Development of Authentication Framework Using GAIT for IoT Communication Models

The data collected by the authorised sensor devices is transferred to the cloud for safe storage. Communication occurs via a secure key management system designed for a decentralised network. Subsequently, it is important to guarantee that only authorised individuals are able to get this data from the cloud. Ensuring user authentication is of utmost importance in resource-constrained IoT devices due to the large number of users accessing sensitive data. The current user authentication methods suffer from accuracy, error rate, and implementation time problems. A novel solution is offered to tackle these challenges by introducing a sophisticated stacked ensemble authentication architecture that utilises cellphones for IoT systems. The suggested authentication approach utilises gait analysis to authenticate individuals by recognising their unique patterns of human movement. A smartphone equipped with an accelerometer and gyroscope sensor is used to gather gait characteristics. The proposed system has two components: preprocessors and classifiers. The preprocessor performs noise reduction and feature extraction. Subsequently, stacked ensemble machine learning classification models, including base and meta classifiers, are used to classify the characteristics, such as linear acceleration, rotation rate, and gravity, for the purpose of user authentication. The accuracy and error rate of the outcomes are compared with the schemes that already exist. The observation indicates that the suggested framework attains a 95% accuracy rate, a lower error rate, and optimised execution time.