Analysis of CPM signals using higher-order statistics

The authors address the problem of the estimation of the modulation index of continuous phase modulated (CPM) signals in noise. In a previous work (1993) they considered the same problem when the CPM signal was sampled at the symbol rate and thus constituted a stationary sequence. In this case, sampling is performed at a higher rate, and thus the discrete sequence becomes cyclostationary. Analytical expressions are derived for the cyclic autocorrelation and fourth-order cumulant sequences that permit the formulation of a maximum likelihood estimator. Simulation examples illustrate the performance of both methods.<<ETX>>