DETECTING HACKS AND MALICIOUS ATTACKS ON IOT DEVICES

The fast expansion of IoT devices has resulted in substantial growth of security weaknesses that cyber threats exploit as principal attack vectors. New cybersecurity techniques struggle to adapt to emerging threats so security demands detection solutions that advance beyond traditional approaches. The paper examined the ways in which machine learning technology improves IoT security by creating algorithms that quickly detect cyber threats and counteract unauthorized access and detect anomalous system behavior. The article utilized a structured approach that begins with obtaining IoT environment data from network traffic logs and device activity records. Data cleaning and missing value handling along with outlier detection were performed during pre-processing to maintain data integrity. The feature extraction process focused on identifying patterns in time series data as well as patterns of packet misbehavior and device network interactions. Standard evaluation methods including cross-validation and hyperparameter tuning provided support for the decision tree, random forest, and deep learning algorithm. Model evaluation included accuracy using recall and F1 score for measurement. The difference of the proposed model is in the use of data encryption, authentication, traffic analysis and machine learning. This model can be used in systems for detecting and preventing malicious actions on Internet of Things devices. The trained security models were able to identify malicious activities occurring in IoT networks. The decision tree analysis achieved an accuracy level of 98%. The obtained experimental results demonstrated that machine learning algorithms differentiate IoT network traffic between authentic procedures and security threats. According to the results, machine learning-based intrusion detection systems effectively protected IoT devices from cyber threats.