August 7th, 2007 by Gordon Cook
We had some informative discussion on spectrum policy on my list in July.
David Reed explained to Roland Cole why the regulatorium’s spectrum auctions cause the blood pressure of the radio engineers to rise.
Here is my translation of what David said: Basically it is because spectrum regulation was built on a 1920s knowledge of spectrum behavior. What we have put in place is something similar to economics as run by the Chicago school of macro economics.
The innovations of complexity economic theory are there petitioning the powers that be to recognize them. The powers that be are the telcos and MSOs whose business models depend on exclusive use monopoly spectrum allocation. They will never be recognized.
Thanks to the continued progress of Moores law enabling digital signal processing to do ever more amazing things it is now possible for software defined radio to operate on radically different principals from the analog power broadcasting of the 1920.
It is now possible to build software defined radios that can listen for other spectrum users in the neighborhood and fine tune their behavior sufficiently enough to stay out of each other’s way and hence not interfere.
The problem was that, even if the Republican FCC were not controlled by the duopolists, whom it pretends to regulate, we would still be shackled by the analogue based presumption of spectrum scarcity. In a mindless effort to raise revenues for the federal government these private equity based non public good companies have paid for exclusive right to use spectrum where it was formerly thought that only a single person could ever talk in the space in a “single time slot” or frequency.
The FCC holds license for the use of those “frequencies.” Savor the absurdity by imagining that
in the fiber world, one telco been able to buy exclusive use of all parts of the spectrum that might be defined as red while another telco bought exclusive use of violet, and an MSO exclusive use of blue. Gosh, the Chicago school missed a wonderful opportunity to help us dig our economic grave even deeper.
David put it this way: â€œthe spectrum isn’t even full, as Shared Spectrum demonstrated by measuring all of the spectrum used in NYC during the RNC, which meant that in addition to regular stuff, there were vast numbers of military and first responders active as well. They found that if you used very sensitive measurement tools, only 13% of the entire RF spectrum had *any* detectable signal energy during the full time period, and the actual occupancy was infinitesimal.
If we imagine using just the technology we know and can make cheaply today, we can make any one band much more efficient in terms of the number of bits/second available for use without confusion. But we have no bands to deploy that new technology into - because every band is allocated 3 x over (primary secondary and tertiary users). “Paradoxically, except for cellular data, pretty much unused, until someone proposes to use it and then the current licensees say: wait, we have made crucial plans to use it someday OR that use might make some grandmother who bought something a long time ago have some speckles on their flickering Dumont TV, so pay us lots of money.”
“Bob Frankston cites digital techniques - in particular what we know about channel coding - as a major direction for improvement. This is a mature subject, but only digital cellular systems and some military systems actually use it well. And beyond that are adaptive techniques - which adapt the signals to the physical environment.”
All of the above are techniques - ones that are not employed, and in many cases, not legal for employment because they postdate the 1934 regulatory structure and are too hard for the lawyers to incorporate into a fundamentally decade-time-constant static frequency allocation scheme. In other words, the rate of change *built in* to the radio systems in the field by the *laws* is that radio system innovations take 10-20 years to introduce, though they now take 12 months to invent, and soon will take days to invent (with Software Defined Radio).”
David then showed that with out too much effort you could separate out an interfering wave form in your radio design and come up with a software defined radio that would run in the 2.4 g hertz band and not be subject to interference by operation of your microwave oven.
He concludes in a way that reminds me of the assumptions that the macro economics used to make with their models. “One of my students recently showed (in a class project to be a published paper) that you can even do the soft-decoding trick with retransmitted packets - or with packets that are detected at multiple points in space - so that one can decode and separate the packets with *far* higher reliability than one could otherwise.”
COOK’s Edge: Software defined radio that is beginning to emerge from the labs into actual tests has the ability to render all spectrum management moot. Small wonder that the legal mandarins there have begun to sneer that open source SDR cannot be trusted.