Exploring the effectiveness of multiclass decision jungle for internet of things security

Abstract

Network intrusion detection systems (NIDS) are vital in protecting computer networks against cyber security incidents. The relationship between NIDS and internet of things (IoT) security is pivotal and NIDS plays a significant role in ensuring the security and reliability of IoT ecosystems. Ensuring the security of IoT devices is critical for several reasons. It helps safeguard sensitive information, guarantees the dependability of crucial infrastructure, meets regulatory obligations, and fosters user confidence. As the IoT ecosystem expands, prioritizing security is essential to minimize risks and maximize the benefits of connected devices. Given the ever-expanding cyber threat landscape, the multiclass classification task is essential to empower the NIDS with an ability to distinguish between various attack patterns in less computational time. The multiclass decision jungle algorithm is investigated to optimize the performance of NIDS. The research has considered permutation feature importance to include only the relevant features from the data. Using a contemporary dataset such as CICIOT 2023, the study has demonstrated an impressive attack detection rate of over 90% for 20 modern attack types. This research has investigated the effectiveness of IoT security measures and its prospective contributions to the field of cyber security.