A New Scalable Mutual Authentication in Fog-Edge Drone Swarm Environment

Abstract

A drone swarm is a preferable way in deploying drones for large-scale missions. In establishing drone swarms, secure communication between a fog drone and several edge drones is essential to mutually authenticate each other. Although many research works proposed the designs of mutual authentications, none of them meets the adequate security for the drone swarm environments, requiring either involvement of the ground station during the authentication process or expensive PKI-based crypto operations. Only a few works proposed more lightweight authentication, however, they are still vulnerable to key compromises in addition to the limited flexibility and scalability. In this work, we propose an efficient and scalable authentication protocol for fog-edge drone swarm environments, enabling mutual authentication between fog and edges without involving the ground station. We also show the design can leverage various hardware-assisted security functions. Protocol evaluations show security requirements satisfaction while achieving 14–20 times less computation overhead as compared to PKI-based models.

Response Generation in PP.3

We briefly show how the $F_{i,j}$ in PP.3 can work in such varied environments. Note that the transactions between hardware-protected security environments (HPSE) [17] and outside are protected as the assumption in the preparation phase (PP). H(X) denotes a cryptographic hash function with an input X.

1.1 Response Generation in Generic Edge Drone

As depicted in Fig. 8(a), the edge drone generates a random unique value $seed_{i,j}$ using the random number generator and securely stores it in the secure storage.

Once $C_i$ is received, $ed_{i,j}$ performs following.

$$ F_{i,j}(C_i): H(C_i || seed_{i,j}) \rightarrow R_{i,j}.$$

Response $R_{i,j}$ is stored within the HPSE and not disclosed to any other entity but only to the valid ground station.

1.2 Response Generation in Edge Drone with PUF

We focus on PUFs in two categories; weak and strong PUFs. That strength depends on the number of challenge-response pairs (CRPs) that can be generated from a single device. For more detail, please refer to [13].

Since weak PUF generates the constant output all the time, as depicted in Fig. 8(b), $F_{i,j}$ generates $R_{i,j}$ as following:

$$F_{i,j}(C_i): H(C_i || PUF(output)) \rightarrow R_{i,j} $$

In contrast, as the strong PUF could generate CRPs without limitation, as depicted in Fig. 8(c), $F_{i,j}$ generates $R_{i,j}$ as following:

$$F_{i,j}(C_i): PUF_{i,j}(C_i) \rightarrow R_{i,j}$$