Detecting greedy behavior by linear regression and watchdog in vehicular ad hoc networks

Abstract

Vehicular Ad hoc Networks (VANETs) aim to provide a high safety level to road users and enhance driving conditions for drivers. Thus, they are exposed to several kinds of attacks such as Denial Of Service attacks (DOS), especially the greedy behavior which affects services availability of authentic users. Greedy behavior attack has been well studied for Wireless local area networks (WLAN) and for Mobile Ad hoc Networks (MANETs). However, this attack is less studied for the case of VANETs. Indeed, the detection of a greedy behavior in quasi-static network is much easier than that in a network with low or high mobility. The aim of this paper is to detect greedy behavior in VANETs. More specifically, we propose a new detection algorithm based on a statistical method, linear regression and watchdog software. This algorithm distinguishes the existence or not of a greedy behavior and suspects the compromised nodes using a new appropriate defined metrics. The proposed algorithm is passive, it can be executed by any node in the network and does not require any modification of the IEEE 802.11p protocol.