A method and system for SDN active deception defense based on constraint reinforcement learning

By using a constraint reinforcement learning-based approach, an SDN controller is constructed as a defensive agent to predict attacker behavior and optimize deception strategies. This solves the problem of real-time perception and response in dynamic network environments for SDN proactive deception defense technology, achieving efficient and stable network defense.

CN122348867APending Publication Date: 2026-07-07SHENZHEN Y& D ELECTRONICS CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
SHENZHEN Y& D ELECTRONICS CO LTD
Filing Date
2026-06-09
Publication Date
2026-07-07

AI Technical Summary

Technical Problem

Existing SDN proactive deception defense technologies lack real-time perception and response capabilities when facing attacker reconnaissance activities, have insufficient dynamic adaptability, cannot effectively defend in dynamic network environments, and may affect network availability.

Method used

We employ a constraint-based reinforcement learning approach to construct an SDN controller as a defensive agent. By predicting attacker behavior through a long short-term memory network, and combining the Actor-Critic architecture and Lagrange relaxation optimization, we achieve dynamic deception strategy optimization and network performance constraint awareness. We also construct a dual-timescale adversarial training framework for real-time deception response.

Benefits of technology

It enables proactive detection of attacker probing behavior and optimization of dynamic deception strategies, ensuring real-time defense and availability of SDN networks, and improving defense effectiveness and network stability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN122348867A_ABST
    Figure CN122348867A_ABST
Patent Text Reader

Abstract

The application discloses a kind of SDN active deception defense method and system based on constraint reinforcement learning, it is related to information security technical field.The present application includes: SDN controller is modeled as defense agent, define state space, action space, reward function, constraint cost function, convert SDN active deception defense problem into the Markov decision process under multiple performance constraints;Build opponent model based on long short-term memory network and double-time scale confrontation training framework, the defense strategy network of training convergence is deployed to SDN controller, and state construction, strategy reasoning, constraint check and deception action response are executed to each arriving probe packet.The present application can actively perceive attacker probe behavior, dynamically optimize deception strategy and strictly guarantee the active deception defense system of SDN network availability, realizes the unification optimization of attack behavior active perception, deception strategy dynamic optimization and network performance constraint perception.
Need to check novelty before this filing date? Find Prior Art