Researches On Full Duplex Cognitive Radio Spectrum Sensing Issues

With the rapid development of wireless communication technology and the wild use of mobile devices,the available spectrum resources are nearly saturated.As a result,spectrum resource shortage is becoming an urgent situation to be dealt with in the communication field.Full duplex cognitive radio combines the advantages of full duplex technology with its effective spectrum utilization and the ability of cognitive radio with its dynamically spectrum access strategy,which enable its maximum use of spectrum.The full duplex cognitive radio takes advantages in spectrum sensing mechanism over traditional cognitive radio,but shows poor sensing performance.The main reason is that,the self interference cancellation technology,known as the core technology of full duplex radios,is still not perfect.Which leads that the performance of sensing legitimate user's signals is limited.This thesis focuses on the spectrum sensing research of full duplex cognitive radio,especially on the how to promote the poor sensing performance which affected by signals from different users cannot be identified because of the non-time slotted sensing mechanism.A series of improvement scheme are proposed.The main work of this thesis is as follows:(1)By establishing a risk function,the differences of sensing mechanism measured from the aspects of data conflict and spectrum waste between full duplex cognitive radio and traditional cognitive radio are analyzed.It is proved that the non-time slotted sensing mechanism is prevailed,which also proves the necessity of the research.(2)Under the centralized network cooperative sensing scheme with the energy detection method,a weight optimized spectrum sensing method is proposed.In this method,weights are set to all users,and then the secondary users with higher sensing accuracy are selected by optimizing the weights.Two kinds of fusion strategy,that is,decision fusion and data fusion are discussed respectively.Convex optimization is used to solve the weights.When dealing with non-convex non-linear problems,the method of Sequential Cone Programming algorithm is proved efficient.Once the optimal weight is obtained,the spectrum sensing performance can be improved.(3)The global power spectrum sensing model is applied to full duplex cognitive radio to sensing the primary and secondary users' signals simultaneously in the same frequency band.Which provides the users' multidimensional information including geographical location,transmit power and other information.These can help to identify the users' identity.It effectively eliminates the influence of secondary users ' signals and provides a more comprehensive spectrum monitoring information for the whole communication network.In this thesis,a multidimensional sensing algorithm based on Laplace prior is proposed under single channel condition.Compared with the existing algorithms,this algorithm has the advantages of high accuracy and fast computation speed.(4)Under the global power spectrum sensing model,when the number of sensed channels increase,the data scale increasing rapidly.Which leads to the real-time sensing cannot be met by existing algorithms.Block sparse structure is added in the original model,and the fast block sparse Bayesian learning algorithm is applied to solving the mathematical model which includes the multidimensional information.Compared with the existing algorithms,the method proposed greatly improves the computation speed while the accuracy of results is guaranteed and the real-time sensing requirement is satisfied.