D.D. Stoyanov - Post-graduate Student, Chair Dynamics of Electronic Systems, P.G. Demidov Yaroslavl State University. E-mail: dmdrstoyanov@gmail.com A.L. Priorov - Dr.Sc. (Eng.), Associate Professor, Chair Dynamics of Electronic Systems, P.G. Demidov Yaroslavl State University. E-mail: andcat@yandex.ru M.A. Dubov - Ph.D. (Eng.), assistant, Chair Dynamics of Electronic Systems, P.G. Demidov Yaroslavl State University. E-mail: michaeldubov@gmail.com

A task of frequency spectrum monitoring has become more urgent nowadays. Not only have the radio monitoring services needed to solve this task but also it is actual for cognitive radio systems. However, most available in the open literature wideband signals search algorithms need a priori information about the bandwidth or cyclostationary internal of detectable signal that in practice mostly unknown. In this paper the modification of nonparametric detection algorithm based on the statistics of the Wilcoxon sign rank is proposed to search for broadband modern radio communication standards signals with digital modulation in presents with additive noise. The proposed algorithm is slightly sensitive to the statistical characteristics of the detected signals. A proposed modification of Wilcoxon detector is to adapt the algorithm to detect radio signals in the frequency domain instead originally time domain. It is done by development of transformation rules and a received statistics and samples of the observed process mapping. The effectiveness analysis of the proposed algorithm is produced by statistical modeling. Comparison of modified nonparametric algorithm with a known parametric algorithm is presented. Also in the paper presents the results of the real GSM signals detection received by a universal hardware-software complex USRP (Universal Software Radio Peripheral). Analysis of the data confirms that the proposed modification of nonparametric detection algorithm allows signals detection in the frequency domain under nonparametric apriority uncertainty. The efficiency of the proposed algorithm is compared with the known earlier algorithm gives 1-13 dB gain by SNR for signals having different statistical characteristics with SNR in range of -16 dB to +16 dB. The proposed algorithm provides a fairly accurate estimate of the bandwidth radio signals detected in contrast to the parametric, which generates a small width frequency bands plurality requiring additional methods of evaluation.