Spectral-correlation based estimation of channel parameters by noncoherent data processing

A recently proposed cyclostationarity exploiting, i.e., spectral-correlation based, method for estimating the channel parameters needs a long integration time to obtain a sufficient level of noise and interference suppression in weak-signal case. The paper overcomes such a limitation by adopting a noncoherent data processing, i.e., by segmenting data and averaging the estimates obtained from each segment. The appropriate length of the temporal segment is determined as a compromise between computation time and estimate accuracy. It is shown that the noncoherent data processing can work well also when the accuracy in knowledge of the cycle frequency (i.e., the parameter characterizing the cyclostationarity) is not sufficient to assure an acceptable performance level with a coherent data processing