Research On Wideband Spectrum Sensing Of Cognitive Radios

In recent years, the rapid development of various radio communication services makesspectrum resources increasingly nervous. Cognitive radio use dynamic spectrum accesstechnology. This technology initiatively find and use the unoccupied licensed spectrumresources, it solves the problems of current fixed spectrum allocation system caused spectrumresources underutilization.Spectrum sensing is the key technology of the cognitive radio, it reflects the essence ofcognitive problems. Because most of the cognitive radio work over a wide frequency range,the traditional base band signal detection technology is limited to apply in broadbandspectrum perception. We need to study from the perception of real-time and reliabilityrequirements. Efficient broadband spectrum sensing ability is a prerequisite for cognitiveradio applications, so this paper depth study on this issue.Firstly, this paper introduces the basic concepts of cognitive radio and its key technology,and summarizes the research status of spectrum sensing at home and abroad. Then introducesthe main detection methods of spectrum sensing: receiver detection, transmitter detection andcollaborative spectrum sensing, and focus on the cyclic spectrum detection method. Thecyclic spectrum detection of anti-noise and interference ability, has more signal informationthan traditional power spectral density function, and can effectively distinguish betweensignal and noise. Cyclic spectrum detection is verified to have anti-smooth-noise performanceby matlab simulation analysis.The sampling rete is higher in the application of broadband signal detection usingtraditional methods of detecting signal. And compressed sensing technology can effectivelyreduce the signal sampling rate, alleviate the pressure of A/D device and DSP processor. Thekey technology of the compression sensing is to reconstruct the signal rapidly and accurately.This paper proposes a compressed sensing reconstruction algorithm using smoothed l0basedon hybrid optimization algorithm. By introducing a hybrid optimization algorithm ofnumerical optimization theory, that is combined steepest descent method and modifiedNewton method, it overcomes " notched effect " steepest descent method in SL0algorithm.The algorithm uses the steepest descent method for initial point less demanding and initialstep characteristics. The algorithm provides a good initial point for modified Newton method, then transfer modified Newton method iteration. The algorithm has high accuracy and shorttime in reconstruction. The experimental results that consist with the theoretical analysis,show that the algorithm has good performance.Taking into account the spectrum the underutilized in reality, spectrum occupancy issparser in the dimension of time, frequency and so on. And spectrum occupied render slowlyvarying characteristics, so this paper presents a broadband compressed spectrum sensingmethod based on differential cyclic spectrum. This method makes use of the wideband signalon the cyclic spectrum sparsity, then directly recoveres detection signal cyclic spectrum fromthe compressed sampled signal, finally completes spectrum occupancy state estimation bydetecting the differential signal instead of the detection of the original sampling signal. Thismethod reduces the sparsity of the signal, and achieves a low sampling rate in widebandspectrum sensing. The simulation results show that the proposed method in low SNRenvironment has higher detection probability. Spectrum peak detecting method based onkurtosis is stable. and has better robustness in time-varying noise environment.