| Cognitive radio (CR) technology has recently been introduced to opportunistically exploit the spectrum. CR users are allowed to utilize licensed spectrum bands as long as they do not cause unacceptable interference with licensed users. Such flexibility alleviates the crowding issue in particular spectrum bands and greatly enhances the efficiency of spectrum utilization. However, such improvement relies on a robust and cost-effective design involving identification and reuse of spectrum opportunities changing over time, frequency, and space. In this dissertation, we focus on efficient spectrum sensing and utilization techniques for high-performance opportunistic spectrum access (OSA). We emphasize the importance of exploring probabilistic information unique for CR and develop novel techniques playing critical roles in CR systems.;One enabling technique for CR is spectrum sensing, which aims at monitoring the usage of licensed spectrum. We first propose probability-based periodic spectrum sensing by utilizing the statistical characteristics of licensed channel occupancy, which achieves nearly optimal performance with relatively low complexity. For the first time, the possibility that a licensed user appears in the middle of a sensing block is taken into account. Based on the statistical model of licensed channel occupancy, we then propose periodic spectrum sensing scheduling to determine the optimal inter-sensing duration. The statistical information of licensed channel activity is also used to vary the transmit power at each data sample to enhance throughput and reduce interference. As cooperative spectrum sensing can further improve the detection reliability with the introduced spatial diversity, we develop a probability-based scheme for combination of local sensing information collected from cooperative CR users, which enables combination of both synchronous and asynchronous sensing information. To satisfy the stringent bandwidth constraint for reporting, we also propose to simultaneously send local sensing data to a combining node through the same narrowband channel. The optimal local processing functions at the CR users and final decision rule at the combining node are discussed when the reporting channel is Gaussian and experiences fading, respectively. Calculation of probabilistic information involved is given as well. In the proposed approach, the bandwidth required for reporting does not change with the number of cooperative users. With proper preprocessing at individual users, such a design still maintains reasonable detection performance.;After determining the availability of certain licensed spectrum bands, CR users must select appropriate transmission parameters to better utilize these bands and avoid possible interference, which includes spectrum shaping and resource allocation. We propose a low-complexity spectrum shaping scheme for orthogonal frequency-division multiplexing (OFDM) based CR systems by mapping antipodal symbol pairs onto adjacent subcarriers at the edges of the utilized subbands. Sidelobe suppression and system throughput can be well balanced by adjusting the coding rate of the corresponding spectral precoder while power control on different sets of subcarriers is introduced to further lower the sidelobes. We also propose a spectral precoding scheme for multiple OFDM-based CR users to enhance spectral compactness. By constructing individual precoders to render selected spectrum nulls, our scheme ensures user independence and provides sufficient out-of-band (OOB) radiation suppression without bit-error rate (BER) performance loss. We consider the selection of notched frequencies to further increase the bandwidth efficiency. The proposed schemes enable efficient spectrum sharing between CR and licensed users and exhibit the advantages of both simplicity and flexibility. Then we provide a new resource allocation approach for OSA based on the probabilities of licensed channel availability obtained from spectrum sensing. Different from conventional approaches, the probabilistic approach exploits the flexibility of OSA to ensure efficient spectrum usage and protect licensed users from unacceptable interference at the same time. It also supports diverse quality-of-service (QoS) requirements in multi-user networks and can be implemented in a distributed manner. |
