Enhancing Security and Privacy in IoT-Based Learning with Homomorphic Encryption

Abstract

The security and privacy of data in an IoT-driven intelligence landscape is a major concern. This research examines the integration of Paillier homomorphic encryption into Federated Learning to enhance security while maintaining individual data privacy in such environments. The interconnectedness of devices in IoT frameworks poses a challenge in maintaining the confidentiality of sensitive information. By using Paillier encryption within Federated Learning, this problem is solved by securing learning parameters while still keeping data private. This approach demonstrates promising improvements without violating privacy through extensive simulations and comparative analyses across different model architectures. The results of this study highlight the potential effectiveness of this method for enhancing security measures in interconnected IoT environments.