Towards the Development of Realistic Botnet Dataset in the Internet of Things for Network Forensic Analytics: Bot-IoT Dataset

The proliferation of IoT systems, has seen them targeted by malicious third\nparties. To address this, realistic protection and investigation\ncountermeasures need to be developed. Such countermeasures include network\nintrusion detection and network forensic systems. For that purpose, a\nwell-structured and representative dataset is paramount for training and\nvalidating the credibility of the systems. Although there are several network,\nin most cases, not much information is given about the Botnet scenarios that\nwere used. This paper, proposes a new dataset, Bot-IoT, which incorporates\nlegitimate and simulated IoT network traffic, along with various types of\nattacks. We also present a realistic testbed environment for addressing the\nexisting dataset drawbacks of capturing complete network information, accurate\nlabeling, as well as recent and complex attack diversity. Finally, we evaluate\nthe reliability of the BoT-IoT dataset using different statistical and machine\nlearning methods for forensics purposes compared with the existing datasets.\nThis work provides the baseline for allowing botnet identificaiton across\nIoT-specifc networks. The Bot-IoT dataset can be accessed at [1].\n