Cognitive Radio Spectrum Sensing Based On Overlap FFT And Blind Channel Estimation Techonlogy

The key issues of spectrum sensing and blind channel estimation technology in cognitive radio are studied in this thesis,including the following two parts: frequency-domain energy detection technique based on Overlap FFT(Fast Fourier Transform)structure and blind channel estimation algorithms based on higher-order statistics.Based on the traditional frequency domain energy detection methods of cognitive radio signals,a frequency domain energy detection architecture based on Overlap FFT frames is introduced.Firstly,the detection probability and false alarm probability of the frequency domain energy detection algorithm based on Overlap FFT architecture are derived in theory,and the relational expressions between the detection probability and the false alarm probability and the number of Overlap FFT frames,the FFT frame overlap ratio,the signal-to-noise ratio(SNR)and the detection threshold are given based on the complex sinusoidal signal with additive white Gaussian noise.Then,a frequency domain energy detector based on Overlap FFT architecture is built on the Matlab platform,and the complex sinusoidal signal is used for Monte Carlo simulation to verify the accuracy of the theoretical analysis of the detection performance of this architecture.And the simulation results show that the detection performance of the architecture changes with the received data length,SNR,and FFT frame overlap ratio.In addition,by using OFDM signals to simulate the detection performance under different FFT frame overlap ratios,and comparing with the commonly used FSM(Frequency Smooth Method)detection algorithm based on the cyclostationary characteristic,the practical meaning and superiority of the frequency domain energy detection algorithm based on Overlap FFT architecture is verified.In the research of cognitive radio blind channel estimation technology,the blind channel estimation algorithms based on higher order statistics in the time domain are compared with each other,and Monte Carlo simulations verify the effectiveness of these algorithms in cognitive radio blind channel estimation problems.However,the inverse filter algorithm of source iteration is not ideal under low SNR,while the fitting error method based on cumulant matching and the equation error method based on subspace decomposition require good channel parameter initialization.In view of the shortcomings of blind channel estimation in time domain,a blind channel estimation technique based on high-order spectral parallel factorization is studied.First of all,the MPD/SPD(Multiple/Single Parallel Factor Decomposition)algorithm based on higher-order spectral tensors are described in detail,and Monte Carlo simulation was used to study the relationship between the estimation performance of MPD/SPD algorithm and the length of observation data,SNR and channel order estimation.Then,based on the inherent ambiguity problem in MPD/SPD algorithm,an EMPD/ESPD algorithm based on MIMO inverse filter ambiguity compensation is proposed,and the effectiveness of the improved algorithm was verified through simulation experiments.Finally,the superiority of the EMPD/ESPD algorithm is verified by comparing with the time domain blind channel estimation algorithm through simulation results.