Enhancing Cloud Storage Security with Homomorphic Encryption and ML-Based Access Control

In this research a new approach of improving security of cloud storage through the use of homomorphic encryption and machine learning access control is proposed. One of the most important deficiencies the proposed system relevant to the critical issue of protecting data during its life cycle while enabling computations on them in an encrypted form without decrypting them. Thus, the system minimizes both the unauthorized access rights and the user behavior affecting the data integrity and confidentiality relying on the approaches of the supervised learning. It uses partially homomorphic encryption based on the Paillier cryptosystem as well as RandomForest and NeuralNetwork models. Experimental evaluation performed in a cloud-like environment shows that a system achieves desired security while providing low latency and high accuracy under high working load. The provided research can be considered as a step to the enhancement of the cloud security technologies and presents efficient solution for the protection of the cloud data storage.