Autoencoder for Design of Mitigation Model for DDOS Attacks via M-DBNN

Abstract

Distributed Denial of Service (DDoS) attacks pose the greatest threat to the continued and efficient operation of the Internet. It can lead to website downtime, lost time and money, disconnection and hosting issues, and website vulnerability. Conventional machine learning methodologies are being harmed by reduced recognition rates and greater false-positive rates due to the emergence of new threats. As a result, high-performance machine learning classifiers with low false-positive rates and high prediction accuracy are required for the DDoS detection system. Here, a deep belief neural network is preferred, upgraded to the modified deep neural network (M-DBNN) to accurately detect DDoS attacks from the network. Enable the database to change a specific format and range, which helps the M-DBNN classifier easily predict the class. An advanced Chimp Optimization Algorithm (ChOA) is used to minimize the error to find the best weight of the M-DBNN classifier; this leads to accurate DDOS attack detection and predict the classes effectively. The proposed method is evaluated for CAIDA “DDoS Attack 2007” dataset. The accuracy of the proposed method is 0.87%, and the outcome is compared with those of other existing methods of deep neural network (DNN), support vector machine (SVM), artificial neural network (ANN), and neural network (NN). The proposed method demonstrates great detection accuracy with a low error.