTech Wireless has produced real world, hardened, wireless for trucks.  Wireless for real men, manly men, men’s men…er, and real women in trucking companies.

WebTech started with products that combined GPS and GSM/GPRS for fleet tracking, location, and logging.  But recently, the company announced a product that includes a Wi-Fi radio, the WT1800 Mobile Access Point.

Adding a Wi-Fi access point (AP) allows the trucker to establish her own in-vehicle 802.11 b/g WLAN to connect a laptop or PDA.  The Wi-Fi AP is bridged to the GPRS/EDGE connection as a backhaul.  With GPRS the connection is going to be pretty darn sluggish, but useful for email I suppose.  HSDPA-based services will improve that performance with 400–700K and bursts to a megabit as they roll out.

What makes this product cool, though, is not the smashing together of radios.  Adding radios expands the breadth of applications, but the radios are much less important than the applications themselves.  Apart from the most obvious location tracking and mapping:

• DOT driver logs include time behind the wheel, breaks, compliance with federal regulations.
• Fuel tax data is automatically sent to a partnering service.
• Direct interface to truck diagnostics flags service issues.
• Remote panic button to driver to summon help.
• Ignition disabling for hi-jack protection.
• Connection could be used for real-time inventory and invoicing apps.

I see a definite lesson in this simple repackaging of commodity radio technology.  The real value and differentiation is in the apps.

Technorati Tags: WiFi, GPS, GPRS, HSDPA, Wireless, 3G, Location, Mapping

In digging into the details of ip.access’ product offering, I ran across a recent television interview with Managing Director, Stephen Mallinson.  When asked about future products for the 3G market, he mentioned that the company was working on an offering for GSM operators that placed a small base station in the home. 

Whoa, Nellie.  A base station for the home?  The company’s current GSM/GPRS/EDGE product, nanoGSM, is designed for in-building deployment and provides backhaul over an Ethernet IP connection.  I would guess that the 3G product would have the same IP backhaul and could be plugged into any broadband service gateway that presents an Ethernet port.  And given the fact that ip.access has already produced a pico cell base station for GSM that looks to be the size of a large-ish consumer access point, I expect the 3G offering to be equally compact.

But wait a minute.  This is not some unlicensed spectrum, Wi-Fi radio widget.  We’re talking about licensed spectrum, right?  This is like dancing with the giants isn’t it?  Presumably, operators like AT&T and Cingular would offer these as part of a 3G service offering — and the license to use the spectrum is held by the operator.  Hmmm.  I guess a license to operate radios on a given band allows mobile operators to put base stations in the home.  Why not?  I wonder what the limitations are on RF transmit power in a case like that?  

Mallinson also said that the consumer 3G base station would be called an access point and priced in the range of “a few hundred euros.”

What is the technical challenge here?  Apart from squeezing the technology into tiny package, someone had better be thinking about automatic channel selection to avoid the problem of base stations interfering with one another.  Today’s cell towers are manually configured by RF engineers.  If my neighbor and I both have 3G access points in our houses, who picks the non-interfering channel?

Technorati Tags: Wireless | GPRS | EDGE | 3G | RF | GSM | Spectrum | Femtocell |

BelAir and ip.access have smooshed together their products.  But the resulting combination is not a galactic, convergent, IMS (IP Multimedia Subsystem) architecture defined by 3GPP — it is a simple combination of BelAir’s Wi-Fi based metro mesh and ip.access’ GSM pico cell products.

How does it work?  BelAir has stuffed a tiny GSM base station from ip.access inside its metro mesh enclosure, the BelAir 200.  Mounted on a utility pole, the unit becomes a self-contained GSM base station with IP-based wireless backhaul over BelAir’s metro mesh.

What’s it for?  Currently, the two companies are positioning the combined solution as a GSM coverage filler.  Existing mobile operators can place a few nodes in urban areas where customers complain about coverage.

But I think that it could have much broader appeal:

1. Low profile cellular.  I live in a little New England town with terrible cell coverage.  Why?  Because no one can agree on a low-impact cell tower implementation that won’t cause an armed revolt led by the protectors of our colonial architecture.  If we let at&t or Cingular Put 70 to 100 of these puppies on utility poles, everybody’s happy.
2. Wi-Fi service offering.  Along with GSM coverage, mobile operators could offer neighborhood Wi-Fi access to make their customers sticky, or to lure subscribers away from the competition.

What’s in the way?  Money.  The operators must weigh the cost of prolonged disputes and lawsuits over cell tower placements against the cost of a metro/GSM system like this.  Let’s face it, three cell towers are probably cheaper to install and maintain than 100 pole-mounted widgets.  Let’s see how it all costs out when pricing becomes available.

Technorati Tags: WiFi, IMS, Metro, Muni, Wireless, GSM, BelAir, Mesh, Femtocell

Given Linksys’ recent announcement of a Wi-Fi SIP phone, I was interested in getting more information on Netgear’s Wi-Fi Skype phone.  Here are some answers.

1. How does the phone handle Wi-Fi authentication in hotspots?  Most hotspots use an authentication Web page with user name and password, but Netgear’s phone will not have a Web page. Users will need to find either a free hotspot, or one that supports authentication via WEP or WPA.  The phone is targeted more for home use, than for hotspots — and Netgear believes the Wi-Fi Skype phone will end up as a second line phone for the home, or in some cases even as a replacement for current phone service.
2. Does the phone have any dependency on a PC?  The Skype phone will be a fully independent Skype client and will not need to synchronize with a computer.  The phone will interact with the Skype network in much the same way today’s Skype softphone does and will automatically download contacts when it connects.
3. Is the firmware upgradable?  All upgrades will be direct via the net.  Because the phone connects via Wi-Fi, it will not require a PC connection in order to upgrade firmware.
4. What about SIP?  Netgear’s phone will not support SIP.  Netgear says it does not rule out providing a SIP-based phone in the future.  If Netgear does produce a product, it will launch with a specific carrier — and its phone will ship pre-configured for a VoIP service provider.
5. What about battery life?  Netgear estimates a 3 hour talk time, a feature deemed to be important for typical Skype calls — hour long international calls versus 2–3 minute local calls.  Idle time battery life is estimated at 48 hours.

Netgear says that it has been overwhelmed with interest in the Skype phone since the CES announcement and expects to ship this spring.

How do I think this market will pan out?

Wi-Fi Skype phones will see strong sales — but they will be a drop in the pond compared to the annual volume of today’s mobile phones at 80 million units.  Wi-Fi phone volume for 2005 was probably under 50,000 units world-wide.  And an optimistic forecast for 2006 would put Skype and SIP Wi-Fi phones in the range of 200,000 units.  Even so, if I had a brother or sister in Europe, or a son or daughter at college, I’d probably buy a pair.

Technorati Tags: WiFi, Skype, Netgear, Linksys, VoWLAN, VoWiFi, VoIP, SIP

Buried in the details of the Sprint Nextel megamerger was the purchase of a tiny company, Diamond I.  The company has almost no revenue but develops interesting technology — wireless gambling systems for cruise ships and casinos.  Diamond I’s product, WifiCasino GS includes:

• Wi-Fi equipped PDAs with biometric authentication in case you lose it.
• Location tracking so the casino can find you to deliver drinks.
• CRM system to analyze trends and behavior and an management system that can watch individual apps.

Why is this interesting?  Like the coming GoogleNet, this is another example of the power of knowing more and more about user network behavior.   Because it is a captive network, WifiCasino knows where you are, what application you’re using, and probably the status of your winnings.  The system can also be used to deliver advertising or event announcements for resort guests.

When the casino hands you the PDA, I’m sure you agree to the terms of being under this microscope.  Honestly, it is not much different than what goes on behind the scenes at all casinos today.  You just don’t notice the state of the art face recognition system that captures your image when you walk in the main entrance.  In both cases, privacy is out the window when you walk in the door.  That is the tacit arrangement you make with the casino.  Like other carefully watched and narrowly defined gaming scenarios, I expect Wi-Fi gambling to be highly profitable.

Technorati Tags: WiFi, Location, Gaming, Gambling, Wireless, PDA, Biometrics, Google

Weenie nets continue to drive network innovation, and this one is as bold as they come. FON was started by Martin Varsavsky who reportedly flew from Spain to Silicon Valley a few weeks ago with a business plan in hand and landed $22 million in funding from Google, Skype, Sequoia Capital, and Index Ventures. He says it was his first time in the valley, so I guess it turned out pretty well.

Okay, so I’m exaggerating the “came out of nowhere” part of the story. Varsavsky started ISP Ya.com and telecom carrier Jazztel and is pretty much a longstanding wild network revolutionary. Now, he is a wild Wi-Fi revolutionary.

FON plans to create 1 million hotspots world-wide in 4 years. How?

1. Give away firmware and seed Wi-Fi routers.  FON is selling Linksys Wi-Fi routers with new FON brains for $25. But braver FON users can optionally download new firmware and re-flash their own Linux-based Linksys Wi-Fi routers for free. Establishing a firewall, the FON firmware carves the routers in two: half for the community, half for the user.
2. Create a movement. FON’s movement creates Foneros — those who are willing to share their Wi-Fi with anyone they reach and create a people’s community network.  FON’s country managers are even called Foneros Leaders. Power to the people, baby, right on, right on, right on. Sorry, I grew up in the 60’s.  Honestly, FON’s marketing approach is phenomenal.
3. Share revenue with ISPs. Before they have heart attacks over the widespread unauthorized resale of their network service, FON plans to strike deals with willing ISPs for revenue sharing.

This an amazing scheme. Can it work? As bold as it is, this is a deal with a heck of lot of moving parts and, like any startup situation, they all have to work perfectly. Nevertheless, I’d love to see it happen. I see challenges in two areas: the firmware injection process, and the ISP deals. My advice?

• Make a real deal with Linksys. Asking the universe of users to reflash their routers is a non-starter. I’ve done it plenty of times, but if you make a mistake you can end up with dead box. If Martin can land Google, Skype, and a couple of venture firms in one trip to CA, one not shoot down to Orange county and convince Linksys to build FON in as an option for all of its Wi-Fi routers? Once you land Linksys, then you can pick off Netgear, D-Link, and Buffalo. Skype has deals with both Linksys and Netgear already. Skype’s biz dev guys should be able to grease the skids.
• Pick a big “ISP-neutral” authentication/accounting partner. Part of the benefit of FON is being able to roam around hotspots and securely authenticate to the network. Scaling to a million users world-wide will require a substantial distributed authentication and accounting system to handle ISP revenue sharing.  Better to do it once, instead of handling it case by case with each ISP. Who has the infrastructure to help? Google? Skype?

So far, this is what we know about Google’s plans for free Wi-Fi in San Francisco.  Apart from making network access free of charge, Google plans to:

1. Make it a playground for devices.  Attract as many users in the widest range of devices as possible.
2. Put it up and see how it’s used.  Gather data on what users search for from Knob Hill or how many X-boxers are gaming in Cole Valley.
3. Allow other Service Providers (SPs) to sell services over it.  But if they want to reach Google’s Wi-Fi users, SPs may need to pay. 
4. Provide location-based ads and maps.  Initial location tracking may be crude — within two city blocks — but must inevitably improve to provide mapping.

What’s this all about?

Extending the franchise to the physical world

If capturing intent on the Internet can make money, why not take it to the physical world?  Better yet, combining the physical world with the Internet will yield rich new data about the behavior of the masses.  If the vast majority of Google Wi-Fi users search for doughnuts when passing through Union Square in San Francisco, Google doesn’t need to explain why.  Data is data.  This is what people do when they are in that vicinity.  Once the data is firmly established, the value of the advertising space in that rough location goes up and up.  If Krispy Kreme wants an ad in that wireless zone, it will have to pay more.

Establishing Google as a peer

Google admits to building its own ISP in the article cited above.  This what a gigantic market cap will do to the brain — cause an attack of hubris that could create a major distraction for the company.  But I think it actually makes sense.  Even if Google only builds out the municipal Wi-Fi portion of the network and skips the construction of a nationwide backbone, the company will have built a network with inherent value to other ISPs.  The offer will be to peer with Google, establishing a balance of power between Google and other ISPs.  The real question will be this: Will Google offer more than connectivity to other ISPs?  Will Google offer a subset of its collected data, namely, where the users of the network are, and what they look for when they get there?