Abstract
Joint estimation of channel impulse response and Doppler drift can be confounding since source motion and media temporal fluctuations can be misidentified. In addition, multipaths can be associated with different though not independent Dopplers. We present a novel algorithm for joint estimation of channel impulse response and mean Doppler drift by assuming prior constraints that favor lower complexity of the posterior model. The resulting estimator is compared with the popular maximum likelihood estimator (MLE). Approximate posterior covariance estimates provide Doppler/channel correlation and confidence bounds. We test this method with synthetic channel data and experimental data from a shallow water environment and compare the method with the classical MLE.