The use of wireless communication resources, namely energy and spectrum, has significantly expanded recently. In order to mitigate incongruity ideas of spectrum scarcity and spectrum underutilization, the Cognitive radio (CR) technology considered as an optimistic solution. In Cognitive radio network (CRN), spectrum detection is a critical step in identifying the spectrum gaps or spectrum holes. To detect whether primary user is present or not, many spectrum detection methods like Energy detection (ED), Entropy detection, Matched filter and Cyclostationary detection have been developed. Among those, Energy detection (ED) has public the most observance from researchers due to its fast sensing time, simplicity on implementation and low computational complexity. Nevertheless, due to noise uncertainty problem, the performance of conventional detector is very low at low SNR. In this article, three-stage spectrum sensing technique is proposed to get better performance of single stage and two-stage spectrum sensing techniques and also to overcome the problem occurred due to uncertainty of noise at low SNR values. Furthermore, the performance comparison among three stages, two stage and single stage spectrum sensing techniques are performed over fading channel. The proposed three-stage detection technique has an about 1 dB and 2.5 dB performance improvement as compared to two-stage ED + Renyi Entropy (RE) and two-stage ED + Kapurs Entropy (KE) respectively. The cooperative three-stage detection technique using OR rule has an about 4 dB and 5 dB performance improvement as compared to single node and AND rule cooperative spectrum sensing (CSS) technique respectively at number of user (NU) = 5.