Research On Cooperative Spectrum Sensing In Full-Duplex Cognitive Radio Networks

With the continuous growth of wireless communication services and the wide application of intelligent terminals,limited spectrum resources have become increasingly scarce.In addition,more spectrum resources need to be exploited for imperious demands which are increased by the emerging wireless communication services such as Internet of Things(IoT)communication,Device to Device(D2D)communication,Machine to Machine(M2M)communication and mobile internet communication.On the one hand,the industry focuses on millimeter waves and visible light frequency band to find a breakthrough.It is difficult to meet the coverage requirements of wireless communication,due to the high-frequency spectrum is limited by its own propagation characteristics.On the other hand,cognitive radio technology has made the improvement of spectrum utilization possible.Spectrum sensing is one of core functions in cognitive radio technology.It not only prevents unlicensed users(secondary users,SUs)from causing interference to licensed user(primary user,PU),but also detects available spectrum holes to improve the utilization of spectrum.However,its detection performance is often affected by problems such as multipath fading,shadow fading,and hidden terminals.In order to solve such unfavorable factors,cooperative spectrum sensing was proposed to improve detection performance by utilizing the diversity of secondary users' location.Based on the full-duplex cognitive radio networks,this thesis discusses and analyzes the key parameters and performance indicators in the model,and also investigates its cooperative spectrum sensing system and channel allocation strategy.The main research contents are as follows:(1)To reveal the advantages of the full-duplex cognitive radio networks model compared with the half-duplex mode,this thesis analyzes the key parameters and performance indicators,including: sensing error rate,transition probability of system state,collision ratio,design of sensing threshold,spectrum utilization and secondary throughput.Full-duplex communication technology realizes the same frequency of spectrum sensing and data transmission.On the basis of sharing the spectrum resources with the PU,the SUs increase the use of spectrum holes and further improve the utilization of spectrum resources.(2)To better adapt to the changing network environment in full-duplex cognitive radio networks,we propose and implement a cooperative spectrum sensing system based on ensemble learning and Synthetic Minority Oversampling Technique(SMOTE).The detection of spectrum holes is described as a binary-class problem in this thesis,and the typical AdaBoost algorithm of ensemble learning is adopted for spectrum sensing.This system can better adapt to the network environment by means of the excellent generalization and robustness of the machine learning algorithm.In addition,the SMOTE as a oversampling technology is used for solving the data imbalance problem during model training.(3)In order to reasonably arrange the SUs to participate in the cooperative spectrum sensing of the designated channel in the multi-channel full-duplex cognitive radio networks.We propose a channel assignment strategy based on adaptive genetic algorithm.An optimization model aiming at maximizing the secondary networks throughput is established under the premise of satisfying the normal communication of the PU.The adaptive genetic algorithm is adopted to solve the problem for approaching the optimal solution and avoiding early convergence.