Spectrum Sensing In Cognitive Radios Based On Sub-Sampling

With the development of wireless communication technology and the widely application of wireless communication services, the spectral efficiency problem in low spectrum has become increasingly prominent and fixed allocation way of spectrum resources can’t meet the springing up applications’ demands. To solve this problem, cognitive radio technology has been proposed. Cognitive radio is designed to achieve the dynamic allocation of spectrum resources to improve spectral efficiency, and in order to achieve this goal, the first problem is the perception of spectrum. The purpose of the spectrum sensing is to find out the spectrum holes from the wireless spectrum and use these spectrum holes to improve the spectral efficiency.The existing spectrum sensing algorithms require the Nyquist frequency on signal’s sampling, and thus raise higher requirements on Analog-to-digital Converter (ADC) and higher computing complexity as a result of increasing sampled data. Aiming at this problem, this paper introduces sub-sampling technology to realize the spectrum sensing.This paper sets up spectrum sensing’s system model with the sub-sampling technology, and proposes a sub-sampling technology based factor graph model spectrum sensing algorithm with consideration of spectrum aliasing’s characteristics. The algorithm uses the factor graph model to describe the relationships between original channels and aliasing channels after sub-sampling, and uses probabilistic deduction algorithm to estimate the original channel statement from the aliasing channel statement after sub-sampling to realize the spectrum sensing. Furthermore, this paper analyzes key problems in the proposed algorithm like the choice of sampling rate, the choice of factor graph model and the choice of judging algorithm, and sets up simulation platform to verify its performance. The simulation results prove that the proposed algorithm has better performance compared with the traditional algorithms like Energy Detection in sub-sampling situation.Aiming at that the proposed algorithm’s performance is not very good in low SNRs, this paper proposes an improved algorithm based on dual-threshold energy detection. Simulation results show that the improved algorithm has a higher probability of detection in low SNRs compared to that of the original algorithm and is more efficient.