Main Article Content
Vulnerabilities in smart home (IoT) platforms make it possible for intruders to perform attacks in a variety of settings, including home automation, industrial automation, and sophisticated health systems. Research has developed a variety of comprehensive security technologies to get around this cyber-attack obstacle. Machine Learning (ML), which is being deployed, has been identified as the most viable method. Consequently, the majority of ML approaches solely concentrate on researching suitable learning models in order to increase the recognition rate. However, a lack of suitable identification characteristics frequently contributes to the limits in terms of recognition rate in a variety of assaults. The present approaches, however, are inadequate to cover the comprehensive security spectral range of IoT environments due to the distinctive characteristics of IoT nodes. Furthermore, the majority of previous efforts lacked implementation structures and methods for defending against cyber-attacks. As a result, in this research, we examine the characteristics of several smart home security threats as well as the value of the information that may be extracted and used in Ml techniques to effectively identify any of these cyberattacks. Due to the increase in internet traffic, it is more difficult to identify cyberattacks in the IoT as well as identify fraudulent traffic in its initial stages. SVM, RF, LR, and decision tree algorithms were successfully used in machine learning systems to determine and alert users of smart IoT devices to potential threats. A methodology for the identification of malicious cyber activity is suggested in this paper.