Evaluation of Different Machine Learning Classifiers on New IoT Dataset CICIoT2023
Anmol Geetesh Kumar, Ansh Rastogi, Virender Ranga
Abstract
In today's society, Internet of Things is playing a profound role in revolutionizing a number of industries. It's use in sectors like transportation, smart home appliances, & healthcare is rapidly growing, with new services constantly being developed. Over the past decade, there has been an exponential surge in IoT connections, and this trend is expected to continue in the coming years across different domains. However, it has also exposed a new frontier for cyber-attacks. The sensitive nature of data collected by IoT devices demands stringent security measures. This paper analyzes the classification performance of various cyber-attacks in the IoT domain. For performance evaluation, the cutting-edge CICIoT2023 dataset is used [1]. The dataset encompasses a variety of attacks carried out against 105 IoT devices, which includes benign traffic, DoS and DDoS attacks, reconnaissance and information gathering, web-based attacks, spoofing, brute-force threats, and Mirai botnet attacks. This dataset encompasses wide range of cyber threats & serves as a valuable tool for understanding and mitigating IoT-related security threats.