Automata learning is a technique for automatically generating finite state machines of black box systems. Especially when learning communication protocols, automata learning has proven to be a useful technique to gain insights into the protocol implementation, which can also reveal bugs in the implementation investigated. In previous work, we have shown that this approach can also be used to learn implementations on real devices such as Bluetooth devices.
For this master thesis, our goal is to learn implementations of the LoRa (Long Range) communication protocol, which is often used for energy-efficient communication of sensors over long distances placed in the environment, e.g. on floating buoys in the water. This master's thesis topic provides a nice intersection between the theoretical insights of grammar inference and work on IoT hardware.
Automata learning for Bluetooth devices: https://learntwins.ist.tugraz.at/wp-content/uploads/sites/24/2022/01/Fingerprinting_Bluetooth_Low_Energy_Devices_via_Active_Automata_Learning.pdf