A Robust Deep-learning System for Detecting and Classifying Cyber-attacks in Cloud Computing

Abstract

Abstract The Internet of Things (IoT) has emerged as the benchmark for low-power lossy networks (LLNs) due to the rapid proliferation of low-power, high-throughput technology. Still, IoT systems are open to cyber attacks because the base devices have limited computing, storage, and communication capabilities. As people rely more and more on cloud computers to store and handle important data, cyber dangers that target cloud settings have become more complex and widespread. The identification and categorization of harmful network activity is crucial in safeguarding the integrity and protection of these systems. Here, we present the full design and implementation of a deep-learning-based system called Enhanced battle royal optimized convolutional neural network (EBRO-CNN) for detecting and classifying cyber attacks in cloud-based telecommunications systems. This optimization method creates tiers for network traffic data depending on characteristics of the cloud. Each layer undergoes its own analysis and optimization to meet its own requirements. This optimization across tiers not only increases detection precision but also makes the system more scalable and flexible in the face of ever-changing cloud workloads. We used the NSL-KDD dataset, which contains demonstrations of all the major cloud computing threats, to gauge the effectiveness of the solution we designed. With an accuracy of up to 99% in various configurations, the findings show that the suggested distributed architecture is useful for detecting various cyber attacks.