Time-frequency analysis of multipath doppler signatures of maneuvering targets

Multipath signals arise in many active sensing modalities, such as radar and sonar. Moving targets cause Doppler effects that could vary for different paths. Target Doppler information corresponding to direct and non-direct paths is important for moving target localizations and classifications, particularly when narrowband signals are involved. This information, however, becomes difficult to reveal when dealing with nonlinear time-varying multi-component Doppler signals. In this paper, we introduce a new time-frequency analysis technique based on the local phase information to accurately extract complex Doppler signature of each signal arrival. This is achieved through short-time polynomial phase modeling of data segments. The global behavior is obtained by fusion of the phases across neighboring segments using the data phase continuity property. The offering of the proposed technique is demonstrated using synthetic data in an over-the-horizon radar platform.