边缘计算中改进ELM的高效入侵检测算法

An Efficient Edge Computing Intrusion Detection Algorithm Based on Improved ELM

摘要:

边缘计算将云计算扩展到网络边缘,在解决了云计算时延高、移动性差和位置感知弱等缺陷的同时也带来了诸多安全问题。针对边缘计算网络开放性、异构型和节点资源受限等特点,研究设计具有６层结构的通用边缘计算入侵检测系统,并在此模型架构上提出了一个边缘计算入侵检测方案,基于该方案提出了一种适用于边缘计算部署的改进极限学习机的入侵检测算法TSS-ELM,TSS-ELM增加了云服务器训练样本筛选环节来优化机器学习中的外权,从而对边缘节点数据实现高效的入侵检测。仿真实验结果和分析表明,该算法在准确性、时间依赖性、鲁棒性和误报率方面与其他现有算法相比具有更优异的性能。

Abstract:

Edge computing extends the influence of cloud computing to network edge.It solves the problems of high delay, poor mobility and weak position perception in cloud computing, and it also brings many security problems. According to the characteristics of open, heterogeneous and limited node resources of edge computing network, this paper studies and designs an architecture of edge computing intrusion detection system with 6-layer structure, and an edge computing intrusion detection scheme is proposed. Based on this scheme, an intrusion detection algorithm TSS-ELM is proposed, which is suitable for edge computing deployment. This algorithm introduces the process of training sample selection and optimizes the external weight in machine learning, so as to achieve efficient intrusion detection for edge node data. Simulation results and analysis show that the algorithm has better performance in accuracy, time dependence, robustness and false alarm rate than other existing algorithms.