Drone Simulator with Attack Enforcers

For the development of drone simulation with attacker modelling using runtime enforcers.

Setup

Requirements:

• Python (it is expected that you have an understanding of python too)
• ArduPilot
• Dronekit

Note: this was all done on a Windows 10 PC

MAVLink and AdruPilot

ArduPilot is the most advanced, full-featured, and reliable open source autopilot software available. (Quote from the ArduPilot project repo)

We will be using ArduPilot Copter. Wiki link.

Follow these instructions to setup ArduPilot..

I installed on Windows 10, it was a hassle, I do not recommend. The provided powershell script was what got it running in the end. Running random scripts people tell you to is always bad. Read it first, it isn't long.

Then I needed to:

• Clone ArduPilot repo
  • git clone https://github.com/ardupilot/ardupilot
• (Must be Cygwin from here) Update and clone dependencies
  • git submodule update --init --recursive
• Navigate to ardupilot/ArduCopter directory.
• Run ../Tools/autotest/sim_vehicle.py --map --console
  • At first this complained about some missing python dependencies. Install these pip install XYZ
  • Eventually it started this huge build which took quite a while
  • Once complete, the quad-copter simulator opened up

Dronekit

Dronekit-Python helps you create powerful apps for UAVs. (Quote from the ArduPilot project repo)

• Can be installed with pip install dronekit
  • I needed to install wheel first pip install wheel (there was an ungraceful fail without it)

In our context it is used to communicate with MAVLink and ArduCopter and abstracts a level of detail away for us.

Examples

Simple take off and land (tests your simulator!)

1. Open Cygwin terminal. Navigate to ardupilot/ArduCopter directory.
2. Run ../Tools/autotest/sim_vehicle.py --map --console
3. Open CMD. Navigate to the directory of this repo.
4. Run python dronekit/missions/simple-test.py
5. Watch as dronekit controls the simulated drone through a simple take off and land!

Take off and land with base station

Very similar behaviour to the above take off and land but with much more control via simulation of asynch comms.

1. Open Cygwin terminal. Navigate to ardupilot/ArduCopter directory.
2. Run ../Tools/autotest/sim_vehicle.py --map --console
3. Open CMD. Navigate to the directory of this repo.3. Run python test-base-TOAL.py
   • Base station (test-base-TOAL.py) communicates with on drone controller (missions/take_off_and_land.py) which interacts via dronekit with the drone.

Enforcer Attacks

Placed between the environment (drone) and controller (base station) the attack enforcers can execute a number of attacks.

Policies in this work are defined as Timed Automata and synthesised into enforcers using easy-rte. Below is an example of an injection attack which inserts the ABORT signal after a specified time (T subscript A).

To demonstrate attack enforcers and the attack manager see the below examples.

Examples: Drone Attack Enforcers

Commands assume you have the simulator running

• Mission: take off and land - python test-base-TOAL.py
• Mission: fly from A to B - python test-base-A2B.py
• Mission: fly from A to B to A -python test-base-A2B2A.py

Enforcer Interchange

Adds an attack manager to perform enforcer interchange, which dynamically enables and disables (through the use of activate signals) appropriate enforcers to perform more complex attacks.

Example: Drone Attack Manager

Uses enforcer interchange to dynamically enable appropriate attack enforcers across a mission that combines the three examples above.

• Combined Missions: all of the above - python test-base-combined.py - Assumes you have the simulator running

The policy below illustrates the attack manager policy for the three basic drone attacks used earlier.