Dimensional Reduction With Fast ICA for IoT Botnet Detection

Abstract The Internet of Things (IoT) has unique characteristics with a minimalist design and has network access with great scalability, which makes it difficult to control access. Setting up an intrusion detection system (IDS) on an IoT system while taking into account its unique characteristics is a big challenge. In this paper, we propose a dimensional reduction approach utilizing the fast independent component analysis (ICA) method to address scalability issues of IDS for IoT systems. Experimental results show that the reduction of dimensions by the fast ICA method overall improves the IDS execution time and does not significantly affect accuracy.