Broadband Spectrum Detection Technology Research

With the development of communication, the spectrum hasbecome more and more crowded. The conventional static spectrumallocation policy has been greatly hampered the utilization of spectrum.In this situation, the emergence of cognitive radio is considered to bethe best solution to the low utilization of spectrum.Spectrum sensing is the most critical technology in cognitive radio.To improve the effectiveness of eigenvalue-based detection, thecooperation circumstance is introduced in this paper. Furthermore, anew exact threshold is proposed based on random matrix theory. Thesimulation results show that the proposed algorithm has an increase ofapproximately35%in the detection rate when the probability of falsealarm PFA=0.1.Afterwards, a spectrum detecting algorithm is proposed based onthe Neyman-Pearson criterion. The optimum threshold formula isobtained under multi-sensor environment. Also, the optimal requirednumber of the sensors is determined which is adapted to the signal tonoise ratio (SNR) in every frequency bin. Simulation results indicate thatunder constraints of the PFA=0.1, SNR=0dB, number of tapes K=4, onlyfour sensors are required to achieve the best detection performanceunder the Neyman-Pearson criterion. Secondly, because the improvedalgorithm does not require singular value decomposition (SVD) fordenoising, the proposed algorithm also reduces the complexity of thesystem compared with traditional algorithm based on multitapermethod with singular value decomposition (MTM-SVD).Finally, since a cognitive radio does not have a fixed spectrum, itmay need to sense multiple bands over a large frequency range.Therefore, the wideband spectrum sensing is studied in this paper. Sincethe conventional wavelet edge detection can only work on the assumption that the PSD of the received signal has irregular structures atthe edges of the occupied bands, we introduce a novel method basedon Multi-Resolution analysis and Bayes classifier, which is robustious inlow SNR circumstance and can still perform well when the edges of theoccupied bands are smooth.