Monday, June 15, 2009

The Radio Rain Gauge

It’s hard to manage what you don’t measure, and regulators have precious little data on radio operation. Information gathering by citizen enthusiasts could make a huge difference. Riffing on amateur weather stations, it’s easy to envisage a network of “radio rain gauges”.

The need

Information on radio operation is an important part of managing wireless systems. Individual licensees can do this reasonably well, but national regulators are flying blind. The case for better monitoring is clear. The 2002 FCC Spectrum Policy Task Force Report called for better information to more accurately characterize radio operations; Ofcom reviewed automatic monitoring systems in 2006 with a view to increasing concurrent operation where it is lacking, and policing unlawful operation. [1] There have been a number of influential measurement campaigns of radio operation with a view to measuring “spectrum utilization” (I use quotes because neither term is well defined) in recent years, notably Shared Spectrum and CRFS. [2] The case for measurement has received a fillip in the US recently via a bill in the Senate, the “Radio Spectrum Inventory Act” which requires the NTIA and FCC to “inventory” of each radio spectrum band they manage, from 300 Megahertz to 3.5 Gigahertz, every two years.

Precedents and models

Centralized measurement is limited by scope, budgets and politics. The obvious complement is a citizen network dedicated to continuous measurement, and there is a strong precedent: amateur weather stations. For example, someone in my neighborhood has put up a very impressive weather site. The Northwest Weather Network is an Internet based group of private weather stations in Washington, Oregon, Idaho, and Montana that aggregates weather information on its web site. The Citizen Weather Observer Program (CWOP) is a private-public partnership with over 8,000 members world-wide that collects weather information contributed by citizens, and makes it for weather services and homeland security.

There are also public/private collaborations in weather monitoring that could have analogs in the radio monitoring space. AWS Convergence Technologies, Inc. claims to have deployed 8,000 WeatherBug Tracking Stations and more than 1,000 cameras primarily based at neighborhood schools and public safety facilities across the U.S. It says WeatherBug started in the education market by pioneering a program which installed professional-grade weather stations at schools then networked them together; since 2002, WeatherBug’s application has come pre-installed on HP and Compaq computers, and Logitech peripherals.

Weather enthusiasts know about building and managing monitoring stations, and sharing data. Most of them probably don’t know a lot about radio – but the hams do. The American Amateur Radio League, for example, is a nonprofit with 156,000 members that promotes interest in amateur radio communications and experimentation. .

Building out the radio rain gauge network could provide new inspiration and motivation for both weather enthusiasts and radio amateurs.

This will only work if the equipment is cheap. Fortunately, the relentless improvement in computer technology means that mainstream PCs will soon be able to do sophisticated radio monitoring. The free GNU radio software toolkit [3] has been available since 1998, and one can buy the add-on hardware required (one still needs a radio tuner) from Ettus starting at $700 for a basic kit. Most hobbyist software radio projects focus on building receivers rather than transmitters, and a radio rain gauge is an RF spectrum analyzer by another name – which is a straightforward “Hello World” application that’s included in the most software-defined radio packages.

The Radio Rain Gauge Network

This leads to radio band observation running as a background task on thousands if not millions of personal computers. Screensavers have been using spare compute cycles for years to tackle tough academic and public-interest problems; for example, http://distributed.net/ started in April 1997, and SETI@Home was released to the public in May 1999. The infrastructure for collecting data from thousands of PCs is therefore already in place. [4]

A data aggregator will be an important component of such a network. It will also help to have standard formats and repositories so that anyone can get access to the data. This might be done by an academic institutions (cf. BOINC’s role in the cycle-scavenging screen saver endeavor), or it could be end-user driven like the Citizen Weather Observer Program.

Caveats

It’s easy to imagine unease in some quarters about such initiatives. National security operations would prefer that their waveforms and operations not be open to scrutiny and analysis (cf. sensitive areas blurred out in Google Maps), and radio regulators (prompted by nervous incumbent operators) may try to limit the operations of software-defined radios. However, once the knee-jerk negativity has died down, the value of the data collected and the energy of enthusiasts will carry the day. Further, receive-only installations (like radio rain gauges, i.e. spectrum analyzers) are less scary because they do not transmit a signal that could interfere with other operations.

It is also true that the interpretation of the data gathered by radio rain gauges is tricky. Two frequently cited metrics are efficiency and occupancy. The FCC’s Spectrum Efficiency Working Group concluded in 2002 that “it is not possible, nor appropriate, to select a single, objective metric that could be used to compare efficiencies across different radio services” because “the difficulty in calculating some of these variables (for example, the capacity and number of users), and the assumptions behind these calculations, make measures of spectrum efficiency highly unreliable.” John MacDonald, reporting on a 2007 survey of spectrum utilization in Chicago, concluded that “there is a need for a better metric of spectrum utilization than spectrum occupancy which can lead to [erroneous] conclusions as [to] the availability of free spectrum.” Finally, a measure of the intensity of radio operation is not the same as its productivity. One desirable metric would be the aggregate value of information transmitted; however, it is impossible to compute it since there are so many incommensurable measures of value. In the end, though, the current lack of real-time, widely gathered data is so profound that any new information will improve decision making.

Finally, radio frequency engineering is a black art. Building and calibrating the “RF front end” that connects the digital world of the computer to the analog world of radio signals requires expertise that lies outside the software realm, and high sensitivity receivers with low noise and an ability to handle both strong and weak signals will be expensive. A class of “pro” equipment that consumer devices cannot match will remain.

References

[1] National regulators on the need for monitoring. In both cases, regulators framed the problem using the spectrum metaphor as measuring “spectrum use”. FCC Spectrum Policy Task Force (2002): Task Force Report; Report of the Spectrum Efficiency Working Group. Ofcom reports on automatic monitoring systems: Phase I (July 2006), Phase II (Dec 2006)

[2] Measurements. Spectrum occupancy measurements from January 2004 until August 2005 done by Shared Spectrum Corporation for the National Science Foundation (NSF) under subcontract to the University of Kansas. Ofcom (2009) “Capture of Spectrum Utilisation Information Using Moving Vehicles,” Report by CRFS, 30th March 2009.

[3] GNU Radio: introductory article by Eric Blossom ; Wired story; Wikipedia; GNU Radio project documentation, more.

[4] Worldwide distributed computing: List of projects; survey article in Science.