Research On Cooperative Spectrum Sensing Of Cognitive Radios

With the fast development of wireless communication, lots of wireless applications come forth while the frequency resources are in shortage. From some surveys, it can be known that the utilization of licensed frequencies is not very frequent. Therefore, the Cognitive Radio (CR) is proposed to solve above problems. CR can detect licensed user quickly and accurately, dynamic access to the available frequency and retreat from it once the licensed user comes back, so one of key technologies in CR is spectrum sensing. Cooperative spectrum sensing (CSS) is greater used for its better performance and anti-fading capability compared to the traditional local methods in the actual communication environment. Consequently, research on cooperative spectrum sensing is done in this dissertation.Background and key technologies about CR are first introduced, and also algorithms of local detection are discussed. Due to the limitation of local detection, CSS is studied to improve the detection performance and fusion rules such as hard fusion, soft fusion are summed up. And then, an algorithm based on soft and hard fusion called double-threshold detection is studied in detail. Furthermore, cognitive users are divided into several clusters to detect licensed frequencies, is called cluster-based CSS. Cluster-based CSS scheme has better performance especially under fading channels compared to general CSS schemes. An improved cluster-based algorithm based on double-threshold is then proposed. The improved scheme can greatly improve system performance at the cost of high complexity and also simulation is done to compare above schemes. From simulation it can be seen that the proposed scheme can achieve better ROC curve compared to other schemes.In order to implement the CSS theory into reality, a CSS simulation method based on hardware platform is put forward, which involves the way to join a cluster, the way to quit, the frame protocol used, the timeslot division, etc. In the end, simulation results based on this cognitive platform are given to prove this scheme can implement CSS very well with high utilization and can access the unused frequency without any interruption to the licensed user.