1

Secure Broadcast Protocol for Unmanned Aerial Vehicle Swarms

Hongpeng Guo*, Tianyuan Liu*, King-Shan Lui† , Claudiu Danilov‡ , Klara Nahrstedt*

*

†

‡

2

• The communication between leader and followers must be encrypted against eavesdropping.

Background

• UAV swarm applications will be pervasive in the near future, such as surveillance, farming, light shows and military operations.

• Leader-followers mechanism is a widely used mechanism for swarm management.

• Drones may join or leave the swarm during the task window due to battery limitation. The communication keys must be managed and updated properly when swarm member changes.

3

Problem of Existing Solutions

leader

• Leader establishes peer-to-peer unique keys for the followers.

• Leader need to encrypt a single broadcast message multiple times for different followers.

• Large overhead for the leader. Not efficient and scalable.

4

Problem of Existing Solutions

leader

• Leader maintains a shared group key for the whole swarm.

• A broadcast message from leader can be encrypted only once and used by all swarm members

• Leader needs to establish public key encrypted peer-to-peer links with the followers for group key updates, which causes large overhead.

• It is challenging to secure in-swarm communication, especially when the swarm membership is changing rapidly.

​

• Forward and backward secrecy.

​

• Light-weight, efficient and fast.

​

• Tolerant unstable network conditions.

• A key management scheme based on Diffie-Hellman Chain.

​

• Make drone join overhead to be O(1) in terms of the swarm size.

​

• A swarm broadcast protocol (SBP) to tolerate unstable wireless networks.

5

Challenges & Our Solution

6

System Overview

leader

7

Method Overview: Diffie Hellman Chain

Generate a shared secret among a group of identities.

Diffie Hellman Chain

Leader

Follower

8

Leader Key Generation

Blind keys generated by followers

Intermediate keys

9

Follower Key Generation: Overview

Blind keys generated by leader

Intermediate keys

Blind keys generated by followers

10

11

Swarm Membership Changes

12

Analysis & Unreliable Network Handling

• Tradeoff between drones’ join and leave events.

​

• Key information will be retransmitted when wireless network is not stable.

​

• Key version control mechanism to synchronize group key in the swarm.

​

• Compare our swarm broadcast protocol (SBP) with two baselines.
• Public Key Broadcast Protocol (PKB):

shared group key, peer-to-peer based key updates;

• Public Key Unicast Protocol (PKU):

Peer-to-peer encrypted links.

​

13

Evaluations & Experiments

• Benchmark evaluation on network overhead and CPU utilization.

​

• Emulate drones join & leave scenarios with network emulator.

​

• Compare with asymmetric key updates (PKB) on drone joining events

​

​

Leader overhead

​

​

follower overhead

​

​

message overhead

​

​

14

Benchmark Evaluation

• Compare with asymmetric key updates (PKB) on leaving events.

​

​

Leader overhead

​

​

follower overhead

​

​

message overhead

​

​

15

Benchmark Evaluation

• CORE network emulator, 90 seconds, start with 10 drones.
• Leader broadcast 8kb data to the followers every second.

​

16

Scenario Evaluation

• SBP and PKB occupy similar bandwidth when the swarm is stable.

​

• PKB will consume much more bandwidth when drones join or leave the swarm.

​

• The overhead of SBP is almost constant even when the swarm is highly dynamic.

• We designed chain-structured Diffie-Hellman mechanism for group key initialization and management in UAV swarm.

​

• The key management mechanism is light-weight and efficient in dynamic membership scenario, which only introduce O(1) overhead in terms of swarm size.

​

• Based on the key management algorithm, we further designed a swarm broadcast protocol (SBP), which synchronize the group key within the swarm and can tolerate unreliable network conditions.

​

​

17

Conclusion

18

Q & A