Adaptive Multiframe Detection Algorithm With Range-Doppler-Azimuth Measurements

Abstract

The multi-frame detection (MFD) with range-Doppler-azimuth measurements has been shown to be significant in detecting and tracking weak targets in active radar and sonar systems. Existing methods suffer from integrated energy loss for a long time noncoherent observation sequences due to approximate target evolution model or rough energy integration strategies. Moreover, it is hard to maintain consistency of multi-frame integration for targets in different near-field and far-field regions. By carefully mapping echo measurements to a discrete grid space without the introduction of new conversion errors, we proposed an efficient solution to build an accurate grid state model based on sensor parameters. Then, the combination of the grid state model and target evolution relationship among adjacent frames enables an adaptive MFD implementation. Each search path during multi-frame integration is adjusted adaptively with different grid state sizes. Finally, an improved strategy based on the predicted estimate is further presented to reduce the complexity of algorithms and improve the accuracy of possible search paths. The proposed methods can effectively integrate target energies among multi-frame range-Doppler-azimuth measurements, and maintain the robustness of the long time integration for targets in different near-field and far-field regions. Numerical results and tests with real radar data further show that the proposed method achieves high detection probability and tracking accuracy for weak targets with range-Doppler-azimuth measurements.