Spectrum Sensing And Resource Allocation In Cognitive Radio Networks

With the rapid growth of the demand for wireless communication, thecontradiction between the scarcity of spectrum resources and the low actual spectralutilization by primary users (PU) is becoming a bottleneck restricting the developmentof the future mobile communications. Cognitive Radio (CR), with the ability ofproactively detecting wireless communication and dynamically accessing unusedspectral bands, is considered as a promising technique to solve the contradictionbetween low spectral utilization and the increasing spectral demand. However, thedynamic and complex spectrum resource makes the design and implementation of theCR system challenging.The dissertation has carried out analyses and investigations for the CR dynamicspectrum sharing system, including spectrum sensing, dynamic resource allocation andpower control. The main innovation results are as follows:(1) Firstly, the OFDM based CR broadband spectrum sensing is studied. A hybridspectrum sensing method combining with the correlation detection based on cyclicprefix and MUSIC algorithm is proposed. The rapidity of the correlation sensingalgorithm and the effectiveness of the MUSIC algorithm for the positioning of spectrumhole are combined together by the proposed method, which can shorten the time ofspectrum sensing with a certain sensing performance guaranteed. Results of thesimulation experiments show that the hybrid spectrum sensing method can effectivelylower the sensing cost and improves the utilization of idle spectrum by cognitive users(CU) compared with the traditional energy detection method based on power-spectrumand the first-order derivative method, and the advantage of which is more prominent inmultipath fading channels.(2) The optimal spectrum sensing cycle of CR system in MAC layer is studied inthe dissertation. The system throughput and the probability of collision between PU andCU are directly influenced by the frequency of the spectrum sensing. A statistical modelfor the PU activity is given through an in-depth analysis into the spectral usage patternsof PU. Based on that, an adaptive spectrum sensing algorithm is proposed. By introducing a “control factor”, the proposed algorithm can adaptively adjust thefrequency of spectrum sensing，while maintaining the CR system stability, as well asreducing the probability of collision between PU and CU. Compared with the cycledetection algorithm, the proposed algorithm which has low implementation complexityfor practical applications, can effectively improve the spectrum utilization, reduces theprobability of system conflicts and detection cost.(3) Next, the dynamic spectrum allocation policy for Non-Contiguous OFDM(NC-OFDM) CR system is studied, and a method for multi-user’s sub-channels, power,and bit joint allocation is proposed in this dissertation. The proposed method canmaximize the system capacity and improve the system spectral efficiency by allocatingthe sub-channel and power following the principle of “proportional fairness” withoutthe PU being interfered. Results of the simulation studies show that the algorithm hasprotected the communication requirement of each CU compared with the maximizesystem capacity algorithm. Meanwhile, the system capacity can be increased comparedwith the method of directly setting the interference threshold as the power of the limitedsub-channel. The way of firstly allocating sub-channels and then allocating power isutilized to reduce computational complexity, which is conducive to the projectapplication of the method.(4) Last, the dissertation discusses the power control of CU in the distributed CRsystem. The game theory is adopted to achieve the power control of CU. TheNPGP_SINR algorithm and the NPGP_NUF algorithm are proposed with the existenceand uniqueness of the Nash equilibrium of each algorithm demonstrated. Simulationresults show that the performance of the two proposed algorithms have significantlyimproved performances compared with both the NPG and NPGP algorithms, whichmeans the two proposed algorithms have relatively lower power and higher utility. TheNPGP_NUF algorithm has the lowest complexity, and the time effectiveness of whichcan be obtained by sacrificing a small amount of user utility. Both the NPGP_SINR andNPGP_NUF algorithm are not subject to the signal modulation and access mode.Therefore, they can be applied to the power control in the CR system.