An Efficient IoT Based Intrusion Detection System Using Optimization Kernel Extreme Learning Machine

The Internet of Things (IoT) is an ever-expanding network that links all objects to the web so that they can communicate with one another using standardized protocols. Recently, IoT networks have been extensively used in advanced applications like smart factories, smart homes, smart grids, smart cities, etc. They can be used in conjunction with artificial intelligence (AI) and machine learning to facilitate a data collection procedure that is both simplified and more dynamic. Along with the services provided by IoT applications, various security issues are also raised. The accessing of IoT devices is mainly through an untrusted network like the Internet which makes them unprotected against a wide range of malicious attacks. The detection performance of current IDSs is hindered by issues including false alarms, low detection rate, an unbalanced dataset, and slow response time. This study proposes a new intrusion detection system (IDS) for the IoT that utilizes the chaotic improved Black Widow Optimization Kernel Extreme Learning Machine (CIBWO-KELM) algorithm to address these problems. Initially, the pre-processing of the dataset is carried out using min-max normalization, changing string values to numerical values and changing IP address to numerical values. The selection of the highest performing feature set is achieved through the information gain method (IGM), and finally, the intrusion detection is performed by the CIBWO-KELM algorithm. Python is the tool utilized for testing, while the BoT-IoT dataset is used for simulation analysis. The suggested model achieves an accuracy level of 99.7% when applied to the BoT-IoT dataset. In addition, the results of the studies demonstrate that the proposed model outperforms other current techniques.