Multiple Linear Regression Prediction Model for DDOS Attack Detection in Cloud ELB
Swathi Sambangi, Gondi Lakshmeeswari · 2021
An ongoing research challenge in cloud computing is to address the security and data availability challenges. Although, DDoS attacks in cloud are not new but still they have been continuously throwing new challenges to the network community which makes detection of these attacks an ongoing research challenge with respect to cloud security. One of the reasons for these challenges is the high non-linearity of the real-world data. Thus, we bring into light the importance of understanding the non-linearity of data. Understanding nature of traffic instances in network datasets helps to build efficient machine learning models. For building a machine learning model, we choose to apply regression analysis. Two datasets namely CICIDS 2017 and CICIDS 2019 are considered for the present study as these datasets show high non-linearity. In this paper, we propose to apply regression analysis after performing feature engineering addressing the problem of DDoS attack detection. We propose to visualize the regression model by plotting residual plot, fit chart. The learning models can also be evaluated by comparing their respective MAPE and accuracy values. To the best of our knowledge, the research addressed in this paper is the first contribution in cloud computing which depicts the importance data visualization in analyzing the machine learning models. We believe that this paper paves a way for future researchers in cloud computing to concentrate on data visualization.