Detection of Deauthentication Threats in Wi-Fi Channels Using Machine Learning Strategies
R Latha, Bommi R.M.
Abstract
Due to many difficulties with Wired Equivalent Privacy (WEP), the original encryption mechanism made available by the IEEE, Wi-Fi connections were susceptible to attacks. WEP’s shortcomings were resolved with the development of the 802. 11 standard, which was based on strong encryption methods and added client authentication (a feature absent from WEP). The IEEE 802. 11 standard’s encryption methods only encrypt data frames, making them susceptible to de-authentication DoS attacks. Attacks are conducted using the unencrypted management and control frames that are necessary for setup, maintenance, and data delivery, particularly de-authentication DoS attacks. Machine learning-based intrusion detection systems (IDS) are helpful for handling WLAN incursions in addition to hardware, software, and common upgrades that aid in attack prevention. The options for classifier algorithms based on probability, kernel, decision trees, or rules enable comparison of the effectiveness of the classifiers’ detection rates. When assessing the effectiveness of DoS, a review of the experimental results reveals that the suggested ML-based IDS is superior in attack identification with a 96% detection capability.