In this paper, we address the problem of estimating some signal spectrum characteristics in passive surveillant systems. The received signal is assumed to be completely unknown. We estimate the spectrum of the received signal using a well known spectrum estimation method, Welch method. By some modifications the proposed method uses an adaptive threshold based on Constant False Alarm Rate (CFAR) processor to first detect the existence of signal spectrum, next estimates the carrier frequency of the received signal. We evaluate the method based on computer simulations and by RMSE measures. Numerical simulations demonstrate the capability of our method to estimate the begin and the end frequencies of a given chirp signal and also its bandwidth, accurately in low signal to noise ratios (SNR).