JOURNAL OF INFORMATION SCIENCE AND ENGINEERING A stable clustering scheme based on adaptive multiple metric in vehicular ad-hoc networks Hamid Reza Arkian, Reza Ebrahimi Atani * , Atefe Pourkhalili, Saman Kamali Department of Computer Engineering, University of Guilan, Rasht, Iran Email: {arkian, pourkhalili, kamali}@msc.guilan.ac.ir, rebrahimi@guilan.ac.ir ABSTRACT Vehicular Ad-hoc Networks (VANETs) are the special class of Mobile Ad-hoc Networks (MANETs) with high mobility and frequent changes of topology. Clustering is applied in VANETs to divide the network into smaller groups of mobile vehicles and improve routing, information dissemination and data gathering. In this paper, we propose a 2-layer stable clustering scheme based on adaptive multiple metric combining both the features of static and dynamic clustering methods. The cluster head is selected among the cluster members based on a new multiple metric called suitability value. It is derived from both mobility metrics such as relative speed, position and time to leave the road segment and Quality of Service metrics including available bandwidth, neighborhood degree and RSU link quality. Due to the proposed adaptive metric, the higher cluster stability as well as QoS are achieved. The simulation results clarify effectiveness of our proposed method in a highway scenario and show that our technique has better results and provides more stable cluster structure compared with the other related methods. KEY WORDS: Vehicular Ad hoc Network; VANET; Dynamic Clustering; Quality of Service; Stability; 1. INTRODUCTION Vehicular ad-hoc network (VANET) is the new emerging technology which provides many applications on roads. However, the mobility of vehicles as wireless devices in VANETs causes frequent disconnection of the communication links and changing topology of the network [1]. Clustering is a technique to form grouping of nodes and can greatly improves network performance. It allows the formation of a virtual communication backbone that supports efficient data delivery in VANETs and also improves the consumption of scarce resources such as bandwidth [2]. The clustering technique has been well studied in Mobile Ad-hoc Networks (MANETs) in recent years. According to the characteristics of VANETs, such as high speed, frequently changes of topology, scale of the network, etc., the traditional clustering schemes are not suitable for VANETs [3]. Therefore, new clustering schemes should be designed specifically based on VANET characteristics. Generally, two different approaches for the clustering of vehicles are defined in VANET: First, static clustering based on Vehicle-to-Infrastructure (V2I) communication that Road Side Units (RSUs) play the role of static cluster heads. In order to have real-time communication and connectivity with the Internet, the vehicles are essentially connected with RSUs [4]. However, according to the large distance between the RSUs, the vehicles with high mobility are not always connected to the RSUs. Second, dynamic clustering based on Vehicle-to-Vehicle (V2V) communication that cluster heads are chosen from cluster members [2]. In fact, dynamic clustering method removes the concept of static cluster heads. The dynamic clusters are in motion on the road and vehicles either join or leave the clusters according to their speed and proximity to identified cluster heads [5]. In this case, since the V2V communications are more flexible and independent of * Correspondence to: E-mail: rebrahimi@guilan.ac.ir, reza.ebrahimi.atani@gmail.com