Signal detection is the primary job and the key technology in cognitive radio. In this paper，in order to get a superior detection performance under low Signal Noise Ratio(SNR)，a method based on the sample autocorrelation of received signals is adopted to detect the existence of authorized users’ signals. Then Binary Phase Shift Keying(BPSK) signals and the captured digital television signals are adopted as signal sources to simulate the detection algorithms in MATLAB simulation. The detection results demonstrate that the adopted method outperforms energy detection in Rayleigh fading channel and under low SNR(RSN≤–14 dB) conditions. Moreover，the overall correlation degree of received signals show some effects on detection probability.