Higher-Order Statistics Based Sequential Spectrum Sensing for Cognitive Radio

• Main Authors: Hsing-yi Hsieh, 謝馨儀
• Other Authors: Meng-Lin Ku
• Format: Others
• Language: en_US
• Published: 2011
• Online Access: http://ndltd.ncl.edu.tw/handle/01326201445990314227

碩士 === 國立中央大學 === 通訊工程研究所 === 99 === In cognitive radio, spectrum sensing is a key enabling functionality to identify the vacant spectrum which is not occupied by primary systems. With good sensing capability, secondary users can effectively recycling the spectrum resource without disturbing active primary users. Energy detectors are commonly used and relatively simple spectrum sensing techniques. However, for low signal-to-noise ratio (SNR) regimes, the performance of energy detectors degrades dramatically as the signal and noise could be mixed together after the operation of energy calculation. In addition, the outputs of the energy detectors are often assumed as Gaussian distribution, which is not necessarily guaranteed in realistic cases. In this paper, a high-order statistics (HOS) based sequential test detector is investigated for sensing spectrum, particularly for low-SNR applications. We resort to high-order statistics, in terms of cumulant statistics, for overwhelming the Gaussian noise effect and improving the spectrum sensing reliability. Based on these cumulants, a binary hypothesis testing problem is formulated and a low-complexity sequential probability ratio test (SPRT) is developed for efficiently detecting underutilized spectrum. Our numerical results show that the proposed detector outperforms more than 10dB detection probability than the conventional energy detectors.