Impact of Road Vehicle Accelerations on SAR-GMTI Motion Parameter Estimation

Kurzfassung

In recent years many powerful techniques and algorithms have been developed to detect moving targets and estimate their motion parameters from single- or multi-channel SAR data. In case of single- and two-channel systems, most of the developed algorithms rely on analysis of the Doppler history. Nowadays it is known, that even small unconsidered across-track accelerations can bias the along-track velocity estimation. Since we want to monitor real and more complex traffic scenarios with a future traffic monitoring system like TRAMRAD, we must know which target accelerations we have to handle in reality. For this reason a common passenger car was equipped with an inertial measurement system and differential GPS to measure accelerations in all three dimensions during rush-hour traffic. In this paper the results of the acceleration measurements are presented and discussed. The standard deviations of the measured accelerations are in the order of 0.5 m/s² for accelerations in driving direction and 0.6 m/s² for radial accelerations. A theoretical analysis (which is verified by detailed simulations) of the Doppler slope shows also that at such high across-track accelerations a reliable estimation of the along-track velocity by means of a Doppler slope analysis without further information is unemployable in practice. Also oscillations of the car body along the vertical axis are investigated in this paper. From the field of vehicle dynamics it is known that the eigen frequencies of the car body are in the range from 0.7 to 2.0 Hz. Deflections in the order of one wavelength (X-band) or higher are possible at such frequencies. The simulation results for spaceborne SAR systems with integration times in the order of one second show that the shape and azimuth shift of the impulse response depend beside the oscillation frequency and the deflection also on the initial phase of the oscillation. However, at practical applications the main part of the energy could also be reflected by double bounce from the road surface. Thus, further investigations in the topic of vehicle oscillations by using real radar data are necessary. Finally, some basic ideas are presented which enable a reliable separation between along-track velocity and across-track acceleration. For example, the easiest way to separate both just mentioned motion parameters is the use of a road database, from which the information about the motion direction of the assigned vehicle can be extracted. Hence, the accuracy of along-track velocity estimation is mainly given by the accuracy of the estimated across-track velocity and the angle of the road section in relation to the flight path of the SAR platform.