Classification of Denial of Service Attacks on Wi-Fi-based Unmanned Aerial Vehicle
Gustavo de Carvalho Bertoli, Loureneo Alves Pereira, Osamu Saotome
Abstract
This paper presents an analysis of denial of service (DoS) attacks on Wi-Fi-based Unmanned Aerial Vehicle (UAV). The platform is a Parrot AR.Drone 2 and uses the IEEE 802.11 protocol for command and control. The threat scenarios are the TCP and UDP Flood Attacks and the de-authentication attack. The de-authentication is a functionality available on IEEE 802.11 Wireless protocol that is misused for DoS attacks. The approach for DoS classification is based on logistic regression and decision tree (DT) using a dataset composed of malicious and normal network traffic captured during UAV flights. The DT model obtained in this paper accomplishes an F1-score to classify DoS attacks (de-authentication, UDP, and TCP flood) of 0.97.