Research On Spectrum Sensing Based On Energy Detection In Stable Distribution Noise

With the rapid development of wireless communication technology,the limited frequency spectrum resources is becoming more and more valuable,the traditional fixed spectrum allocation method has been unable to meet the contradiction between the growing wireless communication services and the scarcity of spectrum resources.In order to solve this contradiction,cognitive radio(CR)has been proposed as a new technology.It accesses and shares the idle spectrum resources through an'opportunistic' way,which improves the spectrum utilization efficiency effectively.Reliable spectrum sensing is one of the key technologies for cognitive radio to realize it's function and researches on it have great practical value.There are several mature methods in signal detection,which can be used to spectrum sensing.But in practice applications,the communication channels are often influenced by non-Gaussian noise and most of the traditional detection techniques are based on the Gaussian noise model,so that they are no longer applicable.In this context,we propose three kinds of spectrum sensing algorithms based on energy detection(ED)in a-stable distributed noise by combining with the traditional energy detection in this thesis.The ?-stable distributed noise is the best model to describe the non-Gaussian noise in the communication channel.However,the energy detection become invalid in a-stable distributed noise because there is no limited second-order statistic for it.To solve this problem,firstly,we propose an energy detection algorithm based on nonlinear transformation in this thesis.This algorithm using energy detection after pre-processing transformation,which makes the received signal has two-order statistics.We define the preservation to suppression ratio to measure the nonlinear pre-processing function.The simulation results prove that the detection performance of the proposed algorithm is better than the detection algorithm based on fractional lower order moments and energy detection algorithm and it has a great detection performance when the preservation to suppression ratio of the nonlinear pre-processing function is high.Secondly,we propose a energy detection based on the kernel function theory,which mapping the received signal to the transformation domain and then using the energy on it to detect.This algorithm has a great detection performance both in Gaussian noise and a-stable distributed noise.Finally,we propose a detection algorithm based on fractional lower order cyclic spectrum density energy according to the cyclostationarity of the modulated signal.For the modulated signals,the detection performance of this algorithm is better than the detection algorithm based on second-order cyclic spectrum density energy and the detection algorithm based on fractional lower order moments in a-stable distributed noise.