ABSTRACT

In a Denial of Service (DoS) attack, an attacker tries to stop authorized users from retrieving data or services. By pointing computer and its internet connection that you are trying to use, an attacker may be unable you in retrieving websites, email, online banking accounts or other information that trust on the infected computer. When a DoS attack occurs, a computer or a network user may not be able to access resources like Internet and email. An attack can be at the network or on operating system. DDoS degrade the performance of a system. System requires a robust detection method from DDoS. To explore and propose the potential issues for detect the attacks. To develop ANN for detecting the DDoS attacks using Neural Networks.First of all, explore and detect the issue then analysis of attacks after that train the detector after training complete then testing was performed. Evaluation on the base of the attacks and detect. The study has used Resilient Back propagation, Scaled Conjugate Gradient, and Conjugate Gradient with Powell/Beale Restarts three algorithms for training data in MATLAB tool to detect attacks. Resilient Back propagation algorithm is accurate and has minimal storage requirements and performs well for complex problems. However, in this research ANN, Conjugate Gradient is also suitable and is able to give similar performance with less number of iterations. The study used Feed forward NN and research also performs experimental analysis of DDoS attacks using artificial neural network and it is highly recommended for other researchers may use these network Regularity Feedback, Radial Base Function (RBF) future researcher may us (RBF) and Recurrent neural network