Correlation Matching Approach for Spectrum Sensing in Open Spectrum Communications

In the new paradigm of open spectrum access, the envisioned radio agility calls for fast and accurate spectrum sensing, this challenges traditional spectral estimation. In this study, we propose three new procedures that are able to sense the known spectrum of the candidate or primary user, fulfilling the requirements of open spectrum scenarios. These procedures are developed by following the framework of correlation matching, changing the traditional single frequency scan to a spectral scan with a particular shape and generalizing filter-bank designs. The proposed techniques are called Candidate methods, because their goal is to react only when the candidate's spectral shape is present. First, Candidate-F is proposed as a spectral detection method, where this is based on minimizing the Frobenius distance between correlation matrices, and can be viewed as an extended version of the weighted overlapped spectrum averaging estimate. Next, Candidate-G is presented, which is a new procedure that is based on a geodesic distance, and that presents the lowest complexity. Lastly, a third procedure is studied, Candidate-M, which provides the most compliant performance with the demanding open spectrum scenario by generalizing the Capon-spectral estimator. By means of the analytical results, simulations of receiver operating characteristics, and estimation variance, this study shows the advantages of Candidate-M over the existing filter bank or cyclostationary detector methods.