Deauthentication Attack Detection in the Wi-Fi network by Using ML Techniques

Abstract

The initial encryption technique Wired Equivalent Privacy (WEP) provided by IEEE, had many shortcomings, making Wi-Fi connections vulnerable to attacks. The shortcomings of WEP were overcome by the development of 802.11 standard, which was based on strong encryption schemes enabling client authentication too, a feature not available in WEP. IEEE 802.11 standard's encryption schemes encrypt just data frames, making it susceptible to de-authentication DoS attack. The unencrypted management and control frames that are used for establishment, maintenance and data exchange are exploited for carrying out attacks, especially de-authentication DoSattacks. While encryption, protocol modifications, besides hardware, software and standard upgradation help mitigate the attacks, machine learning (ML) based Intrusion Detection System (IDS) come in handy to manage the WLAN attacks. The choices of classifier algorithms that are based on - probability, kernel, decision tree or rule enable to compare the performance of the classifiers – detection rate. Evaluating the efficiency of DoS, a look at the experimental results highlights that the proposed ML based IDS is better in attack identification with 96% detection capability.