An optimal DoS attack scheduling method based on a deep learning algorithm

By constructing an optimal DoS attack scheduling method using deep learning algorithms, this approach solves the problem of traditional DoS attacks being difficult to defend against in cyber-physical systems, enabling more efficient and flexible attack strategies and enhancing the challenges of network security defense.

CN122294115APending Publication Date: 2026-06-26BEIJING JINGHANG COMPUTING & COMM RES INST
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
BEIJING JINGHANG COMPUTING & COMM RES INST
Filing Date
2026-03-20
Publication Date
2026-06-26

AI Technical Summary

Technical Problem

Traditional DoS attacks are difficult to defend against in cyber-physical systems. With the development of network technology, attack strategies have become more complex and covert, leading to the depletion of system resources, affecting the normal operation of physical devices, and even causing security incidents.

Method used

An optimal DoS attack scheduling method based on deep learning algorithms is adopted. By establishing a mathematical model of a distributed cyber-physical system, using Kalman filtering to obtain local state estimation information, constructing a Markov decision process model, and combining the signal-to-interference-plus-noise ratio and attack energy allocation, a neural network is used to approximate the optimal attack strategy and output an attack energy scheduling strategy that satisfies stability constraints.

Benefits of technology

It improves the efficiency and effectiveness of DoS attacks, enhances the attacker's advantage, reduces resource waste, increases the adaptability and flexibility of attacks, and elevates its position in network warfare.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122294115A_ABST
    Figure CN122294115A_ABST
Patent Text Reader

Abstract

This invention relates to an optimal DoS attack scheduling method based on deep learning algorithms, comprising: establishing a mathematical model of a distributed cyber-physical system; obtaining local state estimation information through Kalman filtering and transmitting it to a remote estimator via a wireless channel; establishing an attack-channel coupling model based on the signal-to-interference-plus-noise ratio (SINR), whereby the attacker injects attack energy into the wireless channel to reduce the probability of successful packet reception, and obtains the remote estimation error covariance as the system state by eavesdropping on confirmation character information; constructing a Markov decision process model that maximizes the discounted cumulative reward, using the error covariance as the state, attack energy allocation as the action, and the trade-off between estimation error and attack energy as the immediate reward; solving the Markov decision process model using a deep Q-network, approximating the optimal Q-function through a neural network, and outputting an optimal attack energy scheduling strategy that satisfies stability constraints. This invention improves the efficiency and effectiveness of DoS attacks, enhancing the attacker's advantage in network warfare.
Need to check novelty before this filing date? Find Prior Art