Abstract
Industrial Internet of Things (IIoT) devices have been widely used for monitoring and controlling the process of the automated manufacturing. Due to limited computing capacity of the IIoT sensors in the production line, the scheduling task in production line needs to be offloaded to the edge computing servers (ECS). To obtain desired quality of service (QoS) during offloading scheduling tasks, {the precise interaction information between production line and ECSs have to be uploaded to the} cloud platform, which poses privacy issues. Existing works mostly assume all interaction information, i.e., the offloading decision for the subtask in a scheduling task, have same privacy level, which cannot meet the various privacy requirements of the offloading decision for the subtask. Hence, we propose a local differential privacy-based deep reinforcement learning (LDP-DRL) approach in edge-cloud-assisted IIoT to provide personalized privacy guarantee. The LDP mechanism can generate different level of noise to satisfy various privacy requirements of the offloading decision for the subtask. The prioritized experience replay (PER) is integrated in DRL to reduce the impact of noise on the QoS performance of task offloading. The formal analysis of the LDP-DRL is provided in terms of privacy level and convergence. Finally, the extensive experiments are conducted to evaluate the effectiveness, capacity of privacy protection, the impact of discount factor on the convergence, and cost efficiency of the LDP-DRL approach.