Investigating Cross-Platform Robustness for Machine Learning Based IDSs on 802.11 Networks

Security and Intrusion detection in 802.11 networks is currently an active area of research where WiFi specific Data Link layer attacks are an area of focus. While these attacks are very simple in implementation, their effect on WiFi networks can be devastating. Recent research has focused on producing machine learning based IDSs for these attacks. Such IDSs have shown promise. Our work investigates the CrossPlatform robustness of such machine learning based solutions. By cross-platform robustness we mean the ability to train a solution on one network and run it seamlessly on another. We demonstrate that machine learning based IDSs could potentially suffer when employed across different platforms. In order to solve this, we propose a MAC address mapping technique which can achieve a Cross-Platform detection rate, for machine learning based IDSs, on average of 100% and a false positive rate on average of 0.1%.