Yesterday I discovered this nice infographic produced by Drone Industry Insights where you can see the MANY threats in the area of drones. I would have never though there could be so many, even though looking at each single one I have to agree that the threat exists.
One of the interesting aspects of the infographic is the layering of threats, from the human operator (the drone user) to the interface with the control (in the smart phone or as a dedicated control device), the drone platform, the gateway then on to the radio connectivity and the cloud finally addressing the applications (in the cloud and on the drone). As often it is noticed, the human link is the weakest one, in most cases. Hence a future where drones can be autonomous may actually reduce the security risk (but might increase our anxiety!).
This also connects to the panel on Industry 4.0, I listened to, at the inauguration of the new EIT Digital Milan Satellite location on May 23rd. In that panel it was pointed out that one of the issues looming on Industry 4.0 is the potential threats and the lack of a systematic approach to security.
Security remains in many areas an after thought. You have something (a product, a service, a process) you find gaps -or you are subject to an attack- and you react to fix it.
For complex systems, like a manufacturing plant involving hundreds of robot, this is going to be a nightmare. Robots are seldom designed with security in mind. And robots will be interconnected with one another, and with the cyberspace. Worse! They will take proactive action to interconnect and are progressively being designed to promote interactions.
Willem Jonker, the EIT Digital CEO, pointed out during the panel that we should move from a static view of security to a dynamic view where patterns are monitored and tiny deviations from the normal pattern give rise to suspicion and lead to in-depth analyses.
In turns this calls in the concept of digital twins, digital copies of the real entity, like a bit copy of a robot, that can be used for evaluating the observed behaviour against the expected one (according to its specs). It is more than that. Once you have digital twins you can also compare one against the others and look for emerging patterns. This is done using a variety of technologies in the artificial intelligence cluster. You apply deep learning to create the understanding of the behaviour of a single digital twin, then you might apply deep neural networks to create and analyse emerging patterns out of several digital twins and so on.
It is not just Industry 4.0 and robots in the shop floor. Our cars are morphing into autonomous robots and will be using interaction as the stepping stone for their behaviour.
Indeed, moving towards autonomous systems and even more towards symbiotic autonomous systems will make the whole area of threats/security more difficult to address and will absolutely require consideration at the design stage. Design for security, not just design with security in mind!
This was also pointed out at the Quadrato della Radio, yesterday, where a strong call was made to invest in education of students and professionals, something, as I pointed out to the many industry key people present, EIT Digital is committed to.
This is something being addressed in the new White Paper under development at the IEEE FDC Symbiotic Autonomous Systems Initiative.