Knowledge-based moving target indication
Moving target indication is one of the central tasks of airborne radar systems. The efficiency of the signal processing used for this purpose can be significantly improved if additional information relating to the observed scene is taken into account.
The detection and localization of moving objects (moving target indication, MTI) in the air, on the ground or at sea is one of the core tasks of multi-function radar systems on airborne platforms. However, particularly the detection and localization of moving objects on the ground (ground moving target indication, GMTI) is made more difficult by clutter, i.e. interfering echoes from the surface of the Earth. If the radar system has several parallel arranged receive channels, these interfering radar echoes can be suppressed with multi-channel signal processing techniques, thus also allowing the detection of slow moving objects.
Space-time adaptive processing (STAP) is particularly suitable for clutter suppression. It uses spatio-temporal clutter filters, the coefficients of which are estimated on the basis of radar echoes that originate from range cells other than the test cell that is to be investigated for the presence of a moving target. The extent to which the characteristics of the clutter in this secondary data conform with those of the clutter in the test cell, greatly determines the extent to which an adaptive filter such as this can suppress clutter, thus leading to a greater probability of detection and localization precision.
If the characteristics of the clutter – except for power – are distance-independent, the secondary data can be selected arbitrarily. A different scenario applies for inhomogeneous clutters, the characteristics of which change with range and over time. In this case, the selection of the secondary data has a decisive influence on the GMTI capacity of the radar system.
The best performance is achieved when the characteristics of the clutter in the secondary data are as similar as possible to those of the clutter in the test cell. To achieve this, one can either use information derived from the radar data itself or further external data to select the secondary data. In both cases, this is known as knowledge-based signal processing.
At Fraunhofer FHR, knowledge-based signal processing techniques are compared and further developed. The data that is required here is acquired using the SAR/GMTI demonstrator system PAMIR which was developed at the institute. The advantage of knowledge-based signal processing is illustrated in the following by way of an example.
Utilization of maps
Maps are a possible source for the determination of suitable secondary data. They can be used to segment the scene covered by the radar into areas with different backscatter characteristics. The map can be a clutter map, i.e. more or less a high-resolution radar image of the scene, but also a conventional digital map of the area. In the latter case, it would be an advantage if the map also contained information on land cover.
A segmentation is particularly promising in cases where the scene contains both water and land. This is due to the different backscatter characteristics over land and over water. Moreover, due to the fact that the reflection is particularly strong at boundaries between water and land, the shorelines that do not run in or parallel to the line of vision of the radar will be clearly visible in the radar image.
The figures illustrate the advantages of knowledge-based signal processing. Figure 1 shows the clutter power of a coastal scene received by the radar system. The broken lines mark the transition area between the sea (below) and land. Figures 2 and 3 show the result of adaptive clutter suppression with STAP, whereby the secondary data was first selected without knowledge of the scene and then selected after segmenting the scene into three areas, namely sea, shore area and land. The white circles highlight each of the remaining inhomogeneities, which represent potential false alarms. The utilization of map information clearly reduces the number of potential false alarms. In this case, they are attributable to the echoes of trees moving in the wind.