Position tracking protects drone herds against hackers
The technique can tell the difference between real robots and frauds.
Much ado has been made over flying drones in groups, but there hasn't been much thought given to protecting autonomous drone groups against security breaches. What happens if someone impersonates enough drones to hijack their collective decision-making process? MIT might have a viable defense. Its researchers have developed a positioning technique that would prevent and mitigate these kind of impostor attacks. The key is to give each drone a wireless fingerprint that reflects its interaction with the environment -- effectively, the network can tell the difference between the actual drone pack and a fraud operating from the outside.
The approach relies on measuring the signals from a two-antenna WiFi system used in mid-flight. As its signals bounce off the environment at different times and directions, they give the drone's transmitter a unique profile that can't easily be replicated. While the system can't completely ignore bogus wireless transmissions, it downplays those sketchy signals that are likely to come from the same source. Ideally, this prevents intruders from spoofing drones while protecting legitimate drones that happen to be transmitting similar signals.
Early tests are promising. In a theoretical study, the technique kept the drones largely in their planned positions even when three quarters of the drones had been compromised by an attack. A clever attacker still wouldn't have much control, in other words. There's a lot of work to do before this is ready for the real world, but MIT envisions it being useful for delivery drones, self-driving cars and any other robotic herd where one rogue agent could cause serious chaos.
