Blind Detection of Cyclostationary Features in the Context of Cognitive Radio

International audienceThe methods of dynamic access to spectrumdeveloped in Cognitive Radio require efficient and robustspectrum detectors. Most of these detectors suffer fromfour main limits: the computational cost required forthe detection procedure; the need of prior knowledge ofPrimary User’s (PU) signal features; the poor performancesobtained in low SNR (Signal to Noise Ratio) environment;finding an optimal detection threshold is a crucial issue.In this paper, we propose a blind detection method basedon the cyclostationary features of communication signalsto overcome the four limits of spectrum sensors. In orderto reduce the computational cost, the FFT AccumulationMethod has been adjusted to estimate the cyclic spectrumof the intercepted signal. Then, the spectrum coherenceprinciple is used to catch the periodicity hidden in thecyclic autocorrelation function of this signal. The hiddenperiodicity is revealed by the crest factor of the cyclicdomain profile. The detection of PU’s signal is achieved bycomparing the embedded periodicity level with a predeterminedthreshold related to the crest factor. This thresholdvaries randomly dependent on the SNR. Then, we havemodelized the distribution law of the threshold in orderto select the optimal value. Using the crest factor of thecyclic domain profile as a detection criterion has permittedto develop a spectrum sensor which is able to work in ablind context. Simulation results corroborate the efficiencyand robustness of the proposed detector compared with theclassical Energy Detector