Title: LR-SDP: lightweight privacy preservation approach for distributed mobile devices in IoT environment

Authors: Vipin Vijayachandran; Suchithra Ramachandran Nair

Addresses: Department of Computer Science, Jain University, Bengaluru, Karnataka-560069, India ' Presidency College, Hebbal Kempapura, Bengaluru, Karnataka 560024, India

Abstract: A distributed mobile system comprises several mobile devices that process and exchange data or results with one another. Mobile users must protect their privacy when receiving and sending the data to and from the local devices. The raw data in the dataset is pre-processed using mean interpolation, most frequent value mean approach and binning. After the data cleaning procedure, the pre-processed data is fed into the logistic regression-based statistical differential privacy (LR-SDP) algorithm. The Laplace mechanism adds noise with a zero mean and adjustable standard deviation, safeguarding data specifics during noise introduction and removal while making the dataset resilient against inference attacks. The Pima Indians Diabetes Database is employed in this paper, and the proposed model attained an accuracy of 81.16% when using a decision tree (DT) and 79.22% when using a random forest (RF). Similarly, the support vector machine (SVM) and deep neural network (DNN) achieves 82.46% and 83.11% accuracy, respectively.

Keywords: logistic regression; differential privacy; Laplace approach; distributed mobile system; privacy preservation; support vector machine; SVM; deep neural network; DNN.

DOI: 10.1504/IJSN.2024.141788

International Journal of Security and Networks, 2024 Vol.19 No.3, pp.138 - 149

Received: 24 Apr 2024
Accepted: 29 Jul 2024
Published online: 01 Oct 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article