Research And Application Of Independent Component Analysis Based On Periodic Non-uniform Sampling In Wideband Spectrum Sensing

With the development of multi-function communication systems,the spectrum resources become increasingly scarce.Cognitive Radio(CR)as a wireless communication technology can dynamically access to the spectrum by continuously sensing the surrounding spectrum environment,and flexibly assign spectrum holes to secondary users for communication.CR not only saves the cost of opening up new spectrum resources,but also provides a solution to alleviate the pressure of lack of spectrum resources faced by the wireless communications.In the system of wideband spectrum sensing of CR,the sensing band is wide and the specific spectral position of the signal is unknown in advance.In this way,according to the Nyquist Sampling(NS)theorem,the sampling rate required to collect signal information should be at least twice the overseen frequency range if uniform sampling is used.It will greatly increases the difficulty of hardware implementation.Therefore,how to reduce the system sampling rate is an important problem that needs to be solved urgently in CR wideband spectrum sensing.Using Periodic Non-uniform Sampling(PNS),the original signal can be collected and reconstructed at a sampling rate well below the NS.In this paper,the research and application of Independent Component Analysis(ICA)based on PNS in wideband spectrum sensing is studied.The main work is as follows:(1)Explore the practical necessity and advantages of PNS applied to wideband spectrum sensing,to reduce the sampling rate and the difficulty of hardware implementation.Firstly,the basic theory of PNS is analyzed,and then the realization method of PNS and the selection principle of the optimal sampling sequence are introduced in detail.Further more,the unbiased Power Spectral Density(PSD)estimation algorithm is applied to calculate the PSD of PNS sequences.In this way we can sense the active state of the primary user signal in the channel and find the spectrum holes.(2)Combine a Blind Source Separation(BSS)algorithm based on ICA with PNS in order to determine the spectral position of the primary user signal with different transmission power.Firstly,the separation performance of different algorithms under PNS conditions is analyzed.The wideband spectrum sensing scheme based on ICA is simulated and the spectral position of each primary user signal is determined by the Minimum Energy Reconstruction(MER)of non-baseband signal.(3)Use the model of Rayleigh fading to explore the effect of ICA based on dual-channel PNS in improving signal separation,while considering the multi-path effect and Doppler shift may have an effect on sensing performance in actual communication environment.Further more,the validity of spectrum sensing in dual channel sampling is analyzed by simulation.