| Spectrum detection is the basic support technology of cognitive radio(CR),which is the foundation and premise of spectrum allocation and spectrum sharing.However,the current spectrum detection technology is still difficult to support the future application needs of the information society under the interconnection of all things,and there are some urgent problems to be solved: First,the spectrum detection algorithms which are restricted by some factors,such as the signal type of primary user(PU)or the receiver of PU,are difficult to meet the actual application requirements;Second,in the complex and harsh communication environment such as low signal-to-noise ratio(SNR),noise uncertainty(NU),the detection of PU signal becomes more difficult and complex;Third,the development needs of future communication network,such as high energy efficiency,low power consumption and low latency,put forward higher requirements on the complexity,detection time,and detection cost of spectrum detection algorithms.In order to accelerate the implementation of spectrum sharing applications and support the construction of future communication networks,this thesis has conducted an in-depth study on the spectrum detection technology in CR from two aspects of detection algorithm and cooperative strategies.The main work and contents are as follows:Firstly,aiming at the problem that the local energy detection algorithm is extremely sensitive to noise,the energy detection algorithm based on generalized stochastic resonance has been studied,and its detection performance has been theoretically analyzed and simulated for verification.On this basis,in order to reduce the system overhead introduced by the stochastic resonance technology,a double threshold energy detection algorithm based on generalized stochastic resonance has been proposed.According to the change of channel environment,the algorithm can adaptively select a more economical and efficient detection algorithm,which has strong flexibility and adaptability.Simulation results show that the algorithm can effectively reduce the computation and system overhead without reducing the detection accuracy.Secondly,considering the limitations of single user spectrum detection,the cooperative spectrum detection algorithms based on energy detection have been investigated.In order to improve the detection performance under the large noise uncertainty,a double threshold cooperative energy detection algorithm based on generalized stochastic resonance has been proposed.The algorithm adopts double threshold mechanism and two-step cooperation strategy to adaptively select fusion rule based on detection statistics,so as to improve detection performance and avoid the increase of cooperation cost as much as possible.Simulation results show that the algorithm has strong robustness,and still has good detection performance under large noise uncertainty.Thirdly,considering the mobility of PU and secondary user(SU),and the power of the PU signal received by each SU may be different,a cooperative energy detection algorithm based on dynamic grouping has been proposed.The algorithm introduces grouping parameters to dynamically group all SU,and assigns different weighting factors to SU in different groups.The FC only needs to collect and fuse the local decision result of each SU.Simulations show that the algorithm can effectively improve the system detection performance without significantly increasing the cooperation overhead and processing delay. |
