Privacy-Preserving Data Aggregation in IoT Networks Using Homomorphic Encryption

The work being undertaken in this study focuses on the privacy-preserving data aggregation in Internet of Things (IoT) networks via homomorphic encryption. Since data is being collected and transmitted by IoT devices, protection of data is essential particularly at aggregation stage. Most conventional aggregation procedures involve decryption of data which in turn leaves it vulnerable to hacking. Homomorphic encryption is presented as a solution to compute on the encrypted data, so that the details remain private and there is no necessity to decrypt them. Therefore, we discuss about the existing homomorphic encryption like Paillier, BGV, and CKKS to identify the potentiality of those schemes in IoT networks. All the schemes differ in terms of the rate of encryption and decryption, computational complexity, and accuracy. Our approach is to encrypt the IoT sensor data with these schemes, apply the aggregation operations on the encrypted data and see the decryption to confirm the precision. The fact is that according to the performance analysis, Paillier encryption can be considered as a secure enough approach as for the computational load it is lower compared to the BGV and CKKS methods. Even though BGV and CKKS increase the security levels of computing, they require higher levels of computation. Analysis of accuracy results indicate that HE schemes have a minimal level of errors in the aggregated data to attain privacy.