Efficient Large-Scale Intrusion Identification and Prevention in Distributed Cloud Networks Using Artificial Intelligence

The rapid expansion of distributed cloud networks has significantly increased cybersecurity threats, particularly DDoS attacks that disrupt system stability and performance. An employ of AI-driven solutions is necessary since traditional intrusion detection systems are unable to effectively detect changing assault patterns. This study presents a highly efficient intrusion detection system leveraging the LSTM model trained on the CICDDoS2019 dataset. Comprehensive data preprocessing, including encoding, scaling, and adaptive synthetic sampling (ADASYN), enhances model learning. As compared to more conventional machine learning models like Logistic Regression, ENN, and Bayes CNN, the suggested model achieves a remarkable 99.9 percent accuracy. A model’s ability to identify malicious and benign traffic with a high recall and precision of 99.9 percent demonstrates its efficacy in reducing false positives. Confusion matrices, ROC curves, and trends in training-validation loss, among other performance assessment measures, corroborate robust generalization and low overfitting. A dependable solution for intrusion detection and prevention in large-scale cloud networks, LSTM stands out due to its excellent classification capabilities, as shown by comparative analysis.