Search for a command to run...
This paper studied the application of combining biometric security approaches and machine learning algorithms to improve security in IoT environment. With information about IoT devices being inter-linked, the conventional security mechanisms are failing to deliver strong protection mechanism against the advanced cyber threats. Biometrics, identification through unique physical and behavioral features, offers a viable alternative. Biometric security in IoT still suffers from several issues like data privacy risk, computation limitations, and vulnerability to spoof attacks. This study explores different biometric modalities, from scanning the fingerprint at our phone and facial recognition, to iris scanning, voice recognition and behavioral biometric methods, and evaluates their implementations and potential applications with IoT devices. Various machine learning algorithms are evaluated including deep learning, support vector machines, random forests and anomaly detection methods to assess their feasibility for biometric security. Additionally, the study investigates the integration of several biometric modalities to improve authentication performance and mitigate unauthorized access. It achieves this by conducting an experimental analysis, followed by carrying out simulation to assess the accuracy, time consumption, and attack resistance of these machine learning-based biometric security systems. The results show that the deep learning models have better accuracy and robustness to noise in biometric recognition tasks than traditional machine learning approaches, especially convolutional neural networks (CNNs) and recurrent neural networks (RNNs). Moreover, federated learning and edge computing appear as two promising paradigms of providing efficient solutions to data privacy issue and computational restrictions in IoT systems with limited resources. This study underlines the benefits of combining biometric security with machine learning for use in IoT applications, while paying due attention to specific issues surrounding security vulnerabilities, computational efficiency, and user privacy. They help in building efficient mechanisms of secure and intelligent cybersecurity of IoT frameworks in the future.