Ping Flood Attack Pattern Recognition on Internet of Things Network Dataset

Ping Flood Attack Pattern Recognition using K-Means Algorithm in Internet of Things (IoT) Network status: on repository Abstract — This work investigates ping flood attack pattern recognition on Internet of Things (IoT) network. Experiments are conducted on WiFi communication with three different scenarios: normal traffic, attack traffic, and normal-attack combination traffic to create normal dataset, attack dataset, and normal attack (combined) dataset. The datasets are grouped into two clusters i.e.: (i) normal cluster and (ii) attack cluster. Clustering results using implemented K-Means algorithm show the average number of packets on the cluster of attack in total is 95,931 packets, and the average packets on normal cluster in total is 4,068 packets. Accuracy level of the clustering results then is calculated using confusion matrix equation. Based on the confusion matrix calculation, accuracy of clustering using implemented K-Means algorithm was 99.94%. The true negative rate reaches up to 98.62%, true positive rate is 100%, the false negative rate is 0%, and the false positive rate reaches 1.38%.

基于K-Means算法（K-Means Algorithm）的物联网（Internet of Things，IoT）网络Ping泛洪攻击（Ping Flood Attack）模式识别
数据集状态：已收录至公开存储库
摘要 — 本研究针对物联网（IoT）网络中的Ping泛洪攻击模式识别展开探究。本实验以WiFi通信为载体，设置三类实验场景：正常流量场景、攻击流量场景以及正常-攻击混合流量场景，分别构建正常数据集、攻击数据集与正常-攻击混合数据集。随后将所有数据集划分为两个聚类簇：(i) 正常流量簇；(ii) 攻击流量簇。采用所实现的K-Means算法得到的聚类结果显示：攻击流量簇的平均总数据包数为95931个，正常流量簇的平均总数据包数为4068个。随后通过混淆矩阵公式计算聚类结果的准确率指标。经混淆矩阵计算，所实现的K-Means算法的聚类准确率达99.94%。其中真阴性率为98.62%，真阳性率为100%，假阴性率为0%，假阳性率为1.38%。