Anti-eavesdropping attack industrial internet of things task security offloading method

By improving the MMSAC algorithm and combining divergence regularization and distributed RL, the problem of privacy data leakage caused by eavesdropping attacks in industrial IoT systems is solved, achieving high efficiency and low energy consumption for secure task offloading, and is suitable for industrial IoT systems with multiple terminal devices.

CN117499926BActive Publication Date: 2026-07-24HOHAI UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
HOHAI UNIV
Filing Date
2023-11-02
Publication Date
2026-07-24

AI Technical Summary

Technical Problem

Traditional industrial IoT task offloading methods suffer from non-stationarity, slow convergence speed, low convergence accuracy, and low sample utilization when facing eavesdropping attacks. They also cannot effectively resist eavesdropping attacks, leading to the leakage of privacy data.

Method used

An improved Multi-Agent Soft Actor-Critic (MMSAC) algorithm is adopted, combined with divergence regularization and distributed RL concepts, to design a multi-agent turn-based training mechanism. Furthermore, an improved experience replay technique is used to enhance policy convergence accuracy and sample utilization, thereby achieving safe offloading of distributed tasks.

Benefits of technology

While resisting eavesdropping attacks, it reduces task processing latency and energy consumption, and realizes the confidential transmission and processing of task data, making it suitable for industrial IoT systems with a large number of terminal devices.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117499926B_ABST
    Figure CN117499926B_ABST
Patent Text Reader

Abstract

The application discloses an anti-eavesdropping attack industrial internet of things task security offloading method, and the specific steps are as follows: step 1, setting each parameter of the IIoT system model; step 2, an improved multi-agent soft actor-critic algorithm is proposed to make distributed task security offloading decisions for each IIoT device, the goal is to minimize the system long-term total cost while resisting eavesdropping attacks, including time delay cost and energy consumption cost. The algorithm first uses divergence regularization and distributed RL to improve the traditional SAC algorithm, which improves the convergence accuracy of the strategy. Then, the algorithm proposes a multi-agent round-robin training mechanism, which solves the non-stationarity and non-convergence problems of the traditional distributed multi-agent deep reinforcement learning algorithm. The application can effectively reduce the task processing delay and energy consumption in the IIoT system while resisting eavesdropping attacks, realize the secure transmission and processing of task data, and is suitable for IIoT systems with a large number of terminal devices.
Need to check novelty before this filing date? Find Prior Art