A Privacy-Preserving Sensor Aggregation Model Based Deep Learning in Large Scale Internet of Things Applications
Agus Kurniawan, Marcel Kyas · 2019
Privacy-preservation in data aggregation in large scale Internet of Things applications is challenging. Sensitive data from a result of collecting sensor data needs attentions to address privacy issues. We present a privacy-preserving model to protect data aggregation between sensor gateway and storage servers. Our proposed scheme is designed for decentralized networks and passwordless by obfuscating sensor data. We design, implement and evaluate a practical privacy-preserving system using deep learning autoencoder with convolutional neural network architecture. We do a statistical analysis and perform simulation on computer and IoT board machines. Evaluation process involves training and testing phases with a dataset. We measure system accuracy and computation time. The simulation and experimental results show that privacy-preserving-based deep learning model can address privacy issues on data aggregation and guarantee scalability and performance on applications.