Secure Retrieval of Sub-Trees over Encrypted Data Stored in Cloud Storage for Prediction

Abstract

Objectives: A tremendous growth in adoption of cloud storage services is observed since 2010. Shifting towards the services however leads to some security concerns like data leakage, unauthorized access and data privacy etc. The main objective is to develop a subtree retrieval method that securely that is further used for prediction without impacting performance. Methods: Cryptographic methods are used to secure data. In our approach the dataset is partitioned based on classification and encrypted before uploading to cloud. Retrieval query returns a subtree from the partitions only when the secret key is matched. Findings: Our proposed novel approach results in better performance compared to other methods of partitioning. Also securely retrieves subtree over the encrypted data. Application/Improvements: This approach can be applied to classify patient electronic health records in cloud storage and query the encrypted data in order to make decisions.