基于强化学习的自适应网络威胁缓解

Adaptive network threat mitigation based on reinforcement learning

随着互联网信息技术的深入发展,通信网络受到攻击入侵威胁也在不断变化,提出一种强化学习算法用于网络自适应威胁缓解,在SDN框架的基础上,研究使用强化学习算法用于网络安全管理。以D3QN算法为基础并对其结构进行了改进,使用改进后的D3QN深度强化学习方法来学习缓解APT攻击,实现网络威胁自适应控制。最后对实验结果进行了评估并给出了改进算法模型的收敛结果,验证了该强化学习方法用于自适应网络威胁缓解的可用性和有效性。

With the in-depth development of internet information technology,communication networks are constantly threatened by attacks and intrusions.A reinforcement learning algorithm is proposed for network adaptive threat mitigation.Based on the SDN frame⁃work,the use of reinforcement learning algorithms for network security management is studied.Based on the D3QN algorithm and im⁃proved its structure,the improved D3QN deep reinforcement learning method is used to learn and mitigate APT attacks,achieving adaptive control of network threats.Finally,the experimental results were evaluated and the convergence results of the improved algorithm model were provided,verifying the availability and effectiveness of the reinforcement learning method for adaptive network threat mitigation.