Efficient privacy-preserving data replication in fog-enabled IoT

Internet of Things (IoT) devices generate a high volume of data that is processed and stored in traditional cloud computing. The processing including data replication in traditional cloud computing often results in excessive resource utilization, performance overhead, and long response time. Fog computing has been proposed to overcome the shortcomings of cloud computing. Fog computing alleviates the processing and storage burden during data replication from the cloud to the network edge closer to sensor devices. Numerous data replica schemes in fog computing have been proposed to improve the performance efficiency of the data, reduce the turnaround delays of data access, and minimize network latency. However, these schemes do not consider data replication privacy, which is essential for data protection, reliability, and authentication. Therefore, this paper proposes a data replica creation scheme and a data replica placement scheme for preserving the privacy of data in fog computing. Our proposed replica creation scheme is based on a Level of Privacy (LoP) defined by data owners and the service capacity of fog nodes. Our proposed replica placement scheme is based on the priority level of fog nodes. We have conducted a comprehensive experimental analysis to compare the performance of our scheme and the existing schemes. Our results demonstrate that the proposed scheme can significantly achieve efficient replicas privacy, prediction accuracy, as well as outperform the existing state-of-the-art schemes in terms of computational and memory costs. (C) 2021 Published by Elsevier B.V.

Automated summary

This paper introduces how data replication schemes in fog computing can improve performance efficiency, reduce data access delays, and minimize network latency. Proposed data replica creation and placement schemes considering data privacy. Experimental results demonstrate the proposed scheme achieves efficient replicas privacy and outperforms existing schemes in terms of computational and memory costs.