[Updated 2/23/2010 - Based on comments from Peter Stanford, it would seem that some of the information is incorrect.  It turns out that I was wrong about Claudville's White Space implementation and that Claudville is actually using 200 MHz spectrum for last-mile.]

John Greaves wrote this story “The broadband debate: Is White Space the answer?” for the Digital Media Buzz.  Greaves pitted myself and Richard Bennett (ITIF) as the engineers against Rick Rotondo (Chief Marketing Officer of Spectrum Bridge which is the outfit that implemented the White Space backhaul in Claudville Virginia) and Alan Tilles (legal counsel to the Wireless Infrastructure Association) in a debate on the proper usage of White Space sub-700 MHz spectrum.

Richard Bennett and I gave the view that lower frequencies below 3 GHz are better utilized for mobile applications while higher frequencies from 3 to 6 GHz are more suited to backhaul applications.  Rotondo and Tilles argued that white spaces 700 MHz [Stanford clarifies that this is ~200 MHz] is more suitable for backhaul [Stanford says it's not backhaul but actually last-mile] and that Wi-Fi was more suited for the access network (broadband portion).

The debate on backhaul versus access network

“Rick Rotondo, chief marketing officer for Spectrum Bridge disputes this (my suggestion that they should have used 2.4 or 5 GHz backhaul).  “We tried using Wi-Fi at 2.4GHz, 5GHz would never have made it; 2.4 didn’t make it,”

Rotondo’s statement is factually incorrect.  There are plenty of extremely long-range examples of 2.4 GHz and 5 GHz and the world record for 5.7 GHz is 189 miles!  The entire town of Claudville looks to be only a few miles wide on a map.

However, 700 MHz [200 MHz] is [much] easier to refract over the horizon and get around objects [such as hills] which means you don’t need to have as good a line of sight and you can sometimes get away with fewer repeaters.  But while it’s cheaper to build a 700 [200] MHz backhaul than a 2.4 GHz or 5 GHz backhaul, it is an incredible waste of valuable *mobile* spectrum to use it on backhaul.

Let’s say it costs an extra $50K to build the backhaul with 2.4 GHz line-of-sight instead of using 700 MHz backhaul.  But that same 200 MHz spectrum used in high-powered licensed mode can save $500K on the access network (the last mile that connects the Wireless ISP to the home) and operate as a mobile network.  That’s because you only build a backhaul once but you have to build thousands of individual links to the homes.  Even if we have to run a wired backhaul, it is well worth it because we can save a lot more money on the thousands of individual home connections.

But since Spectrum Bridge and Claudville Virginia managed to get free access to 200 MHz from the FCC, it is simply easier to waste that spectrum on backhaul.  Verizon or AT&T on the other hand paid several billion dollars for their 200 MHz spectrum so they can’t waste that spectrum on backhaul.  They would rather spend a little more on backhaul so they can efficiently use the spectrum for mobile and last-mile connections.

Meanwhile, Rotondo and Tilles argued that Wi-Fi would be better for connecting homes and laptops because Wi-Fi is already built into laptops.  Rotondo was even proud to admit that their access network only ran a few hundred feet rather than a mile which is rather odd because it isn’t cost effective or scalable to have to build one Wi-Fi access point every few hundred feet.  If we can use licensed high-power 200 MHz using standard 3G HSPA, 3G EVDO, or better yet LTE, we can cover probably cover 7 miles for mobile in-car or in-home application (mountainous terrain might force us to build more than one tower) and 30 miles for fixed roof-top antenna home installations.  The argument that Wi-Fi is built into laptops is meaningless because 3G and eventually LTE can be built into laptops as well.  In fact, most major laptop makers like Lenovo only charge an extra $30 to embed 3G (HSPA or EVDO) into their laptops.

Debating the best use of white spaces

Alan Tilles also asked some (rhetorical) questions that challenged my assertion that licensed commercial use was the right model.

“Tilles asks, ‘Is there any reason why we cannot exploit white space to the point of having a certain portion of it be licensed, a certain portion of it unlicensed and of the licensed portion have a portion that is set aside for auctioned services and another portion that is set aside for non auctioned services like public safety?’”

Why can’t we have unlicensed white spaces?

Because 54-698 MHz propagation is too powerful for unlicensed use since people will get interference from 50 neighbors instead of only 5.  High-powered unlicensed spectrum usage has never been technically feasible because we don’t have mobile radio technology that can get around the interference issues.  No one in the world has ever successfully implemented a practical unlicensed high-power wireless network which is why Wi-Fi is typically limited to 0.03 – 0.1 watt transmit power.

If someone were to try high-powered unlicensed, it would be extremely inefficient and slow due to massive interference and retransmit problems if it worked at all.  Some would suggest that it is possible get around much of the interference with directional antennas, but that is only applicable to fixed structures and not mobile.

When Google spokesperson Dan Martin called White Spaces the “Wi-Fi on steroids”, he had the right analogy but drew the wrong conclusion that this was somehow a good thing.  Unlicensed Wi-Fi on 54-698 MHz [even with the same power limit] would be more like wireless on “roid rage” where unlicensed radios would smash everything in a 50 meter radius resulting in a tragedy of the commons.

Why not public safety?

At the fall 2008 CTIA conference in San Francisco, dozens of state and local reps at a round table before the FCC testified that their proprietary safety radios failed during the disasters because the agencies couldn’t inter-operate with each other.  The cell phone literally saved the day during Katrina and all the public safety workers resorted to their cell phones.  It was once again extremely effective at saving lives in the recent Haiti’s earthquake.

We can have very redundant and reliable public safety over commercial cellular networks by using quad-band dual-SIM phones that can reach two independent wireless operators.  We can even require the system prioritize data/voice from emergency workers or consumer calls to 911.

Even if the entire public sector adopted a single standard, building a separate network for just a single niche applications such as public safety is very inefficient from a spectrum standpoint.  There simply isn’t going to be much traffic on that network and it will sit idle most of the time.  It is better to pool the spectrum into a generic transport and then use it for as many applications as possible and apply the necessary priortization schemes.  Building special purpose government networks makes as much sense as government building their own power plant and power grid for government use.