SDESA: secure cloud computing with gradient deep belief network and congruential advanced encryption

With the aim to enhance cloud security with higher data confidentiality rate and integrity, we propose a novel technique called Stochastic Deep Encryption Standard Algorithm (SDESA). After the successful registration of the cloud user with the Cloud Server, the authentication process is executed using a Stochastic Nesterov Gradient Piecewise Deep Belief Network model, which distinguishes between authorized and unauthorized users with the greatest accuracy and least time. Gestalt pattern matching method is used to find legitimate users. Stochastic Nesterov gradient approach is adapted to decrease the classification error, thus boosting the attack detection accuracy. Linear Congruential Ephemeral Advanced Encryption Standard Algorithm, which uses the Linear Congruential Ephemeral Secret Random Key for data encryption/ decryption to protect data by preventing unauthorized access, is finally introduced to guarantee the data confidentiality and integrity. The results indicate that SDESA outperforms conventional methods in terms of various performance metrics under different scenarios.