In this work, we introduce a framework for enhancing secrecy in unmanned aerial vehicle (UAV) aided communication under dynamic attack. We formulate the problem as a secrecy rate maximization then decouple it into secure beamforming and secure UAV trajectory subproblems. We exploit the concept of differential privacy (DP) to design a variational autoencoder (VAE) to perform the secure beamforming. The UAV trajectory subproblem is transformed into a convex quadratically constrained program (QCQP) which is efficiently solved via CVX. The proposed framework shows efficiency in improving the secrecy rate.
Enhancing Secrecy with Differential-Based Framework in UAV Communications Under Dynamic Eavesdroppers
2024-04-19
1371534 byte
Conference paper
Electronic Resource
English