Users Number And SNR Estimation In Underlay Spectrum Sharing Mode

The technology of underlay spectrum sharing improves the spectrum utilization with the primary user and secondary uses sharing the spectrum resources, and becomes a research hotspot gradually. In underlay mode, in order to determine the transmission strategy, the new secondary user needs to know the source number and SNR before accessing the licensed band. Therefore, basing on the background of underlay, the research of users number and SNR estimation has important theoretical significance and practical value. The main contents of this paper are as follows:This paper proposes a users number estimation method of time-frequency overlapped signals in underlay spectrum sharing mode. Firstly, according to the signal correlation and threshold, the single receiving signal is expanded into a virtual array signal with dynamic delay. Then based on the singular value decomposition(SVD) of the virtual array signal, the first eigenvector ch1 was constructed by the difference between two adjacent singular values. Then, the second eigenvector ch2 was constructed by the division of adjacent value which come from ch1. Finally, utilizing the characteristic that there are lager difference between the signals' singular value and noise's singular value and the difference among the noise's singular value are very small, the users number is estimated by searching the location of the maximum in the ch2.This paper derives the Cramer-Rao bound(CRB) of SNR estimation and proposed a SNR estimation method of time-frequency overlapped signals. Firstly, the second moment and forth-order moment ware used to construct signals' kurtosis, and the relationship of the kurtosis between the received signal and the every signal component was derived. Then the normalized power of signal component was calculated according to the characteristic that the kurtosis of Gaussian white noise is zero and the MPSK(M=2.4.8) signals' kurtosis have no relationship with the power of itself. Finally, the SNR of time-frequency overlapped signals is estimated by transforming it into signal component's normalized power estimation.Some experiments are conducted to verify the proposed methods in this paper, and the simulation analysis is carried out respectively from several aspects such as spectrum overlapped ratio, power ratio and the number of signal component. The simulation results show that the proposed users number estimation method can estimate the users number more effectively in high spectrum overlapped ratio and low SNR condition compare with the traditional method that the correct rate is bigger than 90% when the SNR is-2d B. The simulation results also show that the normalized MSE of proposed SNR estimation method can close to the derived CRB and isn't sensitive to the spectrum overlapped ratio and power ratio.