I would like to thank Jon Henke for this great opportunity to work with him and the rest of my Digital Society colleagues including Mike Turk, Nick Brown, James Delong, Brett Swanson, and Steve Effros.  It’s been a great team whose individual and combined intellect I value and I will continue to read their insight.

I also look forward to working with my new colleagues at High Tech Forum who are walking libraries of technical insight.  I enjoyed my time here at Digital Society and hope that my readers will add High Tech Forum to their list of sites to read.

The 150 GB cap applies to AT&T’s slower DSL customers while the 250 GB plan applies to AT&T U-Verse and Comcast cable broadband.  Note that for customers not on U-Verse using the older DSL technology, it is unlikely that they have a 6 Mbps connection required for the 4.8 Mbps option because 6 Mbps U-Verse plan is cheaper.  Furthermore, normal DSL is effectively a little faster than 5 Mbps for actual data throughput because the 6 Mbps sync rate has protocol overhead.  I have tested 6 Mbps U-Verse accounts getting 7 Mbps of data throughput on speedtest.net.

Netflix doesn’t give offer a 2.2 Mbps HD option which works better than 4.8P 2.2 Mbps mode, but I’m fairly certain that 3 Mbps connection using the “best” mode could still watch “HD” at sub 3 Mbps speeds with slightly lower quality.  This is what I have confirmed on my home connection.

UPDATE – It should be noted that many of the movies in Netflix are not available in HD which means they’re low bandwidth to begin with.

The good news is that an application like this might bring some sense to people about the benefits of close proximity to cell towers.  People might realize that the closer they are to a cell tower, the weaker their cell phone transmits at.  And because cell phones are around a million times stronger to the person using it than a cell tower, anyone who wants to ignore the overwhelming scientific data that there are no measurable risk to cell phones usage and wants lower radio exposer will need to live near a cell tower.

So does this mean that living far from a cell tower or living in a shielded building is harmful because the cell phone must operate at peak transmit levels?  No, because peak transmit levels are already restricted by the FCC to safe levels.  That means the worst case reading from the Tawkon app simply means the phone is operating at the peak allowable SAR levels.  If it really scares a person that much, they should use a wired or Bluetooth headset.

Note that the sequential capacity measured in MB/sec is frequently confused with the broadband metric of megabits per second (Mbps).  1 MB/sec is 8 Mbps so these SSDs have sequential write speeds of 1760 Mbps.  Furthermore, the real benefit of SSD drives is the random read and write speeds where data has to be retrieved from different parts of the storage device.  Each of these SSD drives have the equivalent random speed of hundreds or thousands of conventional mechanical hard drives with rotating disks.

“We’re happy to work one-on-one with any of your readers to walk through the measurement tool and address any questions,” AT&T spokesman Seth Bloom tells Broadband Reports. “We’re already addressing ways we can make the labels and information on the online tool more clear for customers between now and May…I can also assure you our team is performing checks everyday to ensure accuracy.”

“Other tools may measure at different 24-hour periods than we do, and most likely do not take into account the standard network protocols (e.g. Ethernet, IP) that are used to provide applications and content to our customers via the Internet,”

It’s one thing if a journalist wants to question a company’s statements, but it’s wrong to twist a company’s words and claim that “AT&T says they’re working on meter accuracy“.  Bode’s portrayal of AT&T’s statements suggests that AT&T is admitting to problems and that they are trying to fix them.  But looking at AT&T’s quoted statements above, it is clear that AT&T did not admit there were inaccuracies.  What AT&T said is that they will look into complaints, they will continue ensuring accuracy, and they explained why users might be getting different measurements.

If users measure a different 24-hour interval than the daily metric shown by AT&T’s usage reports, that will produce daily inconsistencies but it has no effect on the monthly metric which is the only relevant number.  There are no daily usage overage penalties and after I checked AT&T’s usage meter, it only shows previous months and the current month’s running tally.  Based on my own usage of around 50 GBs (which is a few times higher than average), I saw nothing that would indicate any problems with AT&T’s usage meter.

As for the complaint that AT&T measurements include IP and Ethernet overhead, that’s sort of like complaining the supermarket sold you a 10 lb watermelon which included the 1 lb of peel and seeds.  With IP and Ethernet overhead, we’re talking about a difference of 2.4% with large file transfers which makes up the bulk of usage.  Most users aren’t even going to see the difference except for people looking for something to make an issue with.  Broadband transparency is good and these details should be in the terms of the contract, but arcane details like IP and ATM PPPoE overhead probably won’t make it on some “nutrition label” because it’s too confusing and insignificant.

There’s been much media criticism of the wireless industry for alleged spectrum hoarding, but the problem isn’t that simple.  Former FCC chief economist Thomas Hazlett who has studied the spectrum market for years stated the following when asked about spectrum hoarding by the wireless industry.

Thomas Hazlett -

The 534 MHz (auctioned to the wireless industry) is way over-stated.

First, through 2009, only about 200 MHz was being used by mobile carriers. That includes 50 Cellular, 120 PCS, and about 20 for SMR (the former dispatch licenses turned into mobile phone spectrum by Nextel).

Second, there were (finally) big auctions in 2006 (AWS) and 2008 (700 MHz). These yielded 90 MHz of licensed spectrum (AWS) and 70 MHz (700 MHz – counting previous auctions in 2002/03). This bandwidth was encumbered. The 700 MHz licenses had TV stations broadcasting on many of them until to June 2009, and still have wireless microphones and (maybe) low power TV broadcasts. It’s not a big issue because most of the bandwidth, owned by AT&T and VZ, is being used for LTE — and this is not yet rolling out. The AWS spectrum is encumbered by government users, and it’s a problem for T Mobile and other carriers who paid the $13.9 billion in 2006. They’re still clearing the band.

Third, there are the 2.5 GHz licenses that Clearwire and some smaller players are aggregating to provide WiMax services. There is up to 190 MHz available here. But it is fragmented insanely, and the companies have to piece together most of the bandwidth via long-term contracts with non-profit educational institutions — Catholic churches, community colleges, etc. How much can be effectively deployed is not publicly known. Clearwire claims that it has something like 90 MHz covering about 120 million pops. And they’re the biggest play, by far, here. Given the licensing problems and the technology problems (WiMax not scaling, not blowing away 3G let alone LTE), and the inherent cost disadvantage in building new networks from scratch (against carriers upgrading 3G to 4G), this spectrum is not yet effectively competitive in the wireless space.

So, when you get down to it, we now have something like this:

190 — deployed by cellular carriers
190 — 2.5 GHz potentially deployable for wireless broadband
160 — likely to be soon deployed by carriers (AWS, 700 MHz)
—–
540

Tossing in the 190 for the 2.5 GHz licenses is dubious, and the fact that the FCC waited so long to auction the AWS and 700 MHz licenses makes this spectrum a future play rather than currently deployed bandwidth. The additional 50 MHz is from AWS-2, and I don’t know how fast the FCC is going to move on that. Could be years. It’s already been four years since the AWS-1 was auctioned.

This isn’t to excuse the comments by Dish Network executives, but Hazlett highlights the fact that the problem of spectrum hoarding is complicated and that it is often overstated.  Bad behavior isn’t the norm in the wireless industry and the major wireless carriers use their spectrum very efficiently and they are in the process of deploying newly auctioned spectrum.  We should address spectrum hoarding and speculation in future auctions but it should not be a reason to halt spectrum waste reclamation in TV broadcasting.

Professor Hazlett offered a detailed proposal for reclaiming that wasted spectrum by eliminating broadcast TV and paying the private sector serve the roughly 10% of our population who don’t have subscription TV from one of the existing MVPD providers.  I offered a slightly different proposal that would leave 40 MHz intact and reclaim 254 MHz of spectrum which gets most of the spectrum back but eliminates the need to create an alternative to broadcast TV which carries the risk of a permanent TV service entitlement.  I like Hazlett’s proposal but I am biased towards my own, but both proposals are vastly superior to the status quo.

This attack highlights a much more fundamental problem with X.509.  A lot of large companies will say “oh but we use more reputable certificate authorities for SSL”, but it doesn’t matter because the fundamental weakness of X.509 allows any one of the many certificate authorities to compromise the entire SSL PKI system.  Any nation (including rogue states) have access to the master keys.  Anyone willing to spend around $40,000 can simply buy themselves access to a root certificate (essentially a “master key”) that would allow them to create any SSL certificate they desire.  Although the terms of the root certification signing authority contractually forbid buyers from abusing their root certificate, it’s a useless trust based on the honor system.

DNSSEC is a new secure Domain Name System (DNS) that also has the ability to replace the fundamentally weak X.509 PKI system.  DNSSEC security is vastly more secure because of the following design principles.

• Only the DNSSEC roots have master keys.  By comparison, there are dozens of root authorities for X.509 and anyone with $40K or anyone who compromises one of the many root authorities have access to the master key.
• Each DNSSEC root doesn’t have a full master key.  The .com root can’t sign for the .ca or .cn root.
• DNSSEC delegates limited signing authority to each domain owner.  Each domain owner can sign their own certificates for their own servers and users, but they can only sign it for their own domain which eliminates the threat of cross-domain signing abuse possible with the current X.509 PKI system.
• Domain owners don’t need to pay hundreds of dollars for each server or user certificate like the current X.509 PKI racket.  Not only does this save companies money, it removes barriers for the adoption of secure communications.  Certificate Authority companies should earn their money providing services of value, not signing a few bits for customers who had no choice before DNSSEC.

Here are some of the more interesting facts from the 2010 report

• There are now 302.9 million wireless subscribers, up 6% from 2009.
• Wireless penetration is at 96%.
• Minutes of voice usage is slightly down in 2010 compared to 2009, from 2.275 trillion to 2.241.  That would indicate an average of 616 minutes per wireless subscriber.
• MMS is up 64% to 56.6 billion messages, which seems to explain a slight decline in voice usage.
• Data traffic is up 110%, and now at 226.5 billion megabytes.
• 78.2 million active smartphones, up 57% from 2009.
• 270 million data capable devices, up 5.3% from 2009.
• 242 million web capable devices, up 1.5% from 2009.
• Wireless enabled tablets, laptops, and modems (not Wi-Fi) is up 14.2% at 13.6 million devices.
• Average data+voice monthly fee is down 2% to $47.21 despite increased data adoption.
• Wireless industry capital expenditures rose 22% from 2009 to $24.9 billion.

From these numbers, it would indicate a healthy segment of the U.S. economy.

Half of that spectrum in broadcast TV is wasted on older MPEG-2 video compression technology, and the former head of the National Broadband Plan Blair Levin has advocated a migration to MPEG-4 AVC (AKA H.264) so that broadcasters can send the same amount of video with the same perceptible quality using half the radio spectrum.

In addition to the compression optimizations, we can again double our spectral efficiency if we got rid of the whitespace waste stemming from an inefficient radio architecture.  Whitespaces can be eliminated if we switched to a single frequency network where all the broadcast towers use the same frequency, but the current ATSC standard used by American broadcast TV uses 8VSB modulation which is not optimized for single frequency networks (SFN).  We would need to switch to OFDM radio modulation (like the international DVB-T standard) to use SFN architecture.  That would require changing out transmitter and receiver equipment, but we can do this when we upgrade to MPEG-4 AVC so that we only need a single transition.

We can go further and eliminate guard band waste if we consolidated the broadcast stations to a single 40 MHz channel.  Consolidating the transmit towers and other equipment can save on upgrade and facilities costs which can be paid for through spectrum auctions.  40 MHz of spectrum without even using aggressive encoding techniques can yield 120 Mbps of usable payload capacity, and more aggressive encoding along with multi-radio solutions that use MIMO could double or quadruple usable bandwidth.  But even with “only” 120 Mbps of capacity, that would allow for 12 HD 1080P channels each consuming 5 Mbps, and another 60 standard definition 480P channels each consuming 1 Mbps.

By comparison, Hulu.com only allocates 0.5 Mbps for their 480P video streams.  If these bitrates are good enough for the Internet – a service people pay for – it is more than generous for free over-the-air broadcasts.  If we matched Hulu in quality, we could broadcast 120 standard definition channels using 60 Mbsps of capacity in addition to the 12 HD 1080P channels or we could do the following breakdown:

• 12 1080P channels – 5 Mbps
• 20 720P channels – 1.5 Mbps
• 60 480P channels – 0.5 Mbps

The programs that garner the highest ratings and most viewers should be allocated the most broadcasting capacity because it fairly allocates capacity in response to demand.  This is more than enough to divvy up between the TV broadcasters and the broadcasters will remain “broadcasters” so that they can sustain their “must carry” status which is their financial bread and butter.

All of this of course requires another transition similar to the DTV transition we just completed in 2009, but it must be done if we want the smooth deployment of the national broadband plan.  If we need to issue another 3 million coupons to low-income households to ease the transition, it’s a very small cost compared to the revenues raised by the spectrum auctions.  The solutions proposed here aren’t entirely painless, but we must avoid half-way solutions that don’t free up much spectrum and require more costly transitions soon after.  Real solutions call for some real changes.