Computer Networks 121 (2017) 152–172 Contents lists available at ScienceDirect Computer Networks journal homepage: www.elsevier.com/locate/comnet A game theory based trust model for Vehicular Ad hoc Networks (VANETs) Muhammad Mohsin Mehdi, Imran Raza * , Syed Asad Hussain Department of Computer Science, COMSATS Institute of Information Technology Lahore, Pakistan a r t i c l e i n f o Article history: Received 12 July 2016 Revised 22 March 2017 Accepted 7 April 2017 Available online 13 April 2017 Keywords: Game theory Nash equilibrium Trust VANETs a b s t r a c t Vehicular Ad hoc Networks (VANETs) facilitate road safety, transportation security, reliability and man- agement. This paper presents a game theory based trust model for VANETs. The proposed model is based on an attacker and defender security game to identify and counter the attacker/malicious nodes. The pa- rameters considered for attackers and defender’s strategy are majority opinion, betweenness centrality, and node density. The outcome of the specific game is determined by the game matrix which contains the cost (payoff) values for possible action-reaction combination. Nash equilibrium when applied to cal- culate the best strategy for attacker and defender vehicles. The model is simulated in Network Simulator (ns2), and results show that the proposed model performs better than the schemes with random mali- cious nodes and existing game theory based approach in terms of throughput, retransmission attempts and data drop rate for different attacker and defender scenarios. © 2017 Elsevier B.V. All rights reserved. 1. Introduction In a highway scenario, Vehicular Ad-Hoc Networks (VANETs) consisting of high speed nodes need to log connection de- tails providing road safety, transportation security, and reliability. The nodes communicate constantly and may share impersonated events such as false accidents, incorrect traffic conditions and false response from Road Side Units (RSUs). Therefore, the reliability of an event depends on the authenticity of a node and some reli- able tagged nodes such as emergency vehicles which are reliable to report an event without any discrepancy. The authenticity of the node is calculated using majority opinion and betweenness cen- trality explained in detail in their respective sections. It is chal- lenging to provide reliable communication in VANETs, especially in highway scenarios, as frequent addition and removal of nodes compromises the detection of malicious nodes. Therefore, VANETs need a robust approach to handle fast changing network topology to defend and react against an attack. The possibility of an attack is a checked exception and hence needs to be handled establishing trustworthy Intelligent Transportation System (ITS). The time crit- ical nature of highway scenarios requires a node to verify the ac- curacy of the received information in real time. Different trust and reputation models [1] have been presented to classify inaccurate * Corresponding author. E-mail addresses: mohsin.mehdi@ciitlahore.edu.pk (M.M. Mehdi), iraza@ciitlahore.edu.pk (I. Raza), asadhussain@ciitlahore.edu.pk (S.A. Hussain). messages and malicious vehicles. Trust establishment schemes are available for peer-to-peer, sensors, and Mobile Ad-hoc Networks (MANETs) [1,2]. However, trust establishment in VANETs is chal- lenging because of dynamic network topology, increase and vari- ance of malicious nodes, and absence of a third party for network monitoring. The only possible communication with the infrastruc- ture takes place with RSUs. Therefore, centralized systems are not suitable to establish trust in VANETs. Furthermore, high speed ve- hicles make it difficult to log quality of experience between them. The existing entity oriented trust models for VANETs are based on the verification of vehicles’ identities and their legitimacy in the network [3–5]. In identity-based system, the trust metric is linked to the vehicle credentials and does not consider its trustworthi- ness. Other trust models are based on a data-oriented approach, where the vehicles are responsible for the trustworthiness of the information that it generates. This paper presents a game theory based trust model for VANETs implementing a robust algorithm that calculates three pa- rameters; majority opinion, betweenness centrality and node den- sity. The proposed trust model verifies the information and mes- sages to identify trusted nodes for reliable communication. This enables a node to have better understanding of the network and its surroundings to acknowledge both destructive and informa- tive events. The proposed model modifies Ad-hoc On-demand Dis- tance Vector (AODV) routing protocol [4] to maintain a log for ev- ery successful route. The proposed model is simulated using Net- work Simulator (ns2) [5] implementing both attacker and defender with a target i.e. an attack is always detected and deceived. In http://dx.doi.org/10.1016/j.comnet.2017.04.024 1389-1286/© 2017 Elsevier B.V. All rights reserved.