Spectrum and transmission range aware clustering for cognitive radio ad hoc networks

Cognitive radio network (CRN) is a promising technology to overcome the problem of spectrum shortage by enabling the unlicensed users to access the underutilization spectrum bands in an opportunistic manner. On the other hand, the hardness of establishing a fixed infrastructure in specific situations such as disaster recovery, and battlefield communication imposes the network to have an ad hoc structure. Thus, the emerging of Cognitive Radio Ad Hoc Network (CRAHN) has accordingly become imperative. However, the practical implementation of CRAHN faced many challenges such as control channel establishment and the scalability problems. Clustering that divides the network into virtual groups is a reliable solution to handle these issues. However, previous clustering methods for CRAHNs seem to be impractical due to issues regarding the high number of constructed clusters and unfair load distribution among the clusters. Additionally, the homogeneous channel model was considered in the previous work despite channel heterogeneity is the CRN features. This thesis addressed these issues by proposing two clustering schemes, where the heterogeneous channel is considered in the clustering process. First, a distributed clustering algorithm called Spectrum and Transmission Range Aware Clustering (STRAC) which exploits the heterogeneous channel concept is proposed. Here, a novel cluster head selection function is formulated. An analytical model is derived to validate the STRAC outcomes. Second, in order to improve the bandwidth utilization, a Load Balanced Spectrum and Transmission Range Aware Clustering (LB-STRAC) is proposed. This algorithm jointly considers the channel heterogeneity and load balancing concepts. Simulation results show that on average, STRAC reduces the number of constructed clusters up to 51% compared to conventional clustering technique, Spectrum Opportunity based Clustering (SOC). In addition, STRAC significantly reduces the one-member cluster ratio and re-affiliation ratio in comparison to non-heterogeneity channel consideration schemes. LB-STRAC further improved the clustering performance by outperforming STRAC in terms of uniformity and equality of the traffic load distribution among all clusters with fair spectrum allocation. Moreover, LB-STRAC has been shown to be very effective in improving the bandwidth utilization. For equal traffic load scenario, LB-STRAC on average improves the bandwidth utilization by 24.3% compared to STRAC. Additionally, for varied traffic load scenario, LB-STRAC improves the bandwidth utilization by 31.9% and 25.4% on average compared with STRAC for non-uniform slot allocation and for uniform slot allocation respectively. Thus, LB-STRAC is highly recommended for multi-source scenarios such as continuous monitoring applications or situation awareness applications