Research On Spectrum Detection In Cognitive Radio

With the high-speed development of mobile communication and the increase of wireless device in quantity, radio spectrum resource as a limited resource has become more and more scarce. On the other hand, the average utilization rate of spectrum resources is very low and extremely unbalanced. Cognitive radio is a proposal to solve the problem, and its core idea is to detect the radio spectrum resource which is not occupied, then the secondary user can use it. It realizes same radio spectrum using by multi-users. So spectrum detection is the key technique for realization of cognitive radio.Spectrum detection technology in cognitive radio is discussed in this thesis. Firstly, common spectrum detection algorithms such as energy detection are analyzed. Then the thesis focuses on cyclostationary detection and compared it with energy detection. It is shown by MATLAB simulation that cyclostationary detection has better performance. Then the thesis discusses the choice of detection criterion and simulates the performance of cyclostationary detection which uses spectral correlation function peak-average ratio as detection criterion. It is shown that cyclostationary detection using Spectral Correlation Function peak-average ratio has better performance.In order to solve the poor performance of cyclostationary detection in time varying channel, an AMCCMF(Adaptive Multi-Cyclic Correlation Matched Filter detection) algorithm which based on genetic algorithm is studied in the thesis. This algorithm reduces the affect of time varying noise by setting weighted coefficient. This thesis analyzes the principle of AMCCMF in detail, then analyzes the performance of this algorithm by MATLAB simulation. Simulation results show that in time varying channel, when false alarm probability and SNR are equal, the detection probability of AMCCMF is higher than original cyclostationary detection by 0.1 to 0.3.