Slashdot Mirror
Why Detecting Drones Is a Tough Gig
An anonymous reader writes with a link to some interesting commentary at Help Net Security from Drone Lab CEO Zain Naboulsi about a security issue of a (so far) unusual kind: detecting drones whose masters are bent on malice. That's relevant after the recent drone flight close enough to the White House to spook the Secret Service, and that wasn't the first -- even if no malice was involved. Drones at their most dangerous in that context are small, quiet, and flying through busy, populated spaces, which makes even detecting them tough, never mind defeating them. From the article, which briefly describes pros and cons of various detection methods: Audio detection does NOT work in urban environments - period. Most microphones only listen well at 25 to 50 feet so, because of the ambient noise in the area, any audio detection method would be rendered useless at 1600 Pennsylvania Avenue. It is also too simple for an operator to change the sound signature of a drone by buying different propellers or making other modifications. It doesn't take much to defeat the many weaknesses of audio detection.
Are they warmer than a bird? Maybe heat detection?
Laws are rules for the court, but merely a bottom bar to hit for life. Think beyond laws in your actions always.
Use a plain old radio direction finder?
“He’s not deformed, he’s just drunk!”
essentially, as of Right Now, drones will avoid and evade at will (of controllers).
my first thought was - eh? drones will spout RF like nuts to work, you could find a drone for sure.
that's what the article says too.
I assume detecting the RF signature of the transmitter controlling the drone is the best way.
Of course there are these problems:
1. There are many signals on the bands used for RC.
2. It is possible to build an autonomous drone.
3. In these days of software defined radio, people can spin up non-off-the-shelf, non-standard radio control systems.
Omne ignotum pro magnifico.
No detection necessary.
It *might* be possible to use an overlapping LIDAR system to pickup on small flying objects. The number of sensors required and the systems integration required would be an enormous task though.
Architectural plans are like computer source code with a couple of differences: You only compile once.
Are you really afraid of a 2 pound multirotor? Do you really think it will ever be possible to achieve that much destruction with one? I really hate seeing fear mongering, be realistic and see there is very little danger from the current hobby sized systems available.
These days you'd set a waypoint, send your drone off, and drive away. There's no RF to speak of, unless you're live-streaming it over LTE.
The typical multirotor "drone" is necessarily built very lightweight; the electronics and motors are not typically shielded much at all. The brushless motors emit stupid amounts of RF energy due to unshielded motors, multiple banks of ESC's covered by nothing but heatshrink, etc. It shouldn't be particularly hard to spot a fast moving, localized source of RF noise at frequencies typical of multirotor motors.
Then there is heat. The ESCs and motors are HOT. Again, mass must be minimized so there is no easy way to hide that heat. A bright infrared spot zipping along at 50' altitude shouldn't be hard to detect.
For things that are piloted with FPV cameras there is a big video return signal coming off the craft. Due to antenna size this is usually some UHF frequency and fairly loud. The range of possible frequencies is vast, but in the real world there are a limited number of cost effective miniature transmitters available, so it isn't difficult to anticipate the likely frequencies.
Seems like there are a number of tell-tails that shouldn't be hard to exploit if you are serious about it have the means...
Maw! Fire up the karma burner!
It's about being able to deflect them and prevent them from doing bad things.
Sure, it's easier to deal with something you can detect in the first place, but if you can effectively block them by putting up barriers, physical and electronic you will have succeeded in your primary goal. So here's my approach..
1. Do your best to detect them, use sound, video and detect the RF signals emitted by the device and the pilot.
2. Erect physical barriers that are not visible to the operator or the device. I'm thinking there is a LOT you can do with simple fishing line in this regard, but I'm sure a lot of tall trees would serve an excellent purpose too. Put up an obstacle course.
3. Put up electronic fences using short range GPS and WiFi jamming around the "protected" area. You can effectively reduce the ability of a drone to find it's way around and make it impossible for it to be remotely controlled.
4. Concentrate your efforts on finding the PILOT. They will likely have an RF transmitter in their hands, so it shouldn't be that hard, unless the drone is self guided (which is why you jam GPS and provide physical barriers).
5. And Finally, if you do detect something flying where you don't want, come up with some non-lethal ways of bringing it down. You don't need to fire anti-aircraft guns at it, there are ways I can think of which wouldn't present much risk to people, but would be effective in bringing down your average hobbyist's drone.
So I say again, detection is but a small piece of the total security puzzle here, and trying to use audio detection is about the LAST way I'd try it...
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
The usual suspects in the MLC and the congress critters who live off their campaign contributions will love this as it is easy to sell to a fearful public.
A "simple" problem that sounds easy to do for the technology challenged which turns into a money sink because in practice is not easy at all.
I can just see the sweet heart deals, cooked QA test and moving goal posts with massive cost overruns.
Star wars any one?
Easy detection method #48:
(1) Send out a large electromagnetic pulse
(2) If it falls out of the sky, it was a drone
Maybe this is gonna force the creation of practical laser guns.
Until there are proper detection mechanisms, guards can just shoot them down manually.
Drones are going to be revolutionary in a way for our society. Soon, people will start thinking of their "air space" and shooting down drones that invade it. And then, there are going to be products to do that.
Why not use cameras instead? I'm not talking about motion detecting ones, which are not going to very reliable, but what about color detection? Most drones stand out distinctly from the sky they're flying in, and you can see glints of light and such from them. You could also simply have human surveillance watch out for them - they tend to be pretty good at telling birds from drones. Machines aren't replacements for humans at everything, you know...
"Set a man a fire, he'll be warm for the rest of the night. Set a man afire, he'll be warm for the rest of his life."
Yeah the master of the drone might get the sick idea to put a weapon on-board and use it to assassinate someone. What? Ohh.
What about a combination of image detection and echo-location? Image recognition and/or regular sound detection would identify candidate objects, and narrowly-focused echo location would then scrutinize the candidates further.
And so what if you take out a few birds accidentally? Just place a KFC nearby to handle such. They'll enjoy the free raw materials.
Table-ized A.I.
Personally I don't think concentrating on the detection problem is the best approach. We can jam, and we can put up barriers and reduce a lot of the risks with very low cost. We already have RF direction finding capability which could be deployed to pinpoint not only the drone but the pilot's location, it's a little costly, but it's out there and would be nearly off the shelf. Trying to build a RADAR or IR sensor to hone in on the drone is a nice idea, but high cost, low reward.
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
Typically? Somebody with enough resources could add shielding with high-end light materials. Catching nuts with deep pockets or lots of time may be a much bigger problem than catching off-the-shelf attempts.
Table-ized A.I.
Just think of a drone as a big mosquito.
Have gnu, will travel.
http://xkcd.com/1523/
The more I think about this, the more I like it.
#include "standard_disclaimer.h"
IANAE (Biologist not an engineer) but since drones are light, the motors can't be well shielded. Appears to be a common problem, at least with readily available commercial stuff.
http://www.rtl-sdr.com/rtl-sdr-running-adsb-on-a-quadrocopter/
So, what you're saying is that if someone builds a drone with RF shields on their motors, with hyper-quiet rotors, GPS only autonomous nav, and perhaps an invisibility cloak system they too can evade the drone-police? Thank you, good sers!
This is the NSA, we're gonna geet U h@x0r5! Also, what is a h@x0r5?
I don't think concentrating on the detection problem is the best approach
Yeah, well word #2 in this story's subject is "Detecting," so I went with that. Silly me.
not only the drone but the pilot's location
Detecting the "pilot" is actually the hardest part. $200 buys a programmable autopilot that will drop a UAV on any GPS coordinate the batteries can reach — sans pilot. Signals can come from any radio system, including ubiquitous ones like cell towers, so good luck finding that needle in the urban haystack.
Maw! Fire up the karma burner!
http://slashdot.org/submission...
http://slashdot.org/submission...
http://slashdot.org/submission...
Because Slashdot is pulling the wool over your eyes.
Still waiting on Serviscope_minor to wake up to fucking reality and realize that Jessica Price isn't going to fuck him.
Somebody with enough resources
...and right there you've just narrowed the pool of potential threats. Defense isn't about guarantees. Raising the bar is all you can ever achieve.
Give them "enough resources" and they'll nuke you from orbit.
Maw! Fire up the karma burner!
The self piloted drone doesn't get to it's destination w/o a GPS fix. Short range GPS jammers are off the shelf, *easy* use, and not expensive.
Remember, I'm saying that detection is down on my list of things to develop, that other things have a better cost/reward and are based on existing technology.
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
Except a lot of drones can now autopilot their way to GPS way points. Set it and forget it, no RF control needed.
When it comes to drones we have a hell of a lot more to fear from the ones coming out of Pennsylvania ave than the ones flying over it.
"I opened my eyes, and everything went dark again"
Good luck jamming inertia
Maw! Fire up the karma burner!
Drone nets. Nuff said.
Is this really a persistent problem? Are people flying paparazzi drones 24/7 just outside the White House fence?
Simple social solution: Just make it a "throw your ass in Gitmo" offense to operate *ANY* drone beyond the White House fence. I shouldn't matter if it was a malfunction or a gust of wind. Your drone = your responsibility. Someone bumps you and your hand slips on the control? Then it's still your fault for flying too close to restricted airspace.
p.s. As for technical solution, why not fight drones with drones? Set up a bot net of White House drones that monitor the exterior and construct a real-time 3D view of the entire lawn from dozens of overlapping 360 degree 3D cameras. It wouldn't be too difficult to detect any motion that's not part of the swarm.
Good luck jamming inertia
Which is why you put up GPS jamming and physical barriers too. Inertial nav is only accurate over short distances, unless you have some external way to calibrate your nav system and can remove the various bias issues caused by vibration, temperature changes and other things that cause changes in the gyros (mechanical, laser ring or otherwise). Usually inertial nav's need to be calibrated, and they do that with GPS (or some other system like LORAN) in order to maintain enough accuracy over time.
Nothing is perfect, but you do the best you can with the resources you have and you live with the risks you cannot afford to fix. I'm suggesting that there is bigger bang for the buck in other things than trying to go out and detect these things in flight.
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
Did you miss the part where I said we'd jam GPS in the general area of the lawn? Really, you just provide false signals and you can pretty much deflect the drone on GPS autopilot away from it's intended target. Easy to do, off the shelf hardware exists.
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
(1) Send out a large electromagnetic pulse
Tomorrows headline: "Secret Service drone jammer interferes with pace makers - Half of congress deceased" :)
What possible downsides could an EMP have...
Audio detection isn't nearly as broken as the article pretends. Sure if all you have is a single mic, then you have no hope. OTOH, with multiple mics, you can *localize* sounds, which means you don't need to recognize the sounds of a drone, just realize that there's some noise coming from something in the air where there shouldn't be anything. With a microphone array, you can actually pinpoint sound sources much weaker than ambient noise. It's certainly not trivial, but within the realm of what's realistic (assuming there aren't simpler solutions).
Opus: the Swiss army knife of audio codec
We need a bladerunner drone that "retires" other drones.
I just don't buy that audio can't be used. With an array of high quality microphones spread over an area fed into a software radio and some pretty hefty computing power you should be able to look for the rhythmic audio that your typical copter type drones will generate. Because even if they change the size and shape of the rotors all that does is change the frequency not the amplitudes you will see from the rhythmic action.
With some proper math and the right computing power you should be able to identify drones pretty rapidly with fourier analysis and the same microphone array could use Doppler effects to calculate position, and the directional vector. Combine this with some systems to double check such as heat, RF and conventional high band doppler radar and you should have a system that will get 99% of the drones and even provide targeting to a shoot down type system.
The key here is you need some pretty dam good microphones spread over a pretty large area feeding into a pretty massive computer array. It wouldn't be cheap or easy. The easy thing is use a doppler radar system that cross-checks it against RF emissions to eliminate birds. But IMO the best system would use all three, high band doppler radar, RF emissions and audio (and maybe even heat). With three cross checks you should be able to get pretty good accuracy.
First you would need very expensive and very powerful ones to actually defend against a drone. The cheap short range ones you talk about only work at short range like their name suggests. You would need very expensive jammers to be effective against drones. The power needed to jam every direction in the sky increases exponentially by distance. You also need a jammer that can jam the Russian GLONASS GPS system which operates at a different frequency range and most GPS chips have supported both GPS systems for years. Soon the Galileo system will also be up using even another frequency.
Jamming would never really be a viable solution in a city except during an actual attack. It would cause more damage then it could ever hope to prevent. It would have to block the radios and GPS used by planes, helicopters, police, firefighters, and anything else remotely near the jammer, to actually be effective at blocking a radio controlled drone.
Most drones transmit 1.5mw (which is a decent amount) on 910mhz, 2.4ghz, 5.8ghz.
With This and a few hours of work in python. Problem solved.
http://singularityhub.com/2010/10/13/new-super-microphone-can-hear-you-in-a-crowded-stadium-video/
Again, you folks act like RF energy cannot be controlled, it just goes off in all directions no matter what you do and that there is no way to isolate the jamming energy to small areas. This is not true. Would there be some residual affect outside the intended area? Perhaps, but if you do this right it would not be wide spread, nor would it need to extend more than a few hundred feet beyond the desired areas....
Tell me it's not possible to limit where you put the RF and control the signal strength outside the desired area to acceptable levels... I think you can do that, if you are careful and think about where you put the jammers, what antennas you use and what direction you point them.
Also, GPS jammers are neither expensive nor rare. They are off the shelf and have been for decades and well with the budget of the Secret Service (In fact I'd bet they ALREADY have a few). Directional antennas are also inexpensive and off the shelf, even at the frequencies of GPS. None of this is rocket science... Just a bit of engineering.
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
when did r/c airplanes get renamed to drones?
I see a point where nets will go up for a while(probably after an attack). But they will be so ugly that there will be relentless pressure to take them down. Then they will have "pop-up" nets. But they won't work and the nets will go up one more time.
Then some genius will come up with a solution and the nets will go away.
But we are all taking about little quadcopters and whatnot. But there are many many types of contraptions that will come along with drone technology. Gliders, missiles, planes, darts, parachutes, combos; so fly then crawl.
Just like the stupid war on terror. The real key is not to fight the war but to prevent the causes of the war in the first place. There will always be a few nutcases so there isn't much that can be done there. But if there were to ever be regular attacks then you are doing something wrong at a much higher level.
You're not looking for the target as a big thing. Birds are trivially distinguishable: look at the Doppler signature: flapping wings look very, very different from rotating blades.
You would need very expensive jammers to be effective against drones. The power needed to jam every direction in the sky increases exponentially by distance.
Jamming would never really be a viable solution in a city except during an actual attack. It would cause more damage then it could ever hope to prevent. It would have to block the radios and GPS used by planes, helicopters, police, firefighters, and anything else remotely near the jammer, to actually be effective at blocking a radio controlled drone.
Nonsense. Get directional GPS antennas and point them straight up into the sky and surround the whitehouse. While you're at it point some EMP guns straight up in the air too. Nothing is suppose to be flying over the whitehouse so you don't have to worry about disrupting civilian planes, etc... Making them powerful enough to go a few thousand feet straight up but not powerful enough to interfere with LEO satelites would be simple. If you used a directional antenna (again pointed straight up and surrounding the whitehouse) to block all signals then the only drones you would have to worry about would be ones that navigated via visual cues which would be considerably harder to do.
This information does not show up much in the media, but there is massive use of small hobby grade drones in Donbass hostilities.
Most of the casualties inflicted on both sides are from 60's era dumb artillery. The modern element is drones, used for both intelligence gathering and (mostly) correcting the artillery fire. It makes old guns much more deadly. Drones are typically supplied and sometimes operated by volunteers. The Russian side also uses newer military grade drones, but they are expensive and there is not a lot of them.
Typical countermeasure is wiping the whole WiFi band with a powerful noise transmitter. Many drones (especially bigger ones) are brought down with barrage of bullets from AK-74s when they are detected visually.
I was going to make an observation about how we immediately launch into solutions. We come running with potential fixes (LOL SO EASY) then flame each other about how they won't work. Standard internet frenzy (accept it, casuals and SJWs) aside, we're really being quite constructive. It's the sort of problem that, even uninvited, we leap at the chance to chew on. It's behavior we've seen at /. before.
/was/, but then I realized, we have an obligatory for that too.
I
Clearly LIDAR needs to be considered. If a guy can accurately detect overhead wires using a "friggin' lazer" then I'm sure a "drone" can easily be detected...
Any UFO (Unauthorized Flying Object), malicious or not, over my lawn or my house gets shot down.
Yes, this is what I suspect the result will be. That birds will be killed indiscrimately whenever and wherever the military / security situation deteriorates. Even in normal times I can see this happening over key locations, over tense borders, et.c.. Birds that keep to small territories will be less affected, those that migrate or have large ranges might be in increased danger of extinction.
Do nothing.
The drone will hit the White House, and then fall to the ground. Broken.
These things are made in China FFS.
Prop driven planes used to play havoc with the reception of tv broadcast signals. I suspect there's enough ambient RF around that you might not even have to add any signal sources of your own. Ducted fans might make it a little trickier but you still have that air flow that will have characteristics of its own. I'd try IR doppler but I don't know enough about how good it is at detecting mini vortexes.
A $1500 drone can lift 30 pounds of C4, fly it 3 miles from a boat crusing the Potomac, then drop it from 500 feet with deadly GPS accuracy.
Why even bother with drones ?
Just hop down to your local Police Station or Sheriff's department, steal the MRAP they bought to intimidate the peasants . . . . er. . . conduct their police business in a safe manner without fear of landmines or IED's, load it with explosives and drive that bastard right onto the front lawn ?
I mean, if we're going to go all Michael Bay on things, why do it with tiny ass drones ? Retrofit that MRAP so you can control IT with a transmitter, pack that baby full of ANFO and REALLY do some damage.
Better yet, in true MB fashion, send in dozens of drones first, cue some epic music THEN send in the MRAP.
If Slashdot editors did the minimal due diligence, they'd find that the writer is advocating for the technology he's selling, i.e. this article is in reality an advertisement.
Phrases like: "The most effective way to detect drones is with radio frequency (RF) methodology, and Drone Detector is the only system utilizing that technology. " should be a dead giveaway....
So, if I want to hit the white house with a few kg of dynamite:
1. No radio. No remote control. Use gps only. Start the drone, release it, and no more radio emissions. No control signal to the drone, I switch my transmitter on only if I want to abort the mission late in the game. No video stream or telemetry from that drone either. Although I may have another drone for video purposes. One that doesn't go too close to the action.
2. As for shielding rc noise, aluminium foil is very light weight. Cover most of the electronics & motors in foil, except the gps & radio receivers. Use coax wiring instead of ordinary for powering stuff. A handful of small capacitors also cuts down on RF. Maybe not perfect, but this drone will have very little RF compared to ordinary ones. Peaceful hobbyists don't bother with even the simplest, easiest shielding.
3. As for heath, a little lump of dry ice attached to each motor & each ESC. Won't last long, but such a mission won't last long anyway.
4. As for people with shotguns, make it a fast drone. Base it some design that won a recent model aircraft speed competition. Also use height, perhaps coming in in a 30-45 degree dive with somewhat erratic movement. High speed, not much time to react, hard to hit something coming from above.
5. As for gps jamming, I only expect that very close to the high-profile target. So the drone is prepared to use inertial navigation for the final descent. Ideally, some image recognition too if I want to hit a particular window.
Ruling out rc and heath detection is not that hard. The drone will still be easy enough to detect with radar, but current radar equipment is tuned to ignore items much smaller than a missile/fighter plane. That is easy enough to change, but hasn't happened yet. So have your fun . . .
This might actually work. Most drone GPS modules are sensitive to noise from the drones own RF noise, so sending out ~1575mhz noise would probably work. It would have to be for some distance outside of the whitehouse as well, as a safe measure. Drones do have failsafes that could be programmed to continue its course should it lose GPS, thus the reason for the extra distance.
this should be rather simple. better or more serious "drones" have emissions- either telemetry or signals for FPV or video recording or sensor output.
If you have a secured airspace like the White House where there should be NO RF transmissions, then if you find one a) in that airspace and b) moving then you've detected a drone. Simple triangulation tells you where it is.
It's not as if most "drones" (getting tired of every hobby grade aircraft called a drone but that's another discussion) are manufactured with any creativity in mind. Almost all of them use the exact same RF spectrum for flight control and communication, and the same for video or sensor return to the pilot. Almost no one "develops" anything in the multicopter space anymore- they simply BUY whatever commercially available (and FCC approved!) bird is out there, or assemble based on step by step instructions developed thru years of design and trial-and-error by dedicated hobbyists.
Hell, most "drone incidents" aren't even using autonomous navigation, which if properly set up would make a device hard to detect.
If anyone were serious, they'd ditch multicopters and go for faster, completely autonomous fixed wing aircraft - get in, perform mission, and if the mission is to get photos or sensor input, get out. Autopilot systems for reasonable prices for any aircraft, using navigation techniques like GPS and positioning based on wifi, are well outside the "build it yourself" hobbyist range now.
But popular hype, and politicians (and businessmen) with no imagination, are focused on multicopters, so that's all we'll hear for another few years.
I'd be inclined to modify that statement, and say. "Audio detection does NOT work, now".
Each rotor of a quadcopter is going to emit sound that depends on the number of prop blades and the prop speed. The four rotors will emit at frequencies that are almost but not quite the same, The four frequencies continuously shift by minute amounts as the control system adjusts power to stay stable in the air.
The quadcopter therefore has a very distinctive sound signature. This signature is out there, waiting to be detected, if the money can be found to develop the technology to do it.
Presumably if that happened, there would be a push for stealth quadcopters. But that's another kettle of fish.
The GPS on mine usually has 12+ satellites locked in, so its unlikely to be directed via false GPS. It'll just kick on the error routine and ignore the GPS and go into compass/accelerometer mode.
GPS Jamming doesn't work as you seem to think. Your receiver would still see quite a few signals, they'd just be lying about the position. Or, if you go with the cheaper model jammer, the receiver wouldn't see anything because the noise floor would be too high.
Switching to alternate navigation modes is indeed possible, but as I've stated in other posts, I'm suggesting that there be physical obstacles added using largely invisible structures to force the drone to fly a complex route to get where it wants, then concentrating the detection capacity along the forced route. In other words, force the bad guys to do specific things that make them more easily detected.
No system is perfect...You do what you can within the resource constraints you have and consign yourself to living with the risks that are left.
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
I know a bit about radio frequency (RF), doppler radar, jamming GPS (passive and active). Near-range drone detection, tracking and characterization (within inches) is simple using doppler radar.
The real issue here is counter measures. When do you shoot (CRAM), capture, use active GPS to shift off course, etc.? Drones (electric or combustion) do emit a unique RF and acoustic signature that could be used for close range real-time counter measure decisions.
It would have to be for some distance outside of the whitehouse as well, as a safe measure. Drones do have failsafes that could be programmed to continue its course should it lose GPS, thus the reason for the extra distance.
Yeah, my guess is they won't do much until after the first successful attack.
Creating a half mile deadzone though wouldn't be a too big of deal. You might
see a comeback of payphones, emergency call boxes, paper maps and the like
in that area to compensate but it really wouldn't be that big of a deal.
The bigger deal is what to do with them everywhere else. How do you protect
schools, hospitals, playgrounds, city halls, restaurants, etc... Granted a backpack
bomb is still fairly easy but it gets alot easier when you can be miles away.