Recently, the Federal Communications Commission has mandated the use of a database to observe the spectral opportunity at a given place over time along with the use of spectrum sensing. It is arduous to obtain the database for spectral opportunity for all places, and since spectral opportunities rapidly vary with location, a new paradigm to obtain the spectral opportunity needs to be determined. We propose a method to obtain the spectral opportunity at a given location for specific channels, such as television (TV) and frequency-modulated (FM) bands, which follow deterministic time usage pattern. Since these channels have a periodic deterministic time usage pattern for every 24 h, learning the spectral opportunities and utilizing these for consequent time periods have better efficiency than other proposed schemes. Once the spectral opportunity is known and if we have multiple channels from which to select, then a decision on the selection of a channel is made. The spectrum decision is done to satisfy the constraints imposed by a user. Since, with our proposed method, the spectral opportunity is known ahead of time, better spectrum decisions can be made. To this end, we propose three different channel selection techniques with different constraints applied. The first channel selection technique provides the selection of channels with a minimum number of channel switching and, thus, reduces the jitters, switching delays, and packet losses. This channel selection technique is appurtenant to real-time applications such as voice over Internet Protocol. The second channel selection technique provides the selection of channels that maximizes the overall throughput of the system and is particularly useful for bandwidth-intensive applications such as BitTorrent. The third channel selection technique provides a compromised value between minimal channel switching and maximal throughput, thus providing a control over the bandwidth demand and real-time requirement. A comparison ...