Communications

How tiered Internet pricing could actually facilitate P2P

Time Warner Cable's planned experiment with tiered charging for Internet access has generated a flurry of coverage in the blogsphere, but no new insights (at least that I've seen).

The primary problem ISP's complain about is that 5% of their customers use 90% of the available bandwidth and when they examine this traffic, it's mostly peer-to-peer file sharing.  A reasonable question is how to allow as much of this traffic as possible without increasing an ISP's variable costs or slowing down their other users.

This may not be as difficult as it appears.   Indeed if Internet access was as competitive as mobile telephony, we might already have seen what I'm about to propose — a combination of bundled pricing equivalent to mobile's "free nights and weekends" and "free on-net calls" with a way to facilitate P2P traffic that leverages exactly these "free" periods.

An ISP's costs

ISPs have some costs which are relatively fixed and others that are tied to usage.  A network is a relatively fixed cost and when it's not full, the incremental cost of adding traffic is zero!  This is the reason mobile operators give away free nights and weekend.  They've built their mobile network for the peak daytime traffic, so it costs them nothing to run promotions that add incremental traffic at off hours.  Peak hours and off hours may be different for an ISP, but the concept is the same. When a data pipe is lightly loaded the ISP's cost of adding incremental traffic is zero.

On the other hand, some ISP costs are usage based, for example "IP Transit" or more properly, Internet Transit.  This is the ISP's upstream cost to send and receive traffic to/from the rest of the Internet.  However, even here, usage-based costs occur at heavy usage.  Light usage periods don't save money.  To understand what's happening, it's worth a digression on Internet Transit.

Internet Transit

Internet access is monopoly or duopoly or a heavily regulated industry.  The middle mile connections from the local network to the Internet backbone may or may not be competitive depending on where you are.  But the Internet backbone itself is extremely competitive.  If you can get to a major Internet Exchange Point in the US or Europe, there are many providers offering extremely competitive rates for Internet Transit.  Typically these services are priced on a megabit per second per month basis (Mbit/s/Month) with lower rates for higher volume commitments.  The other key idea is that charges are based on the 95th percentile of all the five minute data rate samples taken during the month.  So an ISP can have a few bursts above their typical rate, as long as they represent less than 5% of the sampled intervals.

But this also means there is no extra cost to run at or near the typical rate at all times.

Local traffic

Even more important, if file sharing is done with other computers on the same ISP's network, then there is no need to pay for Internet Transit at all.  The question is how to figure out which potential peers are "on-net" and which are "off-net."

Sending signals to P2P software

Most P2P file sharing software has relatively little knowledge of locality.  Some P2P software practices "prefix awareness," for example, Joost gives preference to peers in the same /24 IP address block when they are available.  But if a major operator provided an automatic way for P2P client software to determine whether a prospective peer's IP address was currently reachable "for free", it seems likely the file sharing community would leap on it, and if there's money to be saved, active file sharers would download the new clients immediately.

A standard way to present such information might be via an extension to the XML-based response codes in one of the whois information exchange proposals, e.g. from ICANN or from APNIC.  Also, while what I'm proposing might start as a pricing plan rather like a mobile operator's "free nights and weekends" and "free on-net calling," it's not hard to see extensions where an ISP could offer dynamic access to underused capacity to those programs that were prepared regularly interrogate an ISP's server and use just the advertised off-hours capacity.

In closing

People liked fixed price deals.  Unlimited is great, but there's plenty of experience with bundles of minutes and the idea of data bundles has already showed up in 3G mobile data plans.  The combination of several tiered data bundle prices with the availability of "free" connectivity for "on-net" peers and during off peak intervals is likely to appeal to file sharers and produce better results for both the sponsoring ISPs and file sharers alike.

Google is playing to win in the 700 MHz auctions

Many say Google will bid to lose in the upcoming 700 MHz auctions and many more are equivocating.  The idea is Google's entry alone will induce enough openness, and besides they couldn't afford to become an operator.  This shows a total lack of understanding!

Google is run by idealists who want to change the world and have the money to undertake grand projects.

They are already seeking to index, and make available, all the world's information. As part of this vision they are scanning all existing books and fighting the legal battles this implies.  When their own video repository failed to gain traction, they bought YouTube for $1.65B, and took on their legal hassles. Most recently their Palimpsest project is going to host enormous scientific databases for free.  In short Google is not afraid to spend money or take large risks, including legal risks, to accomplish something they believe in.

Critical for Google's vision, and for their business, is open Internet access.  Fixed access is relatively open today, but mobile is a big problem.  So Google won't seek a little opening on the part of some US operators.  Google needs open mobile access, i.e. dumb pipes, which means the total destruction of existing mobile operator business models.

They've already started from the edge with the Open Handset Alliance and the Android open-source mobile handset stack. Google's made a major investment here, not to compete for revenue on handset software or to control applications on edge devices, but to tip the balance from operators to the edge.  First and foremost, Android is an open platform to encourage innovation.  It's also free.  Google doesn't need or expect a direct return on this investment.

The 700 MHz auctions are just the next step.  Again, look for something unconventional.  Google doesn't need or want to become a mobile operator.  They want to create an industry where mobile dumb pipes are widely available.

 

Assuming Google wins, what might they do?

This is speculation on my part, but a good plan might be two fold. 

First pick the latest mass market technology for mobile broadband access, probably mobile WiMAX, and set up a program to foster numerous independent wireless ISPs (WISPs) rolling out services on Google's frequencies.  In the 1990s, the US had thousands of ISPs providing dial up access over traditional phone lines.  The goal here would be to duplicate that entrepreneurial flurry for both fixed and mobile wireless access.

You may argue fixed is OK as services are local, but for mobile you really need regional and national coverage.  Yes, but Google themselves could offer WISPs the opportunity to participate in a federation for national roaming, perhaps using Google Checkout for payments.

Second, formalize a set of rules for smart-radios based on the work they've been doing for open access to TV White Space.  Throw open their spectrum to anyone who's willing to use appropriate smart radio technology, thus fostering long term innovation.  Open access in the 2.4 GHz and 5 GHz bands has certainly fostered innovation.  Google could achieve something similar.

These two steps, a mass market technology like WiMAX for rapid deployment, and open access for innovation, yield the maximum long term benefit for this spectrum and increase the likelihood that other countries will follow the US lead (as the world has done for 2.4 GHz and WiFi).

========================================================

 

Mobile VoIP in the context of mobile data capacity

When VoIP traffic began to grow in the late-90s, there was some worry that voice traffic would overwhelm the Internet.  Indeed at that time, voice traffic far exceeded total Internet capacity.  But the growth of Internet capacity has far outstripped the growth of voice minutes.  Today a substantial portion of long distance and international voice traffic goes via VoIP and increasingly, local traffic is moving to VoIP, yet VoIP represents only 1% of Internet traffic.  Voice traffic grew relatively slowly.  Total Internet capacity has expanded at ~100% per year during the 1990s and at greater than 50% per year since 2001..

So far, 3G mobile VoIP traffic is negligible and most mobile data traffic is confined to walled gardens.  But that's changing as mobile data capacity expands.  Today, laptops with USB 3G modems are driving 3G data revenues as people seek Internet access for their PCs.  Despite whatever Ts & Cs the operator may impose, what people want is open Internet access, i.e. a mobile dumb pipe.  And given the relatively vibrant competition in many mobile markets, increasing mobile capacity will lead to better and better "dumb pipe" offers.

The issues for successful 3G mobile VoIP are:

• capacity - if you have excess capacity, you don't need QoS.
• latency - round trip latencies must be below 250 ms to avoid a "push-to-talk" feeling.
• battery life - mobile VoIP may require some evolution to achieve the power efficiencies of the custom engineered protocols used for traditional mobile voice.
• devices - mobile VoIP will start on PCs, but must migrate to handsets to achieve volume.

Luckily battery life issues have been explored as part of implementing POC (push-to-talk over cellular) and should be well understood at this point.

The advent of HSUPA and EVDO Rev A radio technology brings latencies into the acceptable range.  Deployment of these technologies has begun.

That leaves capacity.  While there will be congestion during the "busy hour," daily traffic is cyclic.  During most of the day most cell sites have extra capacity.  And a 3G HSUPA modem handles 3-7 Mbps down and 1-2 Mbps up.  That's more than adequate headroom for VoIP.  In other words capacity is becoming available for most hours of the day.

Fixed VoIP took a decade to become pervasive and mobile VoIP will likely take as long, but the next 24 months should see early adoption flourish.

My EU & Asian colleagues' actual residential broadband arrangements

I spent much of this week in meetings with colleagues from Asia and Europe.  During breaks and social time, I ran an informal poll of their residential Internet Access arrangements.  Everyone I polled was either a sales manager or a sales engineer for NMS Communications and thus in the business of selling voice and video development platforms to people in the telecom industry, i.e. this is a fairly high tech bunch.

Why the survey?

There's plenty of country-by-country data on broadband Internet access, for example from the OECD, but speeds and prices vary and there is little information on exactly how many people have which speed. I did find this table in the OECD list,

but it provides the average of advertised broadband services in each country without regard to how many people use each service. On this basis, a service like HKBN's 1 Gbps Home Service would distort already excellent numbers (for Hong Kong, if Hong Kong were an OECD country).  So what do real people actually sign up for and what does it cost them?

The results of my informal poll

Complete data is in a file attached at the bottom of this post.

Japan leads with 100 Mbps symmetric service routinely available for ~US$23 per month.  I've run into this before with others in our Tokyo office going as far back as 2004.  What was new (to me) in this survey was 100 Mbps service in a detached house, not just in an urban apartment building.  As a side note, while it's a 100 Mbps Ethernet connection, typical speeds realized were claimed to be more like 65 Mbps.  I don't know how much of this is due to Ethernet overhead, measurement limitations, TCP parameters? or congestion.

Korea came in second as my colleague from Korea happens to have 50 Mbps VDSL service.  This costs him ~$20 per month.

Regardless of where people were, most people knew the download speed but were unsure of their upload speed.

Hong Kong was interesting.  100 Mbps service is routinely available, but most of my colleagues use ADSL service at speeds between 6 Mbps and 10 Mbps and monthly costs of US$23 to US$26.  One of our folks in Hong Kong is being driven, by his children, to upgrade to 100 Mbps symmetric, but it's quite a bit more expensive than Japan at 600 HK$ or about US$77 per month.

Next best were 20 Mbps services in the US and France.  Mine is Verizon FiOS 20/5 Mbps in the Boston suburbs for ~$55 per month.  The French connection was 20/1 Mbps ADSL from Orange at a detached house fairly far out of Paris.

A lot of people have DSL at various speeds from a low of 512Kbps in both a detached house in Delhi, India and an apartment in Paris, to 1 Mbps in Beijing, China to 2 Mbps near Tel Aviv, Israel to 4 & 8 Mbps near London.  The full detail is in this file:

Download Residential_BB_Survey_Jan_2007.xls

 

Global mobile network operators subscriber data

There's an interesting free source of market data that I've been using for the past two years.  Until two hours ago, I hadn't given this much thought, but when I mentioned it to some colleagues, their reaction was:  Wow!  This is really cool. 

So I thought I'd share it more widely. It's Wikipedia's List of Mobile Network Operators.

 

There's quite a bit of information, by country and by operator, for example:

I've been tracking mobile subscriber growth for more than ten years, in part by purchasing ITU data every other year or so; in part by getting copies of any data than any NMS business unit purchases from traditional market analysts and in part by capturing anything that's published in press releases or other teasers from analyst firms.

Wikipedia doesn't have all data for all operators, but they have almost everything and it's very credible.  They also cover countries that I have never seen in traditional analyst reports, for example:

Somolia is famous for having no functioning central government, no spectrum regulation and no telecom regulations, but none-the-less having the lowest international calling rates in Africa.

VideoSMS — Another cool service that avoids the problems of MMS

Video SMS, like Voice SMS, delivers an exciting service that works on any handset and any network.

Meanwhile, Multimedia Message Service (MMS) is mostly languishing.  In the US and much of the world, MMS is used for picture mail but hardly anything else.  There are several problems.  In most markets, there are a dearth of MMS capable (and appropriately configured) handsets, there are interoperability issues between handsets, particularly with video support or the lack thereof, and in some markets there are interoperability issues between service providers.

So new, simple, inter-operable services have emerged.  The first was Voice SMS, an audio messaging service that has taken off because it actually works with any handset and any network.  I’ve written about Voice SMS several times in the past as it’s a cool service and we supply platforms to many of the vendors in this market.

Today we had our first press release on a new service, Video SMS.  Maxis Communications, the leading operator in Malaysia, has launched their Maxis Video Avatar service in December, based on an application by NGC Systems and platforms from NMS Communications.

Like Voice SMS and unlike MMS, Video SMS is a messaging service that works with any handset.  If both subscribers have 3G handsets, the service leverages the 3G network to deliver an actual video message.  But in the more common case, where one or both of the subscribers are on 2G or 2.5G, the service still works.  Instead of an actual video message, the 2G user generates and/or the 2G recipient receives a voice message with an animated gif image of a lip-sync'd talking avatar.  If the recipient has a voice-only phone the service reverts to Voice SMS, but with a URL where recipient could view the lip sync'd avatar speaking the message if they can get web or WAP access.

From the user's point of view this is a cool new messaging service that serves the same needs as Voice SMS but with an added personalization element — animated avatars.  From the operator's point of view, it's a new revenue source with both message revenue and content revenue, as avatars can be sold just at ringtones and wallpapers are sold.

The key take-aways:

• Easy user interface
• Simple useful service
• Connects with any phone on any network

Smart Communications in the Philippines has also launched this service to their 27M subscribers. As they put it:

Video Avatar is a new P2P 3G video messaging service that combines lip-syncing fun avatars and a 30 second recorded message to create an expressive and memorable video message.

Flying to Vegas this morning

Now that CES is safely over and hotel rates are economical, I'm off to Las Vegas for some business meetings.  I'll be in there through Thursday afternoon. 

If by chance, you are in Las Vegas and want to meet, please contact me on Skype.

The Moral Instinct, cyberjustice and Internet law

There's an excellent article in today's NY Times Magazine, The Moral Instinct, in which Steven Pinker summarizes recent neuroscience research on the human moral sense.  This bears directly on how we think about justice and morality in cyberspace.

He covers the enormous progress (at least since I was in college) in our understanding of where the human moral sense comes from, including what is universal, what is cultural and how we can be easily mislead by "an aura of sanctity, distracting us from a more objective reckoning of the actions that make people suffer or flourish."

Illusions are a favorite tool of perception scientists for exposing the workings of the five senses, and of philosophers for shaking people out of the naïve belief that our minds give us a transparent window onto the world (since if our eyes can be fooled by an illusion, why should we trust them at other times?). Today, a new field is using illusions to unmask a sixth sense, the moral sense. Moral intuitions are being drawn out of people in the lab, on Web sites and in brain scanners, and are being explained with tools from game theory, neuroscience and evolutionary biology.

While the article doesn't mention the Internet, a better understanding of our moral sense can only help temper our approach to law and culture in the new social relationships facilitated by the Internet.  Depending upon the cultural setting, subjects as diverse as ethnic humor, pornography and religious discussion can be completely acceptable or morally outrageous and yet, except for some language effect, the Internet cuts across cultural boundaries.  Then there are Internet-related tragedies like the death of Megan Meier which provoke discussion and calls for sweeping laws to regulate cyber-behavior.

Undoubtedly, our laws and culture will evolve in response to the Internet age, but hopefully we do this with care, taking advantage of what we can learn from history and from the new science of the moral sense.

Here are a few short quotes from Pinker that struck me:

There are many other issues for which we are too quick to hit the moralization button and look for villains rather than bug fixes.

Our habit of moralizing problems, merging them with intuitions of purity and contamination, and resting content when we feel the right feelings, can get in the way of doing the right thing.

Far from debunking morality, then, the science of the moral sense can advance it, by allowing us to see through the illusions that evolution and culture have saddled us with and to focus on goals we can share and defend. As Anton Chekhov wrote, “Man will become better when you show him what he is like.”

If you are at all interested in morality, neuroscience or evolutionary biology, I highly recommend the Pinker article.

Hacking the Boeing 787 -- cartoon

As a frequent traveler, I couldn't help being a little worried by this post on Bruce Schneier's Security blog which discusses several news articles including this:

According to the U.S. Federal Aviation Administration, the new Boeing 787 Dreamliner aeroplane may have a serious security vulnerability in its on-board computer networks that could allow passengers to access the plane's control systems.

But reading his post reminded me of a great cartoon by Ritsch and Renn, originally published in the German IT magazine C'T.

The full original (in German of course) is here.  The English adaption is from the ComPilots website.

"TV white space" spectrum for backhaul - Yipes!

An RCR Wireless article reports:

Sprint Nextel Corp. and T-Mobile USA Inc. urged the Federal Communications Commission to embrace a plan to allocate vacant TV channels -- known as white spaces -- on a fixed-licensed basis for wireless backhaul services.

And quotes Sprint & T-Mobile from their FCC filing (italics are mine):

"Because backhaul comprises a significant cost for wireless carriers, and incumbent local exchange carriers’ special-access charges are exorbitant, Sprint Nextel and T-Mobile must find more affordable alternatives to the ILECS’s special-access offerings.  Despite this need, the amount of spectrum in the lower bands that is realistically available for the provision of wireless backhaul services has declined dramatically over the years. As wireless carriers expand the development of their 3G and 4G wireless networks, the need for reliable and cost effective backhaul will increase.”

Doesn't anyone see how ludicrous this is?

Cell sites are fixed locations.  The vast majority are in urban or suburban neighborhoods.  Their bandwidth requirements will continue to grow, indefinitely.  In any rational world, one would purchase dark fiber to most of these sites.  In the irrational real world that is the US today, one could at least raise hell about the existence of exorbitantly priced special-access services and the lack of dark fiber.

The root problem is we have given Verizon and AT&T privileged access to the public right-of-way, without requiring them to sell dark fiber, or lease dark fiber, or offer anything less than high level service bundles at "exorbitant" prices.

What would it cost for dark fiber to 95% of these sites, if it were part of a community-wide dark fiber build like that being planned in Singapore and already substantially built in many parts of Stockholm?  Construction costs for point-to-point fiber are 20% or so above those of passive optical networks (PON), but the cost of the dark fiber alone (no electronics) is perhaps 80% of the total construction cost, so it's all in the same ball park.  Verizon has directly and indirectly reported various costs for their FiOS construction (FiOS is a PON network), but all current estimates are below $1000 per home passed.

Even at $5K or $10K, the payback for a mobile operator would be measured in months.

Why isn't anyone talking about the root problem?  We've given service providers privileged, i.e. monopoly, access to our public right-of-way.  That might have made sense in 1900, but today we should only give such access to infrastructure providers. 

The only natural monopolies are the right-of-way, poles and conduits and, perhaps, the dark fiber. 

Let's have multiple service providers competing to offer services over dark fiber infrastructure, put wireless backhaul on fiber, and save the "whitespace" for innovative new services.