Detecting and Preventing ARP Spoofing Attacks Using Real-Time Data Analysis and Machine Learning

Abstract

ARP spoofing attacks contain certain risks in networks as they seem to intercept traffic and can lead the leakage of intellectual information. This research paper focuses on enhancing the method through which five algorithms namely: Random Forest, Long Short-Term Memory (LSTM) Networks, Convolutional Neural Networks (CNNs), Support Vector Machines (SVM) and Isolation Forest for ARP spoofing detection and prevention. In the process of the experiment, each algorithm is tested with the dataset of ARP traffic and the results are compared on the five criteria: of data; these are accuracy, precision, recall, F1-score, false positive rate, and the false negative rate. It can therefore be deduced that out of all the algorithms employed, Random Forest has the highest accuracy of 94 and high values of precision and recall thus making it more efficient in real-time ARP spoofing detection. Its effectiveness is equally high as the effectiveness of LSTM Networks and CNNs, which process temporal or spatial data, but work longer. SVMs are comparatively not bad in terms of accuracy to noise ratio, however, they are less accurate as compared to both Random Forest as well as CNNs. This method however lacks good accuracy and has high error values as portrayed above with Isolation Forest. Based on this analysis, conclusions are made that use of higher levels of ML leads to the detection of ARP spoofing implementing Random Forest as the best solution for enhancing the network security.