As network geometry becomes complicated, it is difficult to allocate the frequency to user predeterminately. For this reason, open spectrum access paradigm gets an attention again. In a Flexible Wireless AP (FWAP), proposed by NTT , a unified wireless platform provides wideband frequency, e.g., from 0 Hz to 50 GHz, and it can receive various types of wireless signals. In this system, each mobile device should find a vacant frequency band, which requires to sample a wideband signal. In a mobile device, however, it is a huge burden to sample the wideband signal even though the sub-Nyquist sampler  is implemented. In this paper, we develop a new wideband cognitive radio technique that can provide available frequency candidates for a prospective user. This technique is based on a trade off between a minor performance and a major performance. By trading off a False Alarm (FA) performance, we can reduce a sampling rate for obtaining an appropriate False Negative (FN) performance.