Joint range-velocity estimation with continuous active sonar

Active sonars using linear frequency-modulated (LFM) continuous waveforms are typically processed in sub-bands to facilitate high target range update rates. Target range-rate is then estimated using incoherently processed matched-filter outputs from each sub-band as input to a tracker. Target returns can easily be confused with clutter because sub-band width is typically chosen to avoid decoherence caused by target and/or platform motion. In this paper, we present methods for joint range-velocity estimation which allows for coherent combination of sub-band outputs and accounts for target/platform motion. The approach involves pre-filtering beamformed sub-band outputs using the transmitted LFM waveform, match-filtering each frequency-domain pre-filtered sub-band under a range-velocity hypothesis, and coherently combining the results. A maximum likelihood estimate (MLE) of range-velocity is formed which involves adaptively suppressing clutter and noise using an estimate of sub-band clutter covariance matrices. Simulation results are presented which indicate that joint range-velocity estimation using coherent processing across sub-bands offers a significant improvement over conventional methods and allows for more flexible selection of sub-band processing bandwidth. Moreover, the approach lends itself to handling the case of partial coherence across sub-bands and offers the potential for improved discrimination of slowly moving targets versus clutter discretes. Work supported by ONR.