WEBVTT

00:00.000 --> 00:06.140
The following is a somewhat technical presentation covering the recent evolution of counter-drone

00:06.140 --> 00:09.400
and counter-counter-drone technology.

00:09.820 --> 00:15.880
This talk will focus on the emergence of low-cost, weaponized, unmanned aerial vehicles that

00:15.880 --> 00:20.120
have directly, literally, impacted this region.

00:20.640 --> 00:27.260
The starting point is, as always, the economics driving the rapid advancement and proliferation

00:27.740 --> 00:34.100
of militarized drone systems, both drones and anti, before moving on to an overview

00:34.100 --> 00:39.700
of recent technological developments, mostly from the Russo-Ukrainian War, and how those

00:39.700 --> 00:44.700
are proliferating globally as they are proven and productized on the battlefield.

00:45.280 --> 00:52.540
In state-level conflicts, at least among cost-constrained states, larger Class II and III drones have

00:52.540 --> 00:58.340
become dominant players, delivering the vast majority of munitions to targets, an extraordinary

00:58.340 --> 01:01.280
development in just three years of conflict.

01:01.920 --> 01:07.840
The first broadly successful state-level low-cost drone system was the Iranian Shahid series,

01:07.840 --> 01:13.520
particularly the 136, introduced in 2021, a mere five years ago.

01:13.540 --> 01:19.560
It, and systems like it, offer a nearly two orders of magnitude cost reduction, compared

01:19.560 --> 01:25.620
to similar per-mission costs for legacy delivery platforms, and provide an order of magnitude

01:25.620 --> 01:31.860
more cost-effective method of delivering payload than traditional theater ballistic missiles.

01:31.980 --> 01:39.300
While it is far easier to intercept a low, slow drone than a TBM, the 100-to-1 cost advantage

01:39.300 --> 01:44.400
will almost always win out in economic or strategic analysis.

01:44.440 --> 01:49.120
For non-state actors, the transformation has been even more remarkable.

01:49.120 --> 01:54.460
Ultra-low-cost consumer or consumer-derived drones, which are functionally differentiated

01:54.460 --> 02:00.800
from designed as arms drones by, among other things, shorter range, typically only 5 to

02:00.800 --> 02:05.840
10 kilometers, but still, they were early game-changers for ISIS in Iraq, starting around

02:05.840 --> 02:07.200
2016.

02:07.480 --> 02:12.720
They enabled attacks well behind lines, at trivial cost, and vastly reduced risk compared

02:12.720 --> 02:17.940
to that old insurgent standby, bongo trucks, with Katayusha conversions.

02:18.300 --> 02:23.740
As RF links have become more exploitable, the cost of attacking protected targets for

02:23.740 --> 02:25.720
insurgents has increased.

02:25.720 --> 02:30.260
But almost no critical infrastructure targets are protected, yet.

02:31.000 --> 02:37.460
More advanced drones, such as those using long-range radios, LTE, GPS, or vision-based

02:37.460 --> 02:42.760
targeting, increasingly provide insurgents with the ability to strike transnational targets

02:42.760 --> 02:50.040
accurately and with nearly zero retribution risk, and at trivial cost, which vastly expands

02:50.040 --> 02:52.740
their rational targeting options.

02:52.740 --> 02:58.440
Most civil defense systems are built around a platform consisting of radar, camera, and

02:58.440 --> 03:03.900
a jammer, and possibly a software-defined radio for RF detection, usually mounted on

03:03.900 --> 03:07.380
a single pole: stuff on a stick.

03:08.180 --> 03:11.140
Radar remains the core detection technology.

03:11.140 --> 03:15.440
It is proven, practical, reliable, and mostly weather-immune.

03:15.560 --> 03:20.460
The maximum range of the smaller radar systems typically used for counter-UAS, is usually

03:20.460 --> 03:25.160
limited to about 5 kilometers or so, with some models claiming up to twice that.

03:25.160 --> 03:33.300
The standard target is always a DJI Phantom 4, a phantom for an RCS of 0.01, the equivalent

03:33.300 --> 03:38.020
of a flat metal target of about 10 by 10 centimeters square.

03:38.020 --> 03:43.540
Proliferation of stealth coatings will increasingly compromise radar range.

03:43.800 --> 03:45.640
The radar equation is brutal.

03:46.040 --> 03:51.960
R squared times R squared is R to the fourth, so it takes 16 times better signal to noise

03:51.960 --> 03:57.320
to double the range of a radar system, and any reduction in return signal is multiplied

03:57.320 --> 03:58.560
by four.

03:58.900 --> 04:03.420
Therefore, stealth tech is very effective at hiding drones from radar.

04:04.140 --> 04:09.920
Electro-optics provide visual verification and confirmation of threat and type.

04:09.980 --> 04:14.340
There has been less rapid development in camera technology, and they are pretty standard.

04:14.340 --> 04:19.440
A typical package includes a high-zoom visible-light camera and normally a second camera operating

04:19.440 --> 04:24.560
in a thermal range, which is particularly useful at night and through moderate obscurants

04:24.560 --> 04:26.720
like smoke or dust.

04:27.560 --> 04:33.660
Jammers may be of the low-cost barrage style, which tend to interfere with blue force comms,

04:33.660 --> 04:38.780
or reactive type, which are more expensive but do generally preserve blue team comms.

04:38.800 --> 04:43.120
Omnidirectional units are good for area coverage, while directional tracking units have much

04:43.120 --> 04:46.980
greater jamming range, but must be mounted on a PT platform.

04:48.120 --> 04:54.000
The electronic warfare platform in a modern battlespace is an RF disco party, so most

04:54.000 --> 04:59.160
attack drones now escape EW with countermeasures like controlled radiation pattern GPS antennas

04:59.160 --> 05:01.600
or by fiber-link control.

05:01.600 --> 05:06.820
Even so, jammers remain essential elements of any defensive strategy, if at all possible

05:06.820 --> 05:08.100
to deploy.

05:08.300 --> 05:12.960
Radio direction finding equipment is becoming a must-have add-on.

05:12.980 --> 05:17.860
Short-range systems, like the Kraken SDR, are cost-effective upgrades.

05:17.860 --> 05:24.400
Very long-range, highly capable, time-difference of arrival systems are far from low-cost,

05:24.400 --> 05:30.340
and even they can't track RF silent drones running on GPS or those targeting by AI vision

05:30.340 --> 05:31.360
mapping.

05:31.360 --> 05:37.260
Still, they can do some serious, extremely long-range tracking magic when they do work.

05:37.440 --> 05:43.380
AI-based analytics developments are improving false positive and negative rates by cross-correlating

05:43.380 --> 05:47.920
detection modalities, and can perform some very impressive tricks pulling signals out

05:47.920 --> 05:48.940
of noise.

05:49.260 --> 05:54.960
The core analytics stuff is a SMOP, a simple matter of programming, and so it is advancing

05:54.960 --> 05:56.460
fairly quickly.

05:57.300 --> 06:03.340
There is an ongoing proliferation of counter-drone technology to address the threat of armed drones.

06:03.640 --> 06:09.620
Each category has specific strengths and weaknesses that make them suitable, or not, for particular

06:09.620 --> 06:10.920
applications.

06:10.920 --> 06:16.640
Jammers are a core system, and will typically be proposed as part of any C-UAS strategy.

06:16.800 --> 06:22.260
They can block FPV drone command signals and operate our video feeds out to several kilometers,

06:22.260 --> 06:26.560
and are quite effective at jamming GPS signals, which are deep in the noise floor anyway

06:26.560 --> 06:28.620
by the time they get to Earth.

06:29.040 --> 06:35.000
Jammers are so effective and ubiquitous that CRPA, fiber control, and even Starlink are

06:35.000 --> 06:36.980
becoming default options.

06:36.980 --> 06:42.500
Interceptor drones are emerging as highly capable and justifiable defense strategy.

06:42.580 --> 06:44.280
They do have real limitations.

06:44.540 --> 06:46.480
Detection ranges tend to be short.

06:46.480 --> 06:52.880
An incoming suicide drone, flying at 300 kilometers an hour, and detected 5 kilometers out, has

06:52.880 --> 06:58.260
a one-minute warning, giving a very short detect, verify, and launch window.

06:58.440 --> 07:04.040
If the defender isn't launched from between the incoming drone and its target, intercept

07:04.040 --> 07:05.980
probability isn't great.

07:05.980 --> 07:11.920
Even the fastest interceptors, which are now exceeding 350 kilometers an hour, can't touch

07:11.920 --> 07:17.240
emerging jet-powered drones that cruise at 500 kilometers an hour.

07:17.600 --> 07:24.940
Kinetic kill models, with non-explosive smack-em style, are quite a reasonable balance of capability,

07:24.940 --> 07:30.420
cost, and most importantly, collateral risk for critical infrastructure use.

07:31.040 --> 07:34.380
Emerging, Ukrainian-made models are quite affordable.

07:34.380 --> 07:39.540
It is just amazing every time a firearm proves successful at defeating an incoming drone.

07:39.700 --> 07:45.120
People do get them with their AKs, and good on them, but it isn't a viable defensive strategy.

07:45.160 --> 07:49.960
AI-powered remote weapons stations are an interesting development, but extremely risky

07:49.960 --> 07:52.040
for most civilian stations.

07:52.580 --> 07:57.100
Tracking and targeting advancements are very much worth watching, even if still priced

07:57.100 --> 08:00.760
and regulated out of a lot of civilian defense applications.

08:01.340 --> 08:07.900
One of the few, accurate, relatively symmetrically priced rockets is the APKWS II, but it is

08:07.900 --> 08:11.820
laser-guided, meaning the target has to be locked and visible.

08:12.460 --> 08:17.560
Rocket-repelled, guided systems have not yet been treated to the same cost curve as drones,

08:17.560 --> 08:19.140
but watch this space.

08:19.280 --> 08:25.100
Microwaves hold some promise, but they are expensive, short-range, risky, and can be

08:25.100 --> 08:30.480
cheaply shielded against, with mu-metal shields, like were commonly used in old CRT-era

08:30.480 --> 08:32.320
desktop speakers.

08:32.320 --> 08:37.080
However, the cost-benefit and mass-targeting capability is tempting.

08:37.080 --> 08:41.500
Destructive lasers are promising, but remain extremely expensive.

08:41.500 --> 08:47.200
They are also intrinsically capable of rather horrific collateral damage, and multi-kilowatt

08:47.200 --> 08:51.020
battlefield laser systems are a proliferation risk.

08:51.020 --> 08:56.060
For now, they remain expensive and not well-suited to civilian applications.

08:56.060 --> 09:02.320
Blinding, or dazzling lasers, are a whole different category of cost, risk, and capability.

09:02.920 --> 09:07.840
They can't drop a drone out of the air, but it is possible to blind the imaging sensors,

09:07.840 --> 09:13.840
such that AI-controlled or FPV-operator-controlled drones can't see their targets.

09:14.140 --> 09:19.180
As an add-on defensive measure, they have a good cost-benefit ratio and are light enough

09:19.180 --> 09:22.080
to ride on most camera PT platforms.

09:22.600 --> 09:28.460
A few areas of particularly rapid progress have emerged, mostly out of the Russo-Ukrainian

09:28.460 --> 09:34.040
War, and are rewriting the effectiveness of any C-UAS strategy.

09:34.060 --> 09:41.700
One of the first to emerge, as the EW space became chaotic, was the CRPA GPS antenna,

09:41.700 --> 09:48.020
which used clever maths to zero out EW sources and amplify GPS SAT signals.

09:48.020 --> 09:52.940
They are effective, but remain expensive, $5,000 to $20,000 each.

09:53.140 --> 09:59.720
They depend on beefy processors, like a TI DSP or Xilinx FPGA, to do signal processing

09:59.720 --> 10:01.540
magic in real time.

10:01.800 --> 10:07.520
That such processors are reaching certain markets is an open proliferation issue.

10:07.760 --> 10:13.740
Fiber links run on naked half-millimeter diameter glass, now shrinking to a quarter for better

10:13.740 --> 10:15.700
range and more payload.

10:15.700 --> 10:23.080
Early models, way back in November 2024, had 10-kilometer spools and could unreal at 50-kilometers

10:23.080 --> 10:24.220
an hour.

10:24.720 --> 10:31.020
By mid-2025, the state-of-the-art was 20-kilometer range and 70-kilometers an hour.

10:31.020 --> 10:37.060
And now, in 2026, they run 60-kilometers at 120-kilometers an hour.

10:37.060 --> 10:42.380
This capability has proven frontline-critical and is evolving insanely quickly.

10:42.380 --> 10:47.820
At the same time, EW proliferation has pushed transceivers to just about every frequency

10:47.820 --> 10:52.660
that can carry useful data, from a few megahertz to over 12 gigahertz.

10:52.660 --> 10:59.220
Just this month, a Russian drone was found carrying a 6.1 to 7.2 gigahertz up-converter,

10:59.220 --> 11:04.740
which puts its COM links just outside most compact drone detector ranges, as those rely

11:04.740 --> 11:09.400
on SDRs, like to hack RF, with 6 gigahertz maximum.

11:09.400 --> 11:15.120
However, starting at about 5 to 6 gigahertz, atmospheric losses really start to impact

11:15.120 --> 11:18.520
usable range, so this won't go on forever.

11:19.180 --> 11:26.000
Starlink has become increasingly operationalized and is effective because EW can't reach LEO.

11:26.400 --> 11:33.420
An antenna looking up means not seeing ground-based EW emitters, nor can ground-based RF detectors

11:33.420 --> 11:34.900
find them.

11:34.900 --> 11:39.280
Because Starlink minis were expensive and hard for Russia to source, they are mostly

11:39.280 --> 11:43.000
used on ISR drones that would usually come back.

11:43.200 --> 11:47.720
But in the last few months, they've been found on one-way drones, like Shaheds.

11:47.720 --> 11:52.940
These should not be available at such volumes to antagonists as the satellites know the

11:52.940 --> 11:57.840
real-time position of each receiver, that a sanctioned country is successfully flying

11:57.840 --> 12:03.900
drones from sanctioned territory into the heart of an ally has implications.

12:04.900 --> 12:10.740
AI vision systems started with a lot of promise, even delaying fiber-optic development in Ukraine

12:10.740 --> 12:14.400
as resources were put into refining AI models.

12:14.520 --> 12:20.460
After some initial success with OpenCV, they were, however, slow to operationalize and

12:20.460 --> 12:24.360
are only now becoming reliable at consumer-ish costs.

12:24.660 --> 12:30.820
Running on NVIDIA Jetsons, mostly, and with rapidly improving CNN-based vision detection

12:30.820 --> 12:36.120
and tracking software, like BoT-SORT, which is an open-source tracking library that's

12:36.120 --> 12:41.440
designed to recognize and track targets from moving platforms, and they can run very well

12:41.440 --> 12:48.840
on something like a Sakura II M.2 gumstick with 16GB of RAM and 30 teraflops that cost

12:48.840 --> 12:52.680
just $400 and weighs only 10 grams.

12:52.860 --> 12:57.520
Stealth is an interesting development, with initial reports out of Russia, then China,

12:57.520 --> 13:03.160
of low-cost, iron-based coatings that are durable enough for one-way use and may reduce

13:03.160 --> 13:07.380
the RCS of a drone by 10 or even 100 times.

13:07.560 --> 13:12.840
India and Iran have both recently made announcements, while China has recently announced a new

13:12.840 --> 13:18.560
graphene-based coating that is definitely way more expensive, but still likely a massive

13:18.560 --> 13:20.880
bargain compared to U.S. systems.

13:21.080 --> 13:25.660
These will become more and more ubiquitous on a mid- to high-end one-way platforms.

13:25.660 --> 13:30.600
So, what does reasonable civilian defense architecture look like?

13:30.720 --> 13:36.260
Layered defense is critical to managing evolving threats efficiently and effectively.

13:36.260 --> 13:40.180
Wherever warranted, a radar optical jammer system,

13:40.180 --> 13:42.720
stuff on a stick, is the core defense.

13:42.740 --> 13:48.700
Such a platform is anchored with the radar system, typically a $400,000 to few million

13:48.700 --> 13:52.940
dollar investment per position, depending on the range of the equipment.

13:52.940 --> 13:58.220
Where that cost isn't justified, distributed systems can help but not completely replace

13:58.220 --> 13:59.540
the capability.

14:00.100 --> 14:05.400
While these basic systems remain fundamental to the defense core, that central point, as

14:05.400 --> 14:10.820
a line-of-sight viewshed, there will be significantly compromised by rough terrain or buildings.

14:10.980 --> 14:13.840
Neither radar nor cameras can see through hills.

14:14.120 --> 14:19.760
The command system should be able to usefully integrate additional signal sources and control

14:19.760 --> 14:22.160
additional mitigation effects.

14:22.160 --> 14:25.780
Some sense modalities are intrinsically short-range.

14:25.780 --> 14:30.580
Acoustic signals are not extractable by any technical means beyond about one or two kilometers

14:31.180 --> 14:35.480
due to basic physical limits like Shannon's law and thermal noise.

14:35.820 --> 14:42.920
Common ISR-RF signatures, similarly, disappear into a fog of legitimate radio traffic after

14:42.920 --> 14:46.420
a few hundred meters to at most a few kilometers.

14:46.420 --> 14:53.640
Therefore, short-range field-deployed sensor node meshes are, I believe, going to be increasingly

14:53.640 --> 14:58.480
critical for reliable detection of low-signature incoming threats.

14:58.740 --> 15:05.000
The Zvook project in Ukraine is a good example, and has won deserved recognition with tens

15:05.000 --> 15:11.020
of thousands of distributed acoustic sensors listening for drone engine sounds and reporting

15:11.020 --> 15:15.140
detections back to core defenses and to likely targets.

15:15.140 --> 15:19.920
So what do you do when you detect an incoming, possibly armed drone, other than shut down

15:19.920 --> 15:25.200
critical systems and warn personnel over the PA to shelter in place immediately?

15:25.400 --> 15:31.600
Where jammers fail, intercept drones, explosive or simple impact, are a good balance between

15:31.600 --> 15:38.180
the various critical metrics of risk, cost, range, and probability of intercept, despite

15:38.180 --> 15:41.740
fundamental limitations and operational risks.

15:41.740 --> 15:45.720
The Sting-2 from Ukraine is a good reference example.

15:45.860 --> 15:51.360
Optical jammers, lasers that dazzle, blind, or even damage the optics of vision tracking

15:51.360 --> 15:57.580
and FPV drones are far less expensive and far more exportable than big honkin' lasers,

15:57.580 --> 16:03.580
those with enough power to cause thermal mechanical failure or material degradation, and are comparatively

16:03.580 --> 16:09.560
cheap and exportable and light enough to ride on most robust camera system PTZ platforms

16:09.560 --> 16:11.340
for targeting.

16:11.340 --> 16:14.640
Fiber controlled drones are already in theater.

16:14.880 --> 16:19.620
Hot nichrome wires and mechanical capture devices can break fiber links as the fiber

16:19.620 --> 16:23.260
settles on the mechanism after the drone flies over.

16:23.260 --> 16:28.040
They should be considered as basic as mechanical fencing for access control and deployed on

16:28.040 --> 16:34.140
a similar plan to cut fiber targeting links with as much standoff as geography allows.

16:35.120 --> 16:40.800
A hybrid layered defense has the best chance of success now and into the future as drones

16:40.800 --> 16:42.460
continue to evolve.

16:42.860 --> 16:47.860
The critical takeaway from this data dense slide is that drones started as a game changing

16:47.860 --> 16:52.920
revolutionary technology, but as they've become crucial to modern warfare, that rate

16:52.920 --> 16:58.960
of change has accelerated exponentially across the whole range of drone, counter-drone, and

16:58.960 --> 17:05.060
counter-counter-drone technologies, with extraordinary innovation rates coming out of Russia, China,

17:05.060 --> 17:06.640
and especially Ukraine.

17:07.290 --> 17:10.680
These technologies are proliferating out of theater.

17:10.700 --> 17:16.520
Already four-channel CRPA and ultra-long-range fiber has been found in use in the CENTCOM

17:16.520 --> 17:20.940
theater, both in the Middle East and in various conflicts in Africa.

17:21.200 --> 17:25.980
The rate of spread is also accelerating, as just about everyone realizes how effective

17:25.980 --> 17:29.160
and rebalancing this technology is.

17:29.400 --> 17:34.120
Keeping critical infrastructure and personnel safe from attacks where antagonists may be

17:34.120 --> 17:39.620
operating from thousands of kilometers away, and even a single disgruntled individual can

17:39.620 --> 17:45.020
avail themselves of lethal, destructive, long-range capabilities, is the sort of challenge that

17:45.020 --> 17:47.220
should keep us all on our toes.

17:47.440 --> 17:52.260
This is already a completely different game, and it is just barely started.

18:20.660 --> 18:22.060
You can reach David Gessel at dg@brt.llc