Fast reliable spectrum sensing (SS) is a crucial problem in the cognitive radio systems. To address this issue, cyclostationarity-based detection methods, which are generally more complex but more reliable than energy detection methods, have been proposed. This paper presents a new method to detect the presence of the second-order cyclostationarity in the OFDM-based primary user (PU) signals. The proposed method has a low computational complexity, while it presents a performance close to the well-known GLRT-based Dandawaté-Giannakis's method. Moreover, since the proposed low-complexity method is robust against the noise uncertainty, it can be a good alternative to the energy detection method. We further propose some cooperative spectrum sensing methods to improve the detection performance. Extensive simulation results confirm the superiority of the proposed schemes.