UAVDome - Safe UAV Transportation

With the rise of UAVs/drones in modern transportation and logistics, their security is being challenged and threatened both in frequency and sophistication. This creates the need for better secure drones in the commercial and military sectors. Several solutions are available, mainly relying on encryption or modified antennas, but those are often expensive or require extensive modification to the communication infrastructure, increasing the total cost and not applicable for commercial drones. Moreover, software based solutions are either limited to attack detection and rely on external or limited data sets. Other solutions require a camera which again can be attacked easily and limited during nighttime. This paper presents UAVDome, a machine learning-based solution that detects GPS attacks on drones and provides a unique return-to-home functionality without relying on GPS or camera sensors. The solution implements a virtual GPS that predicts the drone’s relative position using internal sensors only preventing the effect of external attacks without relying on any external data or communication. Experiments present the solution feasibility with minimal loss.