An Unsupervised Machine Learning Algorithm for Attack and Anomaly Detection in IoT Sensors

The development of IoT-based sensor systems, such as many control systems in industries, is essential, and sensors are essential for detecting chemical and biological threats, monitoring the environment, and implementing intelligent home automation. Data from Wireless SensorNetwork is retrieved via IoT devices and software and then sent to a distant location for analysis. The manufacturing process is more intricate, with upkeep expenses prohibitive. Internet connectivity raises concerns about network reliability with 'big data'.Machine learning (ML) has various benefits in all cases involving Internet of Things (IoT) sensor systems. These include the speedy processing of 'big data,' which significantly improves the description of information and modelling of predictions. Threats and attacks in IoT infrastructure are increasing with the widespread adoption of wireless sensor systems across industries. Due to their adaptability and scalability, Mobile Adhoc Networks (MANET) are widely used with Internet of Things (IoT) sensors in various contexts. MANET poses security vulnerabilities in an IoT sensor environment because of its adaptability. The Blackhole and Grayhole assault are examples of dangerous routing attacks that can compromise a whole network's transmissions. The advanced hybridized optimization technique AHGFFA is proposed to avoid these issues using unsupervised machine learning in the MANET-IoT sensors system. MANET-IoT sensors' Security monitors (SMs) offer a new group-based routing method incorporating suggestion filtering. Recommendation filtering in the IoT sensor is optimized for an unsupervised machine-learning technique. Secure Certificate-based Group Formation (SCGF) is a technique that is used to organize the entire network into manageable groups initially. The Recommended Action K-means (K-RF means) filtering uses a trust calculation between set members as a criterion for inclusion. The Advanced hybridized optimization technique (AHGFFA) combines the Genetic style and Fire Fly Algorithm to choose safe and efficient routes. Network simulator-3 confirms the success of the proposed work, and the results demonstrate improved Performance. The IoT sensor is protected by group formation and trust filtering from attacks and anomalies.