Misbehaviour detection mechanism for vehicular systems presented at IEEE MeditCom 2021
INSPIRE-5Gplus partner CTTC presented the latest INSPIRE-5Gplus results on anomaly detection in vehicle-to-everything (V2X) environments at IEEE International Mediterranean Conference on Communications and Networking (IEEE MeditCom) 2021, on 7th September. IEEE MeditCom 2021 is held as a hybrid in-person and virtual conference in Athens, Greece, from 7th to 10th September.
The INSPIRE-5Gplus paper with the title “Reinforcement Learning-based Misbehaviour Detection in V2X Scenarios” was presented online by Charalampos Kalalas from CTTC in Special Session 2, entitled “AI Assisted Technologies for Connected and Automated Mobility (CAM)”. Authors of the INSPIRE-5Gplus paper are Roshan Sedar (CTTC), Charalampos Kalalas (CTTC), Francisco Vázquez-Gallego (CTTC), and Jesus Alonso-Zarate (i2CAT).
In the paper, the authors assess the effectiveness of a reinforcement learning (RL) approach for misbehaviour detection in V2X scenarios using an open-source dataset. In particular, the focus is on the identification of sudden-stop attacks, which can be often difficult to detect due to the attacker’s erratic behaviour over time, and may lead to unnecessary traffic congestion and potential road accidents. By adopting a Markov decision process modelling framework, the authors evaluate the performance of RL-based detection over commonly used detection metrics. The INSPIRE-5Gplus research outcomes reveal that misbehaving vehicles can be accurately detected by sequentially analysing their mobility patterns, i.e., real-time position and speed, in the exchanged beacon messages.