Abstract
In this paper, a deceptive signal-assisted private split learning is investigated. In our model, several edge devices jointly perform collaborative training, and some eavesdroppers aim to collect the model and data information from devices. To prevent the eavesdropper from collecting model and data information, a subset of devices can transmit deceptive signals. Therefore, it is necessary to determine the subset of devices used for deceptive signal transmission, the subset of model training devices, and the models assigned to each model training device. This problem is formulated as an optimization problem whose goal is to minimize the eavesdropped information while meeting the model training energy consumption and delay constraints. To solve this problem, we propose an actor-critic deep reinforcement learning (AC-DRL) framework that enables a centralized agent to dynamically determine the model training devices, the deceptive signal transmission devices, the transmit power, and sub-models assigned to each model training device while considering network topology and device properties (e.g., energy and privacy constraints). The proposed AC-DRL framework integrates an actor network to generate optimal actions and a critic network to stabilize the learning process by evaluating the cumulative rewards. Simulation results demonstrate that the proposed method improves the convergence rate by 3× and the accumulated reward by up to 40% compared with the proximal policy optimization (PPO) algorithm.