IoT-Gen: A DAT ASET GENERATE D FROM REAL loT DEVICES FORANALYSIS OF DoS/DDoSAND OTHERCYBERATTACK SONNETWORKS

Attackers exploit a number of vulnerabilities in Internet of Things (IoT) devices to perpetrate denial-of-service (DoS) and distributed DoS (DDoS) attacks as well as other cyber-attacks in networks. Measures to counter such attacks require availability of existing datasets that can be used to test some network security mechanisms. The 'IoT-Gen' is a new dataset generated from physical and real IoT devices for network traffic. It has 2 attack traffic captures executed in 6 IoT devices and one capture for normal loT devices, totaling 3 packet captures (PCAP) files, which have their converted comma-separated value (CSV) files. The attack traffic was generated by simulating floods of Transport Control Protocol (TCP) and User Datagram Protocol (UDP) with Kali Linux's hping3 in the real loT network testbed. The testbed setup includes a Raspberry Pi 4 configured as router to act as Wi-Fi access point and connects to the internet as well as Raspberry Pi 2 running on Apache Web Server for attack victim. Other testbed hosts include android phone, 3 personal laptop/desktop computers, and the real loT devices that are connected to the Wi-Fi access point. A limitation with the IoT-Gen dataset is that it has yet to be validated using different machine learning (ML) classification algorithms and also evaluating the performance metrics. However, Python programming code has been developed for calculating and visualizing the correlations of the IoT­Gen dataset features, resulting in Pearson Correlation Matrix table and Heatmap.