Implementation of an Artificial Immune System to Mitigate Cybersecurity Threats in Unmanned Aerial Systems

摘要

This project aims to use an artificial immune system to detect cyber-attacks on the dynamics of a minidrone. Using linear transfer functions, a zero-dynamics attack was designed to be injected into the system. To further explore how zero-dynamics translates from linear to non-linear systems, a similar attack was made using a random value rather than an unstable zero of the system. These attacks were then injected via the ground station computer. An offline artificial immune system was created and used to test the effectiveness of detecting the attacks. The immune system uses a discrete, binary, non-self recognition algorithm to create detectors and find disturbances in the data.