Blind Vector Parameter Estimation for Burst Type CPM Transmissions

Short burst continuous phase modulation transmission is of practical relevance in e.g. frequency hopping systems applied in sensor or tactical networks. The channel conditions can be seen as mutually uncorrelated for each burst due to spectral and or temporal separation of those. Because of this time variant nature, a recurring acquisition of the impairment parameters is required for each burst. In this work, a blind joint estimation of several parameters in a flat fading environment for continuous phase modulation bursts is realized by the expectation maximization algorithm. The main contributions are first the formulation of the expectation and maximization steps to enable the joint computation of the maximum likelihood parameter estimates and second the analysis of the likelihood functions to obtain an optimized initialization for the algorithm. It is shown, that the joint estimator produces unbiased estimates and its performance in terms of the mean squared estimation error achieves the theoretical limits, i.e. the modified Cramér-Rao-Vector-Bound and slightly outperforms a state of the art pilot based estimator. Furthermore, the effective throughput is discussed and bit and frame error rates are compared to each other and to the perfectly synchronized estimator. Its computational complexity is analyzed and efficient computation steps and further approaches are outlined to decrease it.