Meta-Learning Based Antenna Positioning and Beamforming in MA-Enabled Secure ISAC Systems

In this paper, we consider a movable antenna (MA)-enabled secure integrated sensing and communication (ISAC) system. An optimization problem is formulated to maximize the minimum secrecy rate among all legitimate users in the presence of an eavesdropper via jointly optimizing the antenna positioning, transmit beamforming, as well as artificial noise. However, the formulated problem is non-convex that is challenging to solve directly due to the tight coupling among optimization variables. To tackle this complicated issue, we propose a meta-learning based optimization framework. By transforming the constraints and constructing penalty terms, the global loss function is used to guide the optimization process within the feasible region. In this framework, the neural network takes gradients as input and outputs update steps for each optimization variable, which provides improved interpretability for the algorithm. Extensive numerical results validate that the algorithm achieves favorable optimization performance in both communication and sensing aspects.