An Improved Spectrum Aware Cluster-Based Architecture For Cognitive Radio Ad-Hoc Networks Nafees Mansoor 1 , *A.K.M. Muzahidul Islam 2 , Mahdi Zareei 3 , Sabariah Baharun 4 , and Shozo Komaki 5 1,2,3,4,5 Malaysia-Japan International Institute of Technology (MJIIT), Universiti Teknologi Malaysia (UTM). E-mail: 1 nafees@nafees.info, 2 akmmislam@ic.utm.my, 3 m.zareei@ieee.org, 4 drsabariah@ic.utm.my, 5 komaki@ic.utm.my *Corresponding Author: 2 akmmislam@ic.utm.my Abstract—To encounter the mounting necessity of radio spectrum, proper utilization of the radio spectrum is must. Cognitive radio practices an open spectrum allocation technique to make efficient utilization of the wireless radio spectrum and reduces the bottleneck on the frequency bands. A robust architecture with suitable communication protocol is a precondition in the deployment of cognitive radio networks. A novel cluster-based architecture for an ad-hoc cognitive radio network is proposed in this paper, where the spatial variations of spectrum opportunities are considered for clustering. Our cluster formation is defined as a maximum edge biclique problem. Each cluster consists with a set of free common channels, which benefits smooth shift between control channels. A set of cognitive radio nodes are grouped into the same cluster if they sense similar free channels and are within the communication range of the leader node called cluster head. The selection of cluster head is based on a parameter called Cluster Head Determination Factor (CHDF). Considering the re-clustering issue for mobile nodes, we also introduce the concept of secondary cluster-head. The secondary cluster-head takes charge of a cluster whenever a cluster-head moves out from the cluster. Proposed clusters adapt themselves dynamically with respect to spectrum availability, and the high mobility of the nodes. Finally, we simulate the proposed architecture to evaluate the performance of our method. Keywords— Cognitive radio networks; ad-hoc networks; cluster-based network; control channel; re-clustering I. INTRODUCTION The technological developments and usages of wireless technology are growing speedily. These bring an ever- mounting demand for radio spectrums. Radio spectrum, a limited natural resource, has been distributed almost fully, which leads to a spectrum scarcity problem for the upcoming wireless technologies and applications. Along with spectrum scarcity problem, the existing radio spectrum is also underutilized. Several surveys on spectrum utilizations show that radio spectrum is underutilized with variance of frequency, time and space [1, 2]. The traditional radio spectrum scheme allocates radio frequencies in the command-and-control model, which does not allow any unlicensed user to use the licensed band while it is free. The main idea of cognitive radio is to use the underutilized or unused radio spectrum in an opportunistic manner. J. Mitola III pioneers cognitive radio [3], which is an intelligent wireless communication system that has the capability to orient itself to the situation and makes corresponding changes in operating parameters (transmit-power, carrier frequency, and modulation strategy) in real time. Primary user (PU) and Secondary User (SU) are two types of users in CRN, where PU is the licensed user for the spectrum and SU uses spectrum opportunistically [4]. The decentralized form of any wireless network is considered as wireless ad-hoc network, which is a self- configuring network [5]. The topology is ad-hoc as it does not rely on preexisting infrastructure, such as routers in wired network. Nodes in an ad-hoc network communicate with each other directly or using intermediate nodes. The decentralized nature of wireless ad-hoc networks enables scalability compared to wireless managed network. Minimal configuration and quick deployment make ad-hoc networks suitable for emergency situations like natural disasters or military conflicts [6]. Mobile Ad-hoc Networks (MANETs) and Wireless Sensor Networks (WSN) are the most common types of ad-hoc networks. These technologies use fixed spectrums, where this static nature prevents users from dynamically reuse unused spectrum for the better utilization of the radio bands. Cognitive Radio Network (CRN) has received a keen interest to the researchers of communication networks in the last few years because of the flexible and dynamic behavior of spectrum usage over other ad-hoc technologies. Choice of transmission spectrum, topology control, and distinguishing mobility from PU activity are the key factors that make CRN unique. Clustering is one of the most widely investigated solutions for scaling down ad-hoc networks. Cluster formation includes arranging network nodes into logical groups with the goal of cutting the signaling overhead required for network operation while upholding the network connectivity. The specific objective of the grouping process generally depends on network characteristics and application requirements. For example, in a dynamic environment, cluster formation seeks to abstract the network topology into a simpler; more stable form so that local changes, e.g., due to the appearance of a primary spectrum user or to node mobility, do not trigger the need for network-wide updates. Another common objective, which improves the efficiency of network functions such as routing and multicasting, is to form clusters such that the number of nodes in the network backbone is kept small. Cognitive Radio Network (CRN) has received a keen interest to the researchers of communication networks in the last few years because of the flexible and dynamic behavior of