Scalable Ddos Attack Detection in Cloud Computing Using Neural Networks

In a Distributed Denial of Service (DDoS) assault, hackers sneakily insert malicious data packets into client networks' internet traffic, causing a catastrophic outage for end users in a cloud computing environment. This is a major issue with cloud computing as detecting and mitigating intrusions is a difficult operation that will impact the whole architecture's functioning. In order to construct machine learning algorithms that can correctly and efficiently detect intrusions, rigorous literature studies are conducted to explore different machine learning approaches, neural network models, and optimization algorithms. The goal is to discover any gaps or flaws and then create these algorithms. Intrusion detection tactics based on Back Propagation Neural (BPN) and Multi-layer Perceptron (MLP) are created in the ANN model to identify DDoS. This paper proposes a hybrid optimization method that combines the exploration and exploitation capabilities of PSO and HHO in order to achieve performance. The suggested hybrid optimiser successfully addresses the drawbacks of traditional algorithms, including the problems of local stagnation, delayed convergence, and optimum trapping of both the local and global levels. The neural network models' weight and bias variables are fine-tuned using the suggested hybrid Harris Hawks Optimization and Particle Swarm Optimization (HHO-PSO) method, which is also used for feature selection. The suggested technique, HHO-PSO with BPN, has a 95.73 percent success rate. When compared to MLP, BPN, and HHO-PSO with MLP, alternative approaches exhibit poorer accuracy, precision, and recall levels that are reflected by their respective F1 ratings. The suggested hybrid method, on the other hand, shows that it is more effective than these models.