Communications

Fiber versus 4G

Yes I know these are extremes and in no way representative, but they both passed through my attention stream within a few hours of each other and I couldn't resist.  :)

I love wireless, for many reasons and many applications, but there is nothing like fiber for enormous capacity (and the attendant speed and latency advantages).

Citylink Fiber in Albuquerque - now that's fast

At this year's Freedom to Connect conference, I had the pleasure of speaking with John Brown who runs CityLink Fiber in Albuquerque New Mexico.  They offer fiber connections to businesses (everything from dark fiber on up) and, in residential parts of downtown Albuquerque, they offer 100 Mbps and 1 Gbps fiber Internet connections.

Of course it's hard to get a single TCP connection to utilize even 100 Mbps but if you're running many things at once and/or have several people at home using the net at once, a 1 Gbps connection can't be beat.  Look at this speedtest by a residential user on their network:

Boy am I jealous...

 

Freedom to Connect (Wash DC right now)

I'm at David Isenberg's Freedom to Connect Conference in Silver Spring MD, i.e. in Washington DC, almost.

There no coverage of wireless, but an excellent roster of speakers and panelists on the Fiber and on the political layers both in the US and internationally.

I particularly liked this morning's panel of six separate entrepreneurs who are running local fiber networks w/o government subsidy:

Big Enough to Succeed: small carriers at the leading edge — entrepreneurial (non-Municipal) carriers show a fourth way (after Telco, Cable and Muni) to the future of connectivity. (60 min)

• John Brown, CityLink Telecommunications
• Gary Evans, Hiawatha Broadband Communications
• Ken Johnson, Conneaut Telephone Company
• Pat Kennedy, Lit San Leandro
• Levi Maaia, Full Channel
• Leslie Nulty, ValleyNet

They are each building fiber access networks independent of the incumbents or the Federal or local government (although all have to deal with local governments of course). Each story is different, but that's excellent.  It means there are multiple paths forward.

Submarine cables

While I have no direct involvement with underseas fiber infrastructure, I've long been interested in the spread of communications, especially to the developing world, so I've tracked international submarine cable deployments for many years.  Today, I was looking at the new submarine cable directory from the Submarine Telecoms Forum and a friend looking over my shoulder said "Wow, where does that come from?"

So if you also want a brief distraction from your busy day, here are the submarine cable sites I follow:

First is Greg's Cable Map.  This is the personal project of a South African technologist, Greg Mahlknecht.

Most of Greg's initial data came from Wikipedia, which not surprisingly has extensive lists of both international submarine cables and domestic cables (those serving just one country).

But for cable by cable data, with maps, I love the the Submarine Telcoms Forum's Submarine Cable Almanac.  It has detailed data and individual maps for over 200 cables systems.  So where overview maps like Greg's above give you a global picture, SubTelForum's Almanac has per-cable maps like this:

FCC's broadband measurements program & Verizon FiOS

If you are interested in the FCC's broadband measurement program or in Verizon's FiOS service, here's some info.  I've been a FiOS customer since July 2005 and I've been an FCC measurement site since their program began in December 2010. Until last month, the monthly reports routinely showed rock solid performance at the advertised rate of 15/5 Mbps with less than 15 ms delay and less than 0.05% packet loss.

Beginning last month and more obvious in this month's report we saw our service degrade and then come back. Of all the measurements, it was delay and packet loss which best reflected what we were sensing.

During the entire period, bandwidth measurements were unaffected:

But over some weeks, something seemed to be happening. Some transactions seemed to take longer than expected. Well, it's visible when you look at latency and packet loss. Delay was creeping up above 20 ms and then above 30 ms with packet loss going up to and above 1%.  Then on Feb 24th, Verizon did something (split a PON segment? added capacity on a backhaul?) that fixed the problems.  Look at these graphs:

Seeing this with Verizon and SamKnows (the FCC's measurement contractor) was reassuring as, at netBlazr, I'd already decided the most sensitive way to measure performance was by looking at latency and packet loss.  We find the open source program SmokePing is the single most useful measure of our network's performance -- a single graphic shows both latency and packet loss.

As an example, here is a SmokePing graphic for a netBlazr member that was having problems, showing before and after the problem was fixed. If only I had been monitoring SmokePing at that time, I could have found and fixed this problem on Monday rather than Tuesday afternoon.  :(   We know better now!

Attending ISP America 2012 in Orlando

I'm off to attend the WISPA conference ISP America 2012 at the Hotel Coranado at Disney World.

If you are attending, please say hello.

I'll be speaking in two marketing sessions (Marketing to SMB's and Bigger on Wednesday and Selling Fat Pipes to Phat Clients on Thursday).  

Then on the Thursday Afternoon I'm participating in the Knowledge Exchange - Technical which is described as "A panel discussion on lessons learned by experienced technical leaders.".  Hopefully I can contribute some theory as the other panelists are likely well ahead of me in practical experience!

 

TMC video interview: asking me about netBlazr

I was recently interviewed by TMC. Check out the full interview by clicking here:

 

ISOC Meeting in DC: Mesh Networks and the Struggle for Universal Internet Access

Next Thursday March 1st from 9:00am to 11:30am, I'll be participating (albeit remotely) in a meeting of the Internet Society of Washington DC along with Bob Frankston and Preston Rhea.  

For those in the DC area, the meeting is at:

SRI International
1100 Wilson Blvd. 
Suite 2800
Arlington, VA 

and Free Eventbrite registration is at:  http://www.eventbrite.com/event/2970182897

For the rest of us, the event will be webcast live at:
          http://www.isoc-dc.org/isoc-dc-tv/

Here's the scope:

Connecting Everyone! Mesh Networks, Public Internet and the Drive Towards Universal Access

Should the Internet be considered public infrastructure? What are the best ways to increase access to the underserved? What are the impediments to more effecient and economic access for everyone?

The Internet is a major force in the world’s economic and political systems, as well as in how people live, work and play in their daily lives. With over 2 billion users worldwide, the Internet still has huge capacity for growth and users have tremendous opportunities today to leverage the technology to develop game-changing innovations that could equally radically change the communications landscape. The Internet is integral to GDP growth, economic modernization, and job creation, generating over 10 percent of GDP growth in the past 15 years in the countries studied.

Providing the opportunity that comes with Interent access to everyone will empower millions and millions of people to leverage the information economy to improve their lives and their communities.  

Please join the DC Chapter of the Internet Society for an important discussion featuring an All-Star panel about increasing internet access everywhere. Free Eventbrite registration required - http://www.eventbrite.com/event/2970182897

Panelists:

Bob Frankston – Frankston.com

Bob Frankston is the co-creator with Dan Bricklin of the VisiCalc spreadsheet program and the co-founder of Software Arts, the company that developed it. Frankston graduated in 1966 from Stuyvesant High School in New York City and in 1970 from M.I.T. Frankston has received numerous honors and awards for his work:

• Fellow of the Association for Computing Machinery (1994) "for the invention of VisiCalc, a new metaphor for data manipulation that galvanized the personal computing industry"
• MIT William L. Stewart Award for co-founding the M.I.T. Student Information Processing Board
• The Association for Computing Machinery Software System Award (1985)
• The MIT LCS Industrial Achievement Award The Washington Award (2001) from the Western Society of Engineers (with Bricklin)
• Fellow of the Computer History Museum

In recent years, Frankston has been an outspoken advocate for reducing the role of telecommunications companies in the evolution of the internet, particularly with respect to broadband and mobile communications. He coined the term "Regulatorium" to describe what he considers collusion between telecommunication companies and their regulators that prevents change.

Brough Turner – Founder, netBlazr, Inc.

Brough Turner is a well established communications industry engineer and entrepreneur. He founded netBlazr to dramatically change the landscape for broadband Internet access in the US. Previously Brough was founder and CTO of Natural MicroSystems (IPO 1994) and NMS Communications, building several successful businesses in fixed and mobile communications equipment. He is an electrical engineering graduate of the Massachusetts Institute of Technology.

While his leading interests are technology and innovation, his career has included roles in engineering, operations, finance, marketing and customer support. He writes and is quoted widely on telecommunications topics in trade and general business publications and he is a frequent speaker at telecom industry events around the world. From 2001-2008, Brough focused on wireless infrastructure and mobile applications. His 3G and 4G tutorials are widely popular (Google ‘3G Tutorial’ for more info). Brough blogs at http://blogs.broughturner.com on the technology, economic and social issues of communications at the intersection of telecom, mobility and the Internet.

Preston Rhea – Open Technology Initiative, New America Foundation

Preston Rhea is a Program Associate for the Open Technology Initiative at the New America Foundation. He supports OTI's mission of digital justice for its Broadband Technology Opportunities Program work with research, analysis, writing and program assistance. Preston also researches and writes on community-based communications and technology activism. Before joining the New America Foundation, Preston spent a year in Beijing, China working for an internet content delivery network. He holds a bachelor of science degree in electrical engineering from the Georgia Institute of Technology, as well as a Spanish minor and an International Plan certificate. Preston also studied electrical engineering at the Universitat Politècnica de València in Valencia, Spain, and spent five years in several countries working with the global student-run organization AIESEC.

 

Fiber problems cause six hour outage

On Friday evening, netBlazr had it's first major outage, unfortunately lasting six hours. In the end, the problem was with one of the fibers in Cogent's riser cable inside the John Hancock Tower. It was not a simple cut. That would have been obvious. Instead we had marginal light levels, perhaps due to a nick, a overly-tight bend or a poorly assembled connector.  

For those that are interested, I wrote up my Friday evening adventures in summary form, followed with more detail for those who are really, really interested. Then this morning I wrote my followup list, again in the netBlazr blog for those who are really interested in the day-to-day issues of being an ISP.

Happy Telephone Patents Day

Mostly we hear about mobile phone patent wars, but today is special. It's the anniversary of telephone patent disputes.

On February 14, 1876, Alexander Graham Bell filed a US patent application for his telephone. On that same day, Elisha Gray filed a "patent caveat" for his telephone.

In 1876 a patent caveat was a preliminary document not unlike today provisional patent application.  The point is they both went on record the same day and that was the start of a long patent dispute.

So today is the 136th anniversay of the start of the unfortunate (or beautiful, depending on your point of view) lasting relationship between telephony and patent lawyers.

 

MIMO is a superset of radio vortices

Back in October I wrote a blog post on radio vortices after reading an article that suggested Twisting Radio Waves Could Give Us 100x More Wireless Bandwidth. I postulated that this might be a viable way to get additional independent paths for a point-to-point MIMO radio link.

I was wrong!

Shortly after I wrote that blog post, Ove Edfors pointed me to a paper that he wrote with Anders J. Johansson, Is orbital angular momentum (OAM) based radio communication an unexploited area? In this paper they prove that radio vortices are a subset of MIMO. Specifically, they show that OAM radio communication, i.e. using radio vortices, is a sub-class of traditional MIMO communication with circular antenna arrays. So, if you have a set of antenna elements that can create radio vortices, the calculations inherent in a MIMO radio system will automatically create and use vortices to the extent they are useful.

Unfortunately they also proved vortices don't give us the additional independent paths I'd hoped for. To get spatial multiplexing gain, MIMO needs additional independent paths, a.k.a. multi-path. Multipath is easy to come by in an office environment (due to reflections from walls, ceiling, floor, filing cabinets, etc.). But in an outdoor point-to-point link with directional antennas there are no reflections, so the only independent paths we've been able to exploit are those resulting from polarization differences (e.g. horizontal vs vertical). This remains the case - radio vortices don't help.

At very short distances, with widely spaced antenna elements at either end, you can get multiple independent paths, just due to the separate fields radiated by separate elements, but this separation gets jumbled as soon as the two ends of a link are separated by more than the Rayleigh distance, as shown in this graph.

At 5.8 GHz with a total antenna apeture of 30 cm, the Rayleigh distance is about 3.5 meters. Since I'm interested in point-to-point links that are typically over 50 meters, we remain limited to two spatial streams. There are no additional paths due to twisted vortices.

 

MassTLC Innovation Unconference

I'm at the MassTLC's annual Unconference. I've been to each of the previous events and each year it gets larger.  I'll post on my Goggle+ account (that also propagates to my Twitter stream - @brough - and my Facebook stream with the has tag #MassTLC.

Valuing the white spaces above 3 GHz

Today, significant money and political capital are being expended to obtain and hold onto usable license-exempt access in the TV white spaces. These efforts are important for applications today but there’s a follow-on spectrum initiative that, if successful, would yield much greater benefits in the long term. We should be seeking similar access to as much as possible of the spectrum above 3 GHz, almost all of which is dramatically under-utilized today.

Throughout the 20th century and right up to today, it's been the case that higher frequencies "don't go as far." But this is the result of technology limits, not ultimate physical limits, and these technology limits are now being overcome.

Within ten years it will be widely apparent that higher frequencies go just as far through the atmosphere, they do just as well at penetrating buildings, and they have other extremely important benefits that lower frequencies lack.

Among their many advantages, directional antennas are smaller and more economical at higher frequencies. Directional antennas reduce received interference and facilitate spatial reuse, thus vastly increasing the utility of higher frequency spectrum.

What’s more, it's easier to send high-speed data because there is more spectrum available at higher frequencies. We'll never be able to send 1 Gbps over any real-world 6 MHz TV channel but, above 3 GHz, we can easily find 200 MHz of spectrum that's temporarily vacant and that's enough to carry more than 1 Gbps of data even with today's technology.

For the next decade or two, TV white spaces will continue to be important for penetrating foliage, but even with foliage, the physics of what is possible differs from 20th century experience. In the future, the real action will be above 3 GHz.

Finally, while it's never easy to persuade existing licensees to accept secondary users in “their spectrum” even while it’s idle and they are non-interfering, it should be easier to fight the political battles now, when most people don't realize the long term value of spectrum above 3 GHz.  Now is the time we should be seeking license-exempt access to as much as possible of the white spaces above 3 GHz.

 

Details for the technically inclined

All photons (light or radio waves of any frequency) go at the same speed (the "speed of light"). In our atmosphere, photons at frequencies above 10 GHz are subject absorption because they excite resonances in atmospheric molecules like water vapor or oxygen. But the atmosphere is transparent to radio signals between 30 MHz and 10 GHz so, with a clear line-of-sight, radio signals at 8 GHz go just as far as signals at 700 MHz or 50 MHz [2].

This physical fact is sometimes missed because the Free Space Path-Loss (FSPL) equation (see: http://en.wikipedia.org/wiki/Free-space_path_loss) commonly used to calculate radio frequency (RF) transmission losses actually encapsulates two effects. These are 1) the actual path loss (which is independent of frequency) and 2) the receiving antenna aperture which is based on wavelength. Thus the FSPL equation assumes smaller antennas for higher frequencies and of course, smaller antennas collect less energy. With equal antenna apertures, unobstructed line-of-sight radio transmissions are frequency independent, even with 20th century technology.

The problems that have favored lower frequencies are reflection, refraction, polarization and diffraction. Higher frequencies have shorter wavelengths and shorter wavelengths signals are more easily scattered. Scattered signals that reach the receiver have taken a longer path and thus arrive a little later. With 20th century technology, these delayed signals (called "multi-path" signals) were just part of the noise degrading the primary signal. Now with Multiple Input Multiple Output (MIMO, see: http://en.wikipedia.org/wiki/MIMO), it's possible to decode multi-path signals, remove them from the noise, align them in time and add them to the primary signal - multi-path signals are no longer a deficit but actually improve system performance!

MIMO technology only began to emerge in the mid-1990s but it is now an option in the latest Wi-Fi, WiMAX and LTE specifications. MIMO uses multiple radios and higher order MIMO requires increasingly sophisticated calculations, but early (2x2) MIMO systems are already widely deployed in 802.11n consumer WiFi products and continued semi-conductor progress (following Moore's Law) will make MIMO calculations and additional radios ever lower cost.

Also inherent in higher order MIMO is beamforming and beamsteering.  As the number of radios and antennas in a MIMO system increases, the system is able to simulate tighter and tighter beams, providing ever more spatial reuse of spectrum and more range or capacity for individual connections. However, tighter beams require more wavelengths of separation between the outer most antenna elements.  Again higher frequencies have shorter wavelengths, so antennas that support tighter beams require less space at higher frequencies. For example, a 10 degree beam at 700 MHz requires an antenna ~10 feet across.  To do the same at 8 GHz, the antenna need only be ~10 inches across.

Long term, the spectrum above 3 GHz will be more valuable than the spectrum below 3 GHz.  Let’s get license-exempt access to these white spaces now, while the political stakes are still (relatively) low.

Some useful references

[1] Spectrum occupancy measurements for frequencies below 3 GHz, see: http://www.sharedspectrum.com/papers/spectrum-reports/.  Fewer measurements have been made above 3 GHz but those done show negligible occupancy (certainly when compared with bands below 3 GHz), for example see:  http://www.sharedspectrum.com/papers/spectrum-reports/

[2] This graphic http://en.wikipedia.org/wiki/File:Atmospheric_electromagnetic_opacity.svg shows the transparency / opacity of our atmosphere at different wavelengths.  Note that 10 meters is 30 MHz and 3 cm is 10 GHz.

[3] While the total atmosphere (as seen from space) is highly absorbing above 10 GHz, shorter range point-to-point connections are still possible as this more detailed spectrum shows for the range 10 GHz - 1000 GHz: http://www.omlinc.com/library/other-references/atmospheric-absorption-of-millimeter-waves.html

[4] NIST, Electromagnetic Signal Attenuation in Construction Materials, NISTIR 6055, http://fire.nist.gov/bfrlpubs/build97/PDF/b97123.pdf  Note that ordinary window glass is essentially transparent to RF at the frequencies tested (500 MHz to 8 GHz) while most other building materials provide substantial attenuation.  One caveat: as this study was done to help design RF-based measurement device for the construction industry, they post-processed their data to remove delayed signals.  In other words, MIMO communications systems will do considerably better than these measurements suggest.

[5]  Okamoto, Kitao & Ichitsubo, Outdoor to Indoor Propagation Loss Prediction in 800 Mhz to 8 GHz Band for an Urban Area, March 2009, http://ds.lib.kyutech.ac.jp/dspace/bitstream/10228/2399/1/IEEE.pdf  Detailed measurements demonstrating frequency independence for signals penetrating real life buildings.

[6] Perras & Bouchard, Fading Characteristics of RF Signals due to Foliage in Frequency Bands from 2 to 60 GHz, http://horwitzinternational.com/PDF%20Files/Trees%20and%20801_11.pdf  Careful measurements show signal attenuation peaks when RF wavelengths equal leaf size and then falls off at even higher frequencies.

[7] RF Engineering for Wireless Networks: Hardware, Antennas and Propagation, by Daniel M. Dobkin, PhD, ISBN 0750678739 has useful chapters on signal propagation in the atmosphere, in the environment and in buildings.

 

Modulating information onto radio vortices

It's well understood that photons carry both energy and linear momentum. It's less widely understood but known for over 100 years that photons can also carry angular momentum. But it's only in the past 20 years that light orbital angular momentum has been studied, and only in the past year or so that serious work has been done with radio frequency photons. Recently I posted the URL of a Discover magazine article, Twisting Radio Waves Could Give Us 100x More Wireless Bandwidth in my Google+ stream (with copies to my Twitter and Facebook streams).

That post prompted one public comment and several off-line email discussions, so let me elaborate. While I don't expect 100x wireless capacity any time soon, this is a real effect that might be used for 2x, 4x or 8x gains within the next decade. Wikipedia has the best visualization:

The relevant scientific article is Radio beam vorticity and orbital angular momentum by Bo Thidé, Fabrizio Tamburini, Elettra Mari, Filippo Romanato & Cesare Barbieri available at arXiv.org.

Thidé et al formed their twisted radio waves by reflecting an RF source off an approximately spiral-formed reflector. This is a cool experiment but you'd need a set of reflectors, one per channel, each with slightly different depths of their spirals to begin making wireless capacity gains. To make this commercially viable, we need an antenna that can create several different orders of vortex in the same physical space.

If someone figures this out, then the resulting separate channels could be used directly with MIMO electronics and an antenna that fits in limited space, much as 2x2 MIMO systems use horizontally and vertically polarized antennas that fit in the same physical space.

It will be very interesting to see if someone can come up with an antenna structure that can produce different order spiral vorteces in the same space. In any event, this is an area to watch.

Wireless ISP Conference(s) - WISPAPalooza 2011

I'm at the Wireless ISP Association's conference in Las Vegas.  The conference name: WISPAPALOOZA with these sessions. Following the main conference, there are several vendor conferences including the Ubiquiti World Conference on Thursday and a MikroTik User Meeting (MUM).  Since netBlazr is using both MikroTik and Ubiquiti gear, this is a good conference to attend.

I'm also speaking tomorrow and moderating a session on Antenna Propagation: Principals and Possibilities.

I'll be taking notes on Twitter, but on the netBlazr twitter account.  If you interested in, follow me at:

http://twitter.com/netblazr

 

No spectrum shortage, just an allocation problem

As a new study from Citi Investment Research & Analysis make clear, the US does not have a spectrum shortage. We've just allowed a relatively small number of carriers to control the spectrum. Quoting the study's summary:

Today, US carriers have 538MHz of spectrum. And, additional 300MHz of additional spectrum waiting in the wings. But, only 192MHz is in use today.

Perhaps if we had an effective "use it or lose it" policy in place, or a heavy tax on unused spectrum a more vibrant market for this spectrum would emerge.  But today, the problem is not a shortage of spectrum but the fact that what's out there is not being utilized.

Obviously things vary by geography, but Citi's summary is completely justified. Their methodology is thorough both as to who owns what and what is deployed county-by-county for 3100 separate counties. Here's the summary of what's in use:

and here are the details on what's currently owned by US carriers:

So why would we repack the TV broadcasters and auction off that spectrum when we've just finished putting in place unlicensed access to TV white spaces?  Unlicensed spectrum will be heavily utilized while more exclusively owned spectrum will just add to the pool of under utilized resources.

Limitations of Carrier Grade NAT, and some workarounds

Qtel, the largest carrier in Qatar (and nearly the only Internet provider) appears to connect all their users (~600K) to the Internet through just one or a very few public IPv4 addresses.  82.148.97.69 was their single public address in 2006-2007.  How can network address translation (NAT) put all those users through just one IP address?

Well, both UDP and TCP include 16-bit port numbers, so for each IP address there are 2**16 = 65,536 unique IP address/ TCP port number combinations which can support network address translation (NAT) for more than 65K simultaneous connections.  Of course, individual browser sessions may open multiple connections so each on-line user can tie up several such connections, even several dozen.

But there are more unique fields that come into play.  On the Internet side of the NAT, each active IP session has five fields of interest:  source IP address, source port, protocol, destination address; destination port.  So you can use the same source address and port number for multiple independent sessions if the sessions are going to different destinations or using different protocols. With this approach, one public IP address can clearly support millions and millions of connections.

But what if most of your users are trying to reach just a few destination addresses?  That’s a problem Qtel had (or at least their users had) when accessing Wikipedia in 2006.

One obvious solution is to devote more than one public IP address for the whole country!  Indeed, if a few hundred people are attempting to access the same address simultaneously, then having a few hundred IP addresses (for the whole country!) would do it. Of course, if the destination IP address is that of Google’s search bar, you might need more than a few hundred IP addresses.

There is another path, one which Geoff Huston alludes to in a recent article on the IPv6 transition.

… we may expect to see a push to re-architect content into Content Distribution Networks that have points of presence in major access networks.

Today, major content distribution networks (CDNs) like Akamai, Amazon Cloudfront, Limelight and Google each maintain thousands of servers close to the edges of the Internet.  These servers cache popular content close to where the users are.

When many, many users in Qatar simultaneously try to access the same content, the load on Qtel’s NAT boxes could be dramatically reduced by putting the appropriate CDN’s caching servers within Qtel’s private address space.  Then requests for popular content wouldn’t even go through the NAT, reducing total traffic and dramatically reducing the occasions when two users are going through the NAT to the same public IP address.

It’s not in line with the Internet’s original end-to-end design, but as IPv4 addresses get more expensive, we can expect to see more and more use of NAT and CDN servers moving into carriers' private address domains.

 

White spaces above 3 GHz

Yesterday, I gave a pitch for more access, on a secondary use basis, to the vast swaths of spectrum above 3 GHz - spectrum that is largely unused in most locations most of the time.  Then I gave an update on netBlazr and how we are using the existing license-exempt spectrum in the 5 GHz band (over 500 MHz in the US).

I'm sorry there's no audio, but here's the deck I used:

White spaces above 3 GHz and an application

View more presentations from Brough Turner

Wireless conferences in Austin Texas this week

While the big event is Internet Telephony Expo, there are two other conferences and a startup pitch fest going on in Austin this week.  I'm leaving in the early morning and will be speaking twice tomorrow afternoon and then pitching netBlazr on Wednesday afternoon.

Tomorrow at 2:30pm I'm appearing on a panel at 4G Wireless Evolution. The panel is entitled "Wireless Fixes Access with Redundancy." The other panelists are from Sprint and Wave2Wave. Sprint is using point-to-point wireless for backhaul from their cell sites since they can't get competitive access to the fiber backhaul that AT&T and Verizon have. Wave2Wave is wireless Internet Service Procider. As netBlazr's approach is rather different, in both our business model and our use of fixed wireless technology, this should be an interesting panel.

Then at 3:30pm I'm the sole speaker for a session in the Super WiFi Summit just down the hall. The description is:

White Spaces: The Radio Evolution

Tuesday - 09/13/11 • 3:30-4:15pm

Brough Turner , Founder , netBlazr.com

Smart antennas and smart radios, Cognitive Radio and Beam Forming are on the verge of being incorporated into product. As we head toward these technologies, the opportunities exist for new models of service sharing and interconnection to deliver broadband solutions.

My talk will be a mix of what's possible today, what will become possible with next generation silicon and how netBlazr is using today's and tomorrow's technology to do an end run around the existing broadband market.  Of particular interest is where additional white spaces would be useful - hint it's not just in the TV bands.

Finally, there's Startup Camp on Wednesday afternoon and early evening.  netBlazr is one of five new companies choosen to present and I'll be doing the honors.  Bob Metcalfe is the keynote speaker and there will be a panel of judges to provide critiques on the startups' pitches.

YouTube video flow control actually causes packet loss (and retransmission)

There's an interesting article in the April 2011 issue of ACM SIGCOM Computer Communications Review. The study by Shane Alcock and Richard Nelson (also accessible here) looks at the way application level algorithms on YouTube's video servers interact with TCP and the actual buffers in a retail DSL delivery system.  I found this particularly interesting as I had already looked into the burstiness of other kinds of traffic in a set of posts in late 2009 (this post in particular).

Here's the way YouTube servers deliver data (at least as of January 2010) to high capacity clients:

By high capacity client I mean the client device has enough buffer space to absorb whatever data the video server wants to send and the client's broadband Internet access service has enough capacity to avoid congestion during delivery of these video bursts.  Unfortunately, the delivery rate YouTube was using in January 2010 was enough to overwhelm the data buffers in typical DSL infrastructure as this graph illustrates.

It appears the YouTube servers send data in blocks of 64 KBytes or 128 KBytes (or multiples thereof) and these blocks can exceed the available buffer space in the DSL delivery infrastructure - mostly likely the buffers in Broadband Remote Access Server (BRAS).  At least one BRAS I was familiar with in 2004-2005 had 200 ms of buffering, obviously not enough for one burst per second.

The good news is the net effect was only a 1% increase in the total data (due to retransmission) so this isn't a big problem and the authors discussed the issue with Google where engineers were working on a solution (in 2010).  What's interesting to me is the extent to which the sizing of buffers for IP infrastructure is still a complex issue and the ease with which a simple application level decision can go wrong.