Research On Wideband Spectrum Sensing Of Cognitive Radios

Cognitive radio technology uses a more flexible way of dynamic spectrum access，so it greatlyimproves the spectrum utilization and solution the spectrum scarcity problem which is caused by afixed spectrum allocation policy. Spectrum sensing is the key technology in cognitive radio field. Itsmain function is to detect the underutilized spectrums named "white space" which has not beenoccupied by the primary users. A secondary user needs to sense its operational electromagneticenvironment and adjust its operating parameters rapidly to use the spectrum resources which areimmediately available. Wideband spectrum sensing can detect multiple channels, which greatlyimproves spectrum sensing capability compared to narrowband spectrum sensing. Therefore,wideband spectrum sensing made cognitive radio use spectrum resource more flexible.The wideband spectrum sensing of cognitive is mainly studied in this thesis.Firstly, the spectrum sensing technology research background and significance are introduced.The narrowband spectrum sensing technology and the wideband spectrum sensing technology arereviewed. The existing wideband spectrum sensing methods have been classification and analysis.Secondly, the wideband spectrum sensing based on OMP algorithm is presented. Becausewideband spectrum sensing detects signals in several GHz bandwidth, excessively high samplingfrequency and large amount of data are require, which are the major challenges for existinghardware devices. By utilizing the sparsity of wideband spectrum, this paper proposes a newspectrum sensing method based on OMP algorithm for wideband spectrum sensing. In the proposedmethod, MWC sampling is used to implement compress sampling for the wideband analog signaldirectly. The compression sample model with finite dimension is obtained by using the symmetrydecomposition property of autocorrelation matrix and the independence of the primary user’s signal.Besides, AIC/MDL criteria is used to estimate the sparsity, which is a threshold of the stop iterationfor the OMP algorithm. As a result, the complexity of the algorithm is reduced greatly. Theestimation of the signal’s PSD is skipped in our method. The occupied channels are detecteddirectly from the compress sampled data in time domain at low rates. Simulation results show thatwhen the in-band SNR is above 9dB, the spectrum detection probability is greater than 90%.Thirdly, the wideband spectrum sensing based on MUSIC algorithm is presented. This methodwhich used multicoset sampling and MUSIC algorithm doesn’t need high sampling rate ADCs andcalculating the signal power spectrum. The performance of spectrum sensing is influenced by noisepower fluctuations slightly, but impacted by estimation accuracy of the number of signals greatly. Inaccordance with this problem, a search matching algorithm with dynamic threshold to estimate the number of signals is proposed. Simulation results show that the estimation accuracy of signalsnumber of the proposed algorithm is better than that of a method based on AIC/MDL criteria. Andthen the detection performances of the wideband spectrum sensing methods based on MUSICalgorithm which utilizes the search matching algorithm and AIC/MDL criteria, respectively, havebeen given. The former is superior to the latter under low SNR.At last, general wideband sensing is discussed. Within a certain wideband range, the type ofthe primary user is limited and fixed. According to this feature, two wideband spectrum sensingmethods in dynamic allocation mode - rectangular spectrum sensing and spectra correlation sensingare proposed, which use the prior information of the primary users’signal power spectral density(PSD). These two methods can work without any bandwidth division, and the primary user couldappear in the testing bandwidth range randomly. Simulation results show that the detectionperformance of spectra correlation sensing is better than rectangular spectrum sensing and theformer method can get a more accurate carrier frequency estimation for it uses more priorinformation.