Meet Drone Shield, an ambitious idea for a $70 drone detection system

[Eugen Leitl]

http://arstechnica.com/gadgets/2013/05/meet-drone-shield-an-ambitious-idea-for-a-70-drone-detection-system/

Meet Drone Shield, an ambitious idea for a $70 drone detection system

Aerospace engineer wants to scan for audio signatures of flying robots.

by Cyrus Farivar - May 2 2013, 1:30am WEDT

HARDWARE
 
Drone Shield

On Tuesday, an ambitious aerospace engineer from Washington, DC began seeking
donations on Indiegogo to create an bopen-source drone detection system.b

The Drone Shield would combine a Raspberry Pi, a signal processor, a
microphone, and analysis software to scan for specific audio signatures and
compare them against what known drones sound like (because obviously a
Predator drone is going to sound very different from a small quadcopter.)
Once a match is found, the Drone Shield then sends an e-mail or SMS to its
owner. As of this writing, the campaign is only closing in on one-tenth of
its goal with $301 out $3,500 raised.

John Franklin, the projectbs organizer, believes it would cost around $60 to
$70 to make one, but hebs hoping to raise funds from other privacy-minded
citizens like himself. He notes the idea here is to counter the rising use of
drones not only in foreign theaters of war, but also in domestic skies.

"I'm a problem solver and I'm trying to gauge if this is a problem that
people are interested in solving,b he told Ars. bThe idea here is that it
becomes an open-source thing and people could contribute their own
signatures.b

Franklin estimated it probably would take babout $100 and two monthsb to
figure out if the idea would work. There are other anti-drone tactics and
devices out there, but none quite as cheap as this onebassuming it functions
as advertised. Not all experts are convinced.

"It would be theoretically feasible," said Chris Kyriakakis, a professor of
electrical engineering at the University of Southern California with
expertise in acoustic signal processing. "A lot of problems to solve,
however, to make that happen. [It's] not clear if a single mic would
sufficebmost likely [you] would need a mic array. Noise mitigation would be
another huge problem. Yes, there are dozens of feature extraction approaches
that would work theoretically, but none that I have seen be effective in the
presence of additive or convolutive noise."

Franklin acknowledged that the device wonbt be 100 percent perfect. He also
told Ars that the system would be based on existing publicly available
documents, such as this 1997-era research paper from the Army Research
Laboratory entitled: "Acoustic Feature Extraction for a Neural Network
Classifier" (PDF). But again, his goal is an open-sourced device that an
audience can build upon and improve. If the interest is there, hopefully the
implementation will be too.

bThis project is yet another indicator of the fact that there is very strong
and widespread sentiment on the ground about dronesband itbs one of
tremendous skepticism,b wrote Linda Lye, an attorney with the American Civil
Liberties Union of Northern California, in an e-mail to Ars.