Overview
Goal-oriented communication lets sensors transmit only the data relevant to the receiver, omitting redundant information — but this efficiency creates a new security risk: an eavesdropper can infer information about the system just from the timing and length of packets, even when their content is encrypted.
This project designs a learning-based eavesdropping attack against an effective-communication system and evaluates both the attacker's performance and possible countermeasures.
Prerequisites
Neural Networks and Reinforcement Learning