Enhancing Cloud Database Security Through Intelligent Threat Detection and Risk Mitigation
Satyadhar Kumar Chintagunta, Siddhesh Amrale · Technix International Journal for Engineering Research · 2022
Cloud computing now forms a significant part of contemporary data storage and management through being able to provide scalable and on-demand access to resources and databases. The availability, confidentiality, and integrity of stored data in cloud databases are at danger from cyber hazards such as DOS attacks, online attacks, port scans, and unauthorized access. Network traffic data is complex, imbalanced and high-dimensional, making it difficult to detect these threats. This research proposes a multi-class LSTM model to enhance cloud database security by detecting intelligent threats and reducing risk. The CICIDS2017 data was used, and the pre-processing activities included the management of missing values, duplications, one-hot encoding of categorical variables, balancing the classes with the help of SMOTE, and data normalization. The LSTM model of multi-class classification and trained with optimized parameters. The proposed framework performed with 99% accuracy (ACC), 99.9% precision (PRE), 98% recall (REC), and 99% F1-score (F1), surpassing the performance of other models, i.e. Support Vector Machines (SVM), Decision Trees (DT), and Deep Belief Networks (DBN). Besides efficient threat detection, the framework helps reduce risks by providing practical knowledge to prevent unauthorized access and proceed with safe cloud operations. This article introduces a very effective, scalable, and feasible solution to real-time threat detection of cloud databases with high performance, strength, and a base to continue improving it with additional volumes of data and enhanced AI algorithms.