Energy Depletion Attacks on Battery-powered IoT Devices

Unmanned aerial vehicles (UAVs), commonly known as drones, are rapidly gaining popularity for applications in delivery services, emergency rescues, and military operations. Drones are often powered by energy-constrained batteries, making them vulnerable to energy depletion attacks. These attacks aim to exhaust UAV batteries by maliciously targeting existing wireless communication channels or directly contacting UAV physical components. Meanwhile, Physical Sensor Attacks (PSAs) and False Actuation Injection (FAI) attacks are two major physical-layer attacks against UAVs that manipulate the victim’s environment by remotely injecting or jamming signals to form an implicit control channel on the victim’s sensors or actuators without establishing any physical interactions or network connections with the victim UAV. Indirectly, such attacks can cause energy depletion on drones. However, to the best of our knowledge, no previous research has investigated the extent to which PSAs and FAI attacks can be used as energy depletion attacks. This thesis addresses the main research question: Can existing PSAs and FAI attacks be used for conducting the energy depletion attack? To answer the question, we propose a probabilistic attack framework algorithm for exhausting the battery of an UAV by considering PSAs and the FAI attack as attack vectors, which are prioritized based on their direct or indirect impacts on rotating the servo motors of the target UAV. We experimentally evaluate the performance of our proposed attacks through simulations in Ardupilot software-in-the-loop (SITL), by modifying the related simulation parameters of a target drone. We performed two categories of attack experiments, namely 1-min and full-battery, and evaluated the impact of such attacks for 1-minute attack duration and until the battery exhaustion, respectively. Based on our experimental results, the most successful attack type is the FAI attack with its maximum attack intensity due to its direct impact on servo speed and rotation, showing over 16% increase in battery depletion for 1-min attack experiments and over 200 seconds (s) decrease in average exhaustion time for full-time attack experiments compared to the observed results under no attack.