边缘计算环境下基于深度学习的DDos检测

A DDoS Detection Methodology Based on Deep Learning in Edge Computing Environment

摘要:

边缘计算作为一种用于降低中心节点计算压力，更靠近终端设备和数据源头的新计算范式，满足了计算业务下沉的需求，也带来了安全问题；其中，对边缘计算安全威胁最大、造成过巨大经济损失和安全事故的当属分布式拒绝服务攻击（DDos）；边缘计算环境下由于算力受限、存储空间有限等原因，传统的防御手段难以应用；因此，提出了一种适用于边缘计算环境下的基于深度学习的轻量级DDos检测框架；采用CIC-DDos-2019数据集来模拟边缘计算环境下的遭受DDos攻击的网络流量，针对数据集进行了适应性强的预处理技术和相似性标签融合，运用SMOTE算法解决了数据集类别不平衡问题，采用一维卷积技术和BiLSTM技术搭建了模型并进行了模型剪枝，构建了一个轻量级模型。结果表明，其针对DDos攻击类别的八分类实验准确率达到了96.8%，二分类实验准确率达到了99.8%。

Abstract:

Edge computing, as a new computing paradigm that solves the pressure of central nodes and is closer to terminal devices and data sources, fulfils the needs of Calculate subsidence, but also brings security crisis. Among them, DDos are the ones that most threaten the security of edge computing and cause enormous economic losses and security accidents. In edge computing environment, traditional defense methods are difficult to apply due to limited computing power and storage space. Therefore, this paper proposes a lightweight DDoS detection framework based on deep learning for edge computing environment. In this paper, cic-ddos-2019 data set is used to simulate the network traffic attacked by DDoS in the edge computing environment. The adaptive preprocessing technology and similarity label fusion are carried out for the data set. The smote algorithm is used to solve the problem of data set category imbalance, and a lightweight model is constructed by using one-dimensional convolution technology, bilstm technology and model pruning technology. As a result, it achieves 96.8% accuracy in 8-class and 99.8% accuracy in 2-class experiments for DDos attack categories.