Communications

Why there is no Internet QoS

In 2006, I wrote a short blog post on why there is no Internet QoS and likely never will be.  That post is a continuing source click throughs and email inquiries so, when Jon Arnold asked me to write a guest article for IP Convergence TV, I thought I'd tackle the definitive story on Internet QoS.

Here is the new article: Why there is no Internet QoS.  It may or may not be definitive, but I have filled in the engineering and economic background, and I investigate where QoS has been deployed in networks in order to identify where there may be QoS related opportunities for ISPs.

My conclusions:

• There is no Internet QoS today, and it’s unlikely any complex QoS scheme will ever be added to the Internet as a whole.
  
• To the extent next generation networks with rich QoS are deployed, this will only happen within walled gardens, not as part of the public Internet.
  
• But there may be opportunities for ISPs to create a simple premium service that could generate incremental revenue.

China's 3G license delay is a smoke screen

Last week there was a flurry of stories about China's 3G plans after Jonathan Dharmapalan of Ernest & Young was quoted as saying he expected it to take 12 to 24 months from the start of China's commercial TD-SCDMA trials, i.e. from now, until 3G licenses were issued.  But there was little analysis or comment on what's really happening.

3G licenses are a formality.  They delay the deployment of 3GSM & CDMA 2000 which could otherwise happen rapidly — just plug new cards into existing radios and offer established handsets (already being manufactured, in China, for the world market).

China 3G is happening, without licenses.  China Mobile Group already has deployment-scale TD-SCDMA radio networks on assigned, but not "licensed" frequencies, in eight cities.  The bottleneck is TD-SCDMA handset silicon.  Commercial trials were finally enabled by the recent delivery of an initial batch of 60K handsets.  This also allows China to meet, at least technically, their pledge to have 3G service in time for the Olympics.

Dharmapalan is speculating that it will take 12-18 months of commercial trials to get the bugs out and the system to scale.  That's plausible.  If so, it will be 12-24 months before "licenses" are granted, i.e. before 3GSM and CDMA 2000 deployments are allowed to proceed in parallel with TD-SCDMA.

The problem is China's patent position.  The core patents for 3G and for 4G are already held by the likes of Ericsson, Nokia, and Qualcomm.  TD-SCDMA provided a way for China to obtain patents on a specific 3G implementation and the size of China's market makes that implementation interesting to Ericsson, etc.  Since China missed the boat on core patents for both 3G and 4G, expect TD-SCDMA and it's 4G successor to exist on a parallel path for the next 10-20 years.

The good new is commercial TD-SCDMA deployments have finally started.

I'm most interested in 3G applications.  China has a vibrant market in value-added services for 2G.  As more and better TD-SCDMA handsets get deployed, we should see some really interesting innovation coming out of China — innovation that will be applicable to any 3G technology, anywhere in the world.

Mobile video-on-demand Yes! Mobile broadcast TV not so hot

I have an article, Going Mobile (TV), that's recently been published by MobileIN, a wireless and mobile information site.  In it I basically argue that major investments in mobile TV broadcast capability are less likely to pay off than investments mobile video-on-demand.

The biggest trend in commercial television viewing is personal video recorders like TiVo.  People want to watch TV content when they want, not when broadcasters schedule it.  The only exception is major sports events (the Superbowl or World Cup matches).  Even the evening news is frequently rescheduled for later in the evening.

The second relevant trend is growth in YouTube and similar web-based video content.  Broadcast TV went from 2-3 channels in the 1950s to hundreds of channels on a typical cable system today.  But consumers are also interested in the long tail of millions of videos that can only be served over the Internet today and, potentially, over the mobile Internet in the future.

Finally, survey's of early adopters of mobile video show music videos, movie trailers, weather, sports action clips, comedy videos, cartoons and amateur video shorts – typically a few minutes long at most – are the most popular content. In addition, it appears 85 percent of mobile video viewers watched viral videos (content sent or pointed out by others) rather than content they found themselves.

All and all, mobile consumers are looking for video -on-demand, not pre-scheduled broadcast TV.

So what's the logic for massive investments in spectrum, followed by even more money in new wireless infrastructure, followed by the need to sell everyone new handsets that can receive the new broadcast mobile TV channels?

Models for Muni WiFi completely neglect technology evolution

Modern travel means interminable waits, but it's a good time for reading. I finally read Wireless Pittsburgh: Sustainability of Possible Models for a Wireless Metropolitan-Area Network by Jon M. Peha, published in February as a working paper of the New America Foundation.

The good news — it’s full of interesting cost estimates and projected subscriber take rates based on specific demographics in Pittsburgh, Minneapolis and Philadelphia.  The paper also examines a range of business models, in detail, from complete monopoly to structural separation (wholesale–retail) to let-the-market-take-care-of-it.

The bad news — all of the models turn out to be extremely sensitive to revenue assumptions, i.e. to the estimates of subscriber adoption and willingness to pay.

The flaws in this study

All of the models compute a net present value based on five years of stable operations, but there is no mention of technology evolution or adoption rates of competing broadband services, i.e. cable and telco (DSL or FiOS) services since this is a US study.  Technology is evolving at a great rate.  You can’t bet on stability.

During the past five years we saw WiFi go from 11 Mbps to widely deployed 54 Mbps systems and bleeding edge (pre-standard 802.11n gear) systems doing well over 100 Mbps.  The last five years also saw costs decline to the point where we see widespread deployment of WiFi by individual consumers, a significant percentage of which are running open WiFi hotspots. 

On a recent drive through three different neighborhoods in Portland Maine, I was interested in looking up real estate information on the web.  On each of a half dozen occasions, I was able to find a open WiFi hotspot within one city block of deciding I wanted to connect.  In January, I was in north New Hampshire and spent two nights at small motel (not part of any chain) in Littleton, NH.  They had no Internet on offer, but a quick check for WiFi signals revealed two within range of our room.  If you don’t like my anecdotal information, look at the WiFi hotspots that have been mapped by Navizon.  It’s very different than five years ago.

No matter how much it simplifies the analysis, you can’t bet on stability.

What might happen with WiFi in the next five years?  The latest WiFi specifications add multiple input, multiple output (MIMO) support, additional modulations and other goodies.  As low cost WiFi routers incorporate these advances we’ll see speeds go over 300 Mbps, but more importantly MIMO technology increases both range and directionality.  This means overlapping systems work better (despite their overlap) and the signals from isolated systems reach further.

If you’re worried today’s open systems will be locked down, then spend your time thinking about schemes like FON which offer more secure ways for consumers to share WiFi bandwidth.

If you’re worried consumer solutions won’t reach the inner city, perhaps someone needs to relook at where WiFi has already been deployed, and then forecast what might happen over the next five years, given the cost of a basic PC is approaching that of a television and the Cable and Telco duopolists both push triple play bundles. 

Don’t short change technology evolution. 

Dark fiber policy — my talk at eComm

At eComm 2008 in March, I gave a short presentation (15 minutes including questions) on dark fiber as a key element of broadband policy.  Brad Templeton has a good summary in the last four paragraphs of this post.

Recently Lee Dryburgh posted all the eComm presentations on Slideshare. Since my presentation was mostly pictures, you need the script that went with the slides. The only way I can see to support this on Slideshare is to add one comment per slide, which I have now done.  If you are interested, go to the version on Slideshare.net, click on the "Comments on Slide 1" tab and then click forward from there.

I discussed the clear line between rights-of-way, conduits, poles and dark fiber - all of which are long-lived elements - and the technologies that light up fiber or run traffic over it.  The latter get obsolete very quickly, so it makes sense to foster rampant competition at these layers, or give control to users. The idea is those that want the latest electronics ought to be able to get them.

Dark Fiber

Turns out there are multiple examples of communities and even whole countries (Sweden) that have fostered the availability of dark fiber.  Not surprisingly, these locations lead in broadband performance (and in price-performance!).

NY Times grossly misreads WEF report

Today's New York Times includes an article by John Markoff entitled "Study Gives High Marks to US Internet."  But either John Markoff is fuzzy about exactly what the Internet is or he didn't actually read the report.  His title is way off base.  He did interview a few people who are quoted in the latter part of the article, so there is some information in the article.  But he's done a major disservice for the many who read only the title or perhaps first paragraph.

The study in question is the Global Information Technology Report recently published by the World Economic Forum.  While the printed report costs money, the summary is on-line and an interactive version gives access to all their data.

This is a report on information technology not specifically on the Internet (just look at the title).  The researchers measured 68 different attributes of information technology, only a few of which directly pertain to the Internet, e.g.

Attribute	US Rank
Internet users (per 100 pop.)	7
Internet access in schools	12
Broadband Internet subscribers	17
Internet bandwidth	19

To his credit, the first section of the report is about "network readiness" and the US is ranked #4, however the study's definition of network readiness includes all sorts of features of the broad business environment, regulatory aspects and computing infrastructure.

Prior to this, I remember John Markoff only for the book Takedown, which I enjoyed.reading even though my sympathies were with Kevin Mitnick.  :-)  So this article is a major disappointment.

Here are the 68 attributes the World Economic Forum study examines:

Visualizing African Internet Connectivity

Here is a great illustration of the not-so-fast or reliable connectivity at African Universities as seen from Trieste, Italy.  Only 46 seconds.

Thanks to Marco Zennaro at the Science Dissemination Unit (via TIER at UC Berkeley).

WiMAX in the US, a limited window of opportunity?

Their first service launch has been delayed, but Sprint Nextel CEO Dan Hesse repeated his vow to blanket the US with a WiMAX network in his talk at CTIA this week.

I sincerely hope he succeeds.  We need as many competing data networks as possible, if we're to see any measure of open mobile Internet access. However, I'm worried.

WiMAX has been quite successful in emerging markets, providing fixed wireless Internet access in countries as diverse as Pakistan, Bulgaria, Nigeria, Georgia, Ethiopia and Georgia (the country!).

The US is another story.  The only significant US deployment is Clearwire's with roughly 400K subscribers, but on a mostly “pre-WiMax” network.  Again, the application is fixed wireless Internet access. 

It’s one thing to use fixed WiMAX to provide Internet access in Pakistan.  It’s another thing to compete for fixed access in the US.  Yes, our telephone & cable duopoly is moving slowly, but they are going after all the more profitable neighborhoods with broadband offerings substantially faster than what fixed WiMAX provides.

What about mobile WiMAX?  Mobile WiMAX is in trials today, using technology and providing performance that the 3GSM community will only see 2–3 years from now.  Sprint clearly hopes to use WiMAX as a springboard past its competitors and past concerns about its declining user base.  Presumably, in the longer term, they hope to converge their dissimilar networks (Sprint EVDO and Nextel iDEN) on mobile WiMAX.  But can mobile WiMAX build a large enough market soon enough? 

Volume drives down cost and cost advantages win in the end — witness Verizon’s announcement that they are jumping ship on Qualcomm’s CDMA evolution in favor of the 3GSM community’s long term evolution (LTE).  Today, GSM networks (GSM/ EDGE/ W-CDMA/ HSPA) have 80% of the world mobile phone market with billions of subscribers and a billion or so handsets manufactured each year.  Right now the entire Sprint-Nextel customer base is ~54 million subscribers.  Perhaps emerging markets will also adopt mobile WiMAX, thus driving up volumes?  Unfortunately emerging markets are even more price sensitive with the high volume application being basic voice telephony plus SMS.  GSM is the lowest cost solution by a wide margin. 

I hope I'm wrong.  I hope mobile WiMAX's performance lead (over LTE) is enough to carry the day.  And, in particular, as I’ve written before, I would very much like to see Sprint succeed, with or without Clearwire, because their WiMAX network would represent yet another source of wireless Internet access.  With four of more networks capable of mobile broadband Internet access, competitive pressures alone should give us what the FCC is currently ignoring, i.e. mobile data plans that are both open and reasonably priced.