Blind free band detector based on the sparsity of the Cyclic Autocorrelation function

In this paper, we will firstly show that the Cyclic Autocorrelation function (CAF) is a sparse function in the cyclic frequency domain. Then using this property we propose a new CAF estimator, using Compressed Sensing (CS) technique with OMP algorithm [1]. This estimator outperforms the classic estimator used in [2]. Furthermore, since our estimator does not need any information, we claim that it is a blind estimator whereas the estimator used in [2] is clearly not blind because it needs the knowledge of the cyclic frequency. Using this new CAF estimator we proposed in the second part of this paper a new blind free bands detector. It assumes that two estimated CAF of two successive packets of samples, should have close cyclic frequencies, if a telecommunication signal is present. This new detector is a soft version of the detector already presented in [3]. This methods outperforms the cyclostationnarity detector of Dantawate Giannakis of [2].