Detection and Prevention of DOS Attacks in Cloud Data Using J48 Algorithm Compared with Random Forest Algorithm for Improved Prediction Rate

Abstract

Aim: The aim of  this research is to detect the Denial of Service (DOS) attacks using two machine learning algorithms, the Novel J48 algorithm and Random forest algorithm and compare accuracy to evaluate efficiency of two machine learning algorithms. Materials and Methods: Considering Multiple Novel J48 algorithms as group 1 and random forest algorithms as group 2  process was implemented to predict DoS attacks  and to get a prediction rate to compare algorithms. The algorithm should be efficient enough to detect the exact type of DoS attack . The sample size considered for implementing this work was N=20 for each of the groups considered. The sample size calculation was done with spss. The pretest analysis was kept at 80%. Sample size is estimated using G-power. Results:  Based on statistical analysis, the significance value for calculating accuracy  was found to be 0.048. The Novel J48 Algorithm gives a slightly better accuracy rate with a mean Flow_Packets_Sec percentage of 89.69% and Random forest algorithm has a mean Flow_Packets_Sec of 75.29% with a significant value of two tailed tests is 0.048 (p<0.05) with 95% confidence interval. Conclusion: Through this, prediction is done for detection of DoS attacks and the Novel J48 algorithm gives a slightly better prediction rate value than the Random forest algorithm.