Supervisors

Suitable for

Abstract

Existing work has shown physical-layer fingerprinting to be an effective technique for authenticating satellite communication, particularly in the absence of cryptographic authentication. In fingerprinting, physical-layer characteristics of the signal are observed in order to identify the transmitter and distinguish legitimate transmitters from attacker-controlled radios.

In this project a student will extend an existing satellite fingerprinting system, which has been primarily evaluated using the Iridium satellite constellation. The student will extend the system to work with additional satellite systems, assessing the transferability of the techniques. This will require the capture of a new dataset of satellite communication from a different constellation, which may include building a decoder for the messages, depending on availability of open-source decoders. An extension of the work might involve the development of new techniques to work better across multiple satellite systems.

Prerequisites: This project will require experience with machine learning (specifically TensorFlow), and an understanding of signal modulation and coding (the project will likely require building a message decoder). Experience with software-defined radios is preferable but not required.

Useful URLs: See also https://arxiv.org/pdf/2305.06947.pdf (current work) https://github.com/ssloxford/SatIQ