How AI, IoT, and Other Technologies Will Need to Coalesce into Comprehensive Counterdrone Defenses
Drones are the foundation of some of the most devastating weapon systems in many countries’ arsenals. They are also a growing focus of smart cities everywhere. Indeed, drone-related projects are likely to claim a significant portion of the projected $135 billion annually that public agencies worldwide will invest in smart cities by 2021, according to International Data Corporation estimates.
For all of our sakes, let’s hope that counterdrone initiatives benefit from some of that investment. Drone technology can monitor our urban environments, aid in traffic management, and accelerate disaster response. But it can just as easily be leveraged by evil people to poison our air and water, block our roads and tunnels, and detonate explosives at lightning speed and with surgical precision.
For that reason, and because this technology is easily accessible to even the most casual terrorist, counterdrone projects need to become an integral component of smart communities initiatives going forward. However, implementing counterdrone civil-defense infrastructure will be trickier than it sounds. The reasons are several:
- Political: Much the same way that swarming insects can easily overpower a lightly defended human, counterdrone initiatives will have trouble fending off the glut of commercial drones taking flight in many countries. Considering that the president of the United States, the world’s largest drone producer, has every intention of loosening export controls on the technology, expanding drone sales to allied nations, and unapologetically deploying drones in every military operation, the drone glut will get worse before it gets better. And these political forces will almost guarantee that terrorists everywhere get their hands on the most sophisticated new drone technology before effective countermeasures can be put in place.
- Legal and regulatory: The legal rights and responsibilities surrounding drone use, and legal recourse when others abuse the technology, vary widely by jurisdiction. Some counterdrone tactics may be illegal in various jurisdictions. Depending on the jurisdiction, the legal obstacles that prevent cities, neighborhoods, or individual property owners from defending themselves against unwanted drones may be considerable. For example, even if the law allows you to use your shotgun in self-defense, you may not have the legal right to train it on a drone that’s overflying your property, operating in a menacing fashion, and invading your privacy. And there will almost certainly be conflicting regulations at the local, state, and national levels that complicate efforts at building stable counterdrone defenses.
- Public safety: Societies may have legitimate concerns against citizens being able to take vigilante-grade counterdrone tactics into their own hands, especially in densely populated urban areas where other people and property may be in harm’s way and many of the drone overflights might serve civic purposes of compelling importance. Even when there’s a governance infrastructure that keeps vigilante counterdrones at bay and invests this responsibility in public safety authorities, some of these technologies may incorporate classified military technology that is unavailable or simply to beyond the budgetary reach of most state and local governments.
- Technological: Counterdrone technologies are very much a work in progress, through there’s a lot of entrepreneurial energy working on the problem from many directions. Drone detectors run the gamut from radar, radio wave receivers, audio sensors, and of course smart cameras galore. In addition to remote hacking of drone software, counterdrone defense technologies being commercialized include lasers, sonic blasters, radio-frequency jammers, radars, electromagnetic pulse generators, directed-energy microwave systems, infrared cameras, acoustic sensors, and ballistically targeted drone-snagging nets.
In counterdrone defenses, artificial intelligence (AI), smart sensors, wireless technologies, and the Internet of Things (IoT) are playing key roles. Startups are bringing many promising AI-driven counterdrone technical solutions to market incorporating the following capabilities:
- Automated detection of drones, pinpointing of locations, and predict likely flight paths of drones
- Automated classification of approaching drones by model, operator, and threat profile, taking care to minimize false positive and false negatives;
- Automated triggering of security alerts and activation of destruction, disabling, neutralization, distraction, deflection, commandeering, and interdiction tactics, including taking over of drones by sending “go home” commands, and landing of seized drones in safe locations to minimize collateral damage.
What’s missing from today’s counterdrone landscape is any systemic focus on knitting together these approaches into a unified civil-defense infrastructure. Promising as many of these approaches may be in isolation, none of them has been proved out in comprehensive systems that have been deployed in production civic infrastructure. Here’s what’s needed, in terms of federal regulation and infrastructure, for the maturation of counterdrone technologies into robust civil-defense infrastructure:
- Mandatory strong identity for licensed drones: Drones are licensed, so they and their owner/operators should have standard identities that are used to track, control, and regulate them continuously. In the US, the FAA is developing a rule for remote identification and tracking of drones. Once this is finalized, the agency plans to require that most drones in US airspace implement it in a transponder-like capability that permits accurate, continuous, and real-time identification and location. This capability will be absolutely essential for counterdrone platforms to be able to authenticate all remote-ID-compliant drones in their vicinity and facilitate automated tracking, permissioning, control, and interdiction of drones in their jurisdictions by the relevant authorities.
- Built-in safeguards to protect authorized drones against cyberhacks: Let’s remember that the vast majority of drones in the skies will be legal, licensed, authorized, and generally beneficial to society. We can’t fully protect society against malicious drones unless we also secure benign drones against cyberhacks. Just as the autonomous vehicle industry is investing deeply into defending against wireless hijacking, drones manufacturers will need to strengthen the authentication, access control, encryption, tamperproofing, and other security features of their products around common standards. There should also be a federally managed registry of hack-vulnerable drones that may be banned from operating or sharply circumscribed in where, when, and for what uses they may be deployed.
- Embedded AI for ensuring that licensed drones always comply with all relevant mandates: This may sound like a no-brainer, but there’s no such regulatory requirement anywhere for drones. Instead, many drones are essentially dumb and their behavior is partially or entirely under the manual control of human operators (aka “pilots”). In other words, society is relying on the good behavior of humans who are licensed by the FAA to operate drones, who are trusted not to use the technology for evil purposes, and who are expected to stay informed and compliant with the many, overlapping, changing, and confusing requirements that they must meet to operate their drones in various circumstances. In compliance situations, society must use AI to strictly circumscribe what actions the human pilots may take.
- Nationwide counterdrone coordination center: Effective counterdrone defenses must seamlessly span the entire country and be closely coordinated with neighboring nations. The US has granted limited counterdrone authority to the Departments of Defense and Energy to defend appropriate federal assets and facilities within these agencies’ jurisdictions. However, a truly comprehensive counterdrone capability would naturally fall under FAA, or perhaps Homeland Security, in a federated governance structure with state and local authorities. Without an operational counterdrone coordination center—and perhaps a nationwide drone air-traffic control system with an AI-driven monitoring backbone, there’s every possibility that terrorists and criminals will exploit the vulnerabilities in current defenses by deploying consumer-grade drones for nefarious purposes. At the very least, such coordination center could ensure that federal, state, and local law enforcement and security officials are trained, equipped, and certified to counter hostile drones.
- Ongoing counterdrone gaming simulation exercises: When evil drones enter the picture, they are just as likely to do so in coordinated groups or swarms as in individual one-off attacks. Defending against such scenarios may require that the counterdrone defenses be agile and intelligent enough to predict against attacks that play out over extended geographic and temporal operating theaters and impact diverse public and private sector assets and parties. It may become necessary to train the AI behind these
counterdrone defenses within complex “multiplayer gaming” simulation environments. Much of this work will draw on the same frontier research in game theory, decision theory, and other computational topics that military has long relied on for their war-gaming exercises. For an interesting research program in multiplayer gaming that has potential counterdrone applicability, check out OpenAI Five. This involves games in which people and bots engage in complex, interactive, real-time, and team-oriented interactions. It’s clearly applicable to drone/counterdrone scenarios, which involve real-time continuous interactions with a mix of strategic and inconsequential moves over long time-horizons, with high-dimensional action and observation spaces, and partially observed states.
What’s clear from all this is that counterdrone defenses must become an integral component of every community’s public safety infrastructure. This is not a responsibility that should be left to vigilantes who get their kicks blasting pesky quadcopters to smithereens.
About the author: James Kobielus is SiliconANGLE Wikibon‘s lead analyst for Data Science, Deep Learning, and Application Development. He was previously a data science evangelist for IBM.