This paper addresses the problem of dynamic spectrum access, discussing a new method for spectrum sensing in cognitive radio networks. In particular, an innovative technique for the detection of an OFDM-based primary user signal is here proposed. Performance analysis is carried out in comparison with conventional spectrum sensing method that exploits the autocorrelation coefficients. Unlike the conventional method, our strategy is completely blind and can be applied with no a priori knowledge of any characteristics of the signals of interest. Simulation results have been carried out under operating settings, typical of wireless local area networks.