| Cognitive Radio (CR) is a promising approach to efficiently utilize the scarce RF spectrum resource. In this paradigm, knowledge about spectrum occupancy in time, frequency, and space that is both accurate and timely is crucial in allowing CR networks to opportunistically use the spectrum, and avoid interference to a licensed primary user (PU). In particular, information about PU location will enable several key capabilities in CR networks including improved spatio-temporal sensing, intelligent location-aware power control and routing, as well as aiding spectrum policy enforcement. The PU localization problem in CR networks is in general different from localization in other applications such as Wireless Sensor Networks (WSN) and Global Positioning System (GPS), due to the following two features. First, a PU does not cooperate or communicate with CRs since they are opportunistic users of the PU spectrum band. Therefore, very limited knowledge about PU signaling, such as transmit power or modulation scheme, is available to CRs. As a result, passive localization techniques should be applied. Second, since CRs need to detect and localize PUs in the whole coverage area at a very low signal-to-noise ratio (SNR), in order to avoid interference to the primary network, the required number of CRs is relatively large and cooperation among CRs is necessary.;In this work, we aim to answer the following research question: How to efficiently acquire and utilize PU location information in CR networks? We first provide theoretical analysis of promising localization algorithms to understand the impact of various system parameters on the localization accuracy. We also develop distributed implementations of such algorithms to improve their robustness and reduce transmit power consumption. We then proceed to derive the Cramer-Rao lower bounds on the achievable localization accuracy, in order to provide system design guidelines and reveal potential improvements for algorithm development. We also consider algorithm design and performance analysis for advanced localization scenarios, such as localization of multiple overlapping PUs and localization using cost-efficient sectorized antennas. For localization of multiple overlapping PUs in time and frequency, we propose an adaptive algorithm that exploits statistical information at each CR, which obtains better accuracy and lower complexity compared with prior art. We introduce the usage of sectorized antennas to solve the PU localization problem, and provide localization solutions with theoretical performance analysis. We make some primary contributions on utilization of the PU location information in CR networks, by proposing a CR power control algorithm that considers uncertainties in both wireless channel and location estimation. The proposed algorithm greatly reduces interference probability to the PU compared with previous work. |
