Dynamic Spectrum Sharing In Cognitive Radio Networks

Nowadays, the way of spectrum allocation is fixed and each spectrum band serves for one specified service exclusively. In this circumstance, each radio system works independently without the interference of other systems. However, a band may become crowded because of bursting heavy traffic while other bands remain idle due to light traffic. This contradiction would cause inefficient use of spectrum resources and reveals the disadvantage of fixed way of spectrum allocation. Therefore, an innovatively new way of spectrum usage is indispensable, and cognitive radio network technology is born to solve this problem.Based on Cognitive Radio (CR) technology, Cognitive Radio Network (CRN) has become one of the hottest research topics in communication area. The key characteristic of CRN lies in the Cognitive Cycle, which consists of sensing, studying, planning and decision making, correspondingly. All the steps in the cycle aim at raising end to end efficiency, spectrum usage and radio resources usage.In the before-mentioned cycle, planning and decision making is one of the key steps, which refers to choosing the optimizing behavior based on the study of cognitive information and experience, and adjusting the corresponding parameters through reconfiguration. This step contains spectrum management and radio resource management, aiming at maximizing the usage of radio resources and raising the performance of radio systems. Dynamic Spectrum Sharing (DSS) is one of the key research areas in this step, which is the main concern of the paper. DSS can change the fixed way of spectrum usage and raise the usage efficiency remarkably. To sum up, the work of the paper is listed as follows:First, the detailed concept, background and characteristics of CRN are introduced. Based on the introduction, the importance of DSS in CRN and the current research on it are described.Following that, the main innovative research work is depicted. The first one is the power control in horizontal sharing scenario, where all the users have the same priority. In order to optimize the transmission power automatically, the tool of Q-learning is borrowed, trying all the transmission power combination. We assume the transmission power of all the users to be discontinuous, so that the number of power combinations is finite. Through trying all the combinations, the transmission power can be optimized automatically under the control of Q-learning.The second research topic is the opportunistically spectrum usage in vertical sharing scenario, where there are two user classes:the primary user, who has spectrum license;the secondary user, who has no spectrum resources. Both the primary and secondary service follows Poisson distribution and the secondary user has to use the licensed spectrum band opportunistically. Therefore, spectrum sensing is indispensable when the secondary user needs to use the spectrum. However, spectrum sensing errors often occur, i.e. miss detection and false alarm. The former happens when the busy channel is detected as idle, which may interfere the QoS(Quality of Service) of primary user seriously. The latter occurs when the idle channel is detected as busy, which may result in some spectrum opportunities wastage. Therefore, it's necessary to design a mechanism, which can not only protect the QoS of primary user, but also make full usage of spectrum opportunities. Stimulated by this point, the paper develops an innovative spectrum access model based on continuous-time Markov chain and achieves the before-mentioned goal effectively.In the end of this thesis, a summarization is presented and the potential work in this research area is pointed out.