Detection of DDoS Vulnerability in Cloud Computing Using the Perplexed Bayes Classifier

Abstract

Cloud computing security has been a critical issue with its increase in demand. One of the most challenging problems in cloud computing is detecting distributed denial-of-service (DDoS) attacks. The attack detection framework for the DDoS attack is tricky because of its nonlinear nature of interruption activities, atypical system traffic behaviour, and many features in the problem space. As a result, creating defensive solutions against these attacks is critical for mainstream cloud computing adoption. In this novel research, by using performance parameters, perplexed-based classifiers with and without feature selection will be compared with the existing machine learning algorithms such as naïve Bayes and random forest to prove the efficacy of the perplexed-based classification algorithm. Comparing the performance parameters like accuracy, sensitivity, and specificity, the proposed algorithm has an accuracy of 99%, which is higher than the existing algorithms, proving that the proposed algorithm is highly efficient in detecting the DDoS attacks in cloud computing systems. To extend our research in the area of nature-inspired computing, we compared our perplexed Bayes classifier feature selection with nature-inspired feature selection like genetic algorithm (GA) and particle swarm optimization (PSO) and found that our classifier is highly efficient in comparison with GA and PSO and their accuracies are 2% and 8%, respectively, less than those of perplexed Bayes classifier.