Research On Sensing Parameters And Sensing Scheme Design In Cognitive Networks

Cognitive radio can effectively solve the problems of limited spectrum resources and low efficiency of spectrum.As one of the key technologies,spectrum sensing play very important role.Secondary user periodically sense authorized spectrum with using spectrum sensing technology and dynamically access and using authorized spectrum on the premise of not interfering the primary user.Thesis have a deep research around the perception of optimization of perception parameters in cognitive radio and designing of intelligent spectrum perception.Specific content is as follows.Firstly, thesis introduces the basic knowledge of spectrum sensing, including spectrum sensing parameters and several common algorithms of spectrum sensing. At the same time, make simulation of the energy detection algorithm and several commonly collaborative spectrum sensing algorithms,reflecting the advantages and disadvantages of various methods through simulation curves.Secondly, the influence of the sensing parameters on the performance of cognitive radio system is studied. Under the condition of single-user spectrum sensing, the impact of sensing slot on the detection probability and frequency bandwidth utilization is analyzed. And proposed a compromise detection algorithm based on detection probability and frequency bandwidth utilization. In the case of cooperative spectrum sensing, the influence of sensing slots and the number of collaborative users on throughput is analyzed.Then, a optimization algorithm of perception slots length and number of users based on maximize throughput capacity is proposed.The simulation results show that the algorithm can get a set of optimal time slot length and collaboration users perception to make the largest system throughput.Finally,an intelligent cooperative spectrum sensing mechanism is proposed based on support vector machine(SVM) and fuzzy integral, in which cooperative spectrum sensing model is converted into a multi-classifier fusion model based on fuzzy integral. Every secondary user is considered as a separate SVM classifier. Sampling data obtained during sensing periods will be used as the input of SVM classifier. And the probability output of SVM classifier will be sent to the fusion center, where fuzzy integral algorithm is carried to get the final sensing results. In a word, the ability of learning in spectrum sensing stage and the ability of intelligently judging in information fusion stage are fully reflected in this mechanism. And simulation results further show that the proposed mechanism has higher detection probability and lower false alarm probability than a single classifier model.