P2CADL : Privacy Preserving in Cloud using Autoencoder based Deep Learning Classifier for Smart City Applications

Nowadays, smart cities enhance the life style of urban people in various disciplines, viz., smart transports, smart environment, smart parking, smart health care, etc. As the cities become smarter, privacy is the major issue and the private sensitive data needs protection. To address this issue, this paper proposes an autoencoder based deep learning classifier by preserving the privacy in cloud for smart city applications. The private data is protected using homomorphic encryption and deep learning is performed in cloud servers. Experiments are conducted on three datasets, viz., Road Traffic Data, Pollution Data, and Parking Data provided by CityPulse Smart City Dataset. It is seen that the proposed approach is efficient compared to conventional classifiers as the non-disclosure of private data in cloud computing. The proposed approach is scalable and it is suitable for big data based smart city applications.