An Implementation of Effective Machine Learning Approaches to Perform Sybil Attack Detection (SAD) in IoT Network

Abstract

The rapid expansion of the technology industry has resulted in the emergence of a number of new areas of research, one of which is known as "intrusion detection." The objective of an Intrusion Detection System, often known as an IDS, is to categorize user behaviors as either benign or malicious, and then to notify the relevant parties in accordance with this classification. However, a variety of strategies for attack detection have been developed, such as the Sybil attack; however, present techniques are restricted to concentrating on both aspects simultaneously because to constraints in detection accuracy and energy consumption. One such strategy is the Sybil attack. In order to circumvent these restrictions, a framework for the detection of Sybil attacks that is more effective in terms of detection accuracy (security) and energy usage (power) has been presented. Nevertheless, the suggested structure is simple, uncomplicated to understand, and does not need any computing requirements. In this particular architecture, in addition to the Cooja-Contiki simulator, three distinct machine learning methods are used. The NSL-KDD dataset is used in order to test the performance of the proposed framework. This dataset attained the maximum accuracy possible, which was 99.1s%. In addition, in this study, a comparison examination of the suggested work with an existing technique that is considered to be state-of-the-art is carried out in order to determine whether approach is more effective.