Multidimensional Scaling based Localization for Cognitive Radio Networks

Abstract

The rapid growth in wireless networks demand on the deployment of new wireless services in both licensed and unlicensed spectrum. However, recent studies show that the xed assignment of spectrum leads to under utilization of the electromagnetic spectrum. Cognitive radio networks (CRNs), has emerged as a promising technology to mitigate both under utilization and spectrum scarcity problems. In CRNs, location information of primary users (PUs) and secondary users (SUs) is valuable in order to create a spectrally ecient CRN that can opportunistically take advantage of the spectrum with no or little interference to the PUs. Since PUs are not cooperative with SUs in nature, localization of all the users, including PUs, for the whole CRN is a challenging task. Number of algorithms are proposed to estimate the coordinate of PUs in CRNs, however, the attained accuracy in real-world applications is still far from the theoretical lower bound, Cramer-Rao lower bound (CRLB). Therefore, the main objective of this dissertation is to develop ecient and accurate localization schemes for the joint estimation of PUs and SUs in CRNs. A hybrid connectivity and estimated distance (HCED) based model is developed to jointly localize PUs and SUs in CRN, in which the available distances are considered between SUs, whereas connectivity information is considered for the links between PUs and SUs, because estimating the distances between PUs and SUs using conventional ranging techniques are generally not possible. By using HCED model, a robust multidimensional and Procrustes analysis based approach for the localization of PUs and SUs in CRN is also proposed. A novel accurate energy ecient localization algorithm for CRN is proposed. The key idea underlying this proposed algorithm is that it provides the optimal transmission range of the SUs to minimize the power consumption of the CRN. Furthermore, the optimal positions of the SUs with known location are determined to improve the nal localization accuracy for the whole CRN. A novel cluster based multidimensional scaling algorithm for CRN localization (CBMSCL) is also developed to improve the localization accuracy in irregular CRNs.