Target Classification Using Bistatic SAR Images

In a radar system, a bistatic configuration is the more generic version of a monostatic configuration. Hence, it is expected and has also been verified to a large extent, that bistatic configuration can increase the range of applications of radar systems. This has given rise to the recent thrust in research activities in bistatic radar development. However, it is an established convention that the new version of a system should be able to perform all the tasks, that its older version could perform 1. Hence, the applications that are currently possible using a monostatic system 2, should be achievable using a bistatic system. One such application is that of automatic target recognition (ATR), for which monostatic radars have proved to be quite effective. The present paper would present some experimental results, which in turn would establish the feasibility of bistatic ATR. In the present paper, ATR exercises have been performed on bistatic synthetic aperture radar (SAR) images. These SAR images have been generated synthetically, using an electromagnetic simulator. In the main paper, few sets of experiments along with their results would be discussed, to analyse various aspects of the bistatic ATR problem. First of all, the dependence of bistatic ATR performance on the bistatic angle of imaging, would be analysed. Bistatic ATR performance decreases with an increase in the bistatic angle of imaging. The reason behind this would also be expounded. In the second set of experiments, the performance of monostatic and bistatic ATR would be compared. It would be shown that contrary to held reserves, bistatic ATR performance is not drastically worse than the monostatic ATR performance. Lastly, the effect of multi-polar bistatic data on ATR performance would be looked into. Even though, multipolar data analysis in bistatic domain is more complicated, still multipolar data do give more information (than single polarisation data) about the target and help in improving the bistatic ATR performance. Conclusions from the work presented in the paper, are quite encouraging. Even though bistatic data handling is more complicated, still a good ATR performance can be achieved by using bistatic radar. Bistatic multi-polar data is also exploitable. These conclusions in turn make a case for further research into bistatic radar systems