ceeding 
y Con- 
Models 
aneiro, 
léSsung 
ation", 
Maps, 
1986 
spects 
, Italy 
Satel- 
Arbor, 
  
DEFINING A KNOWLEDGE BASED SAR PROCESSOR 
W. Noack 
Deutsche Forschungs- und Versuchsanstalt 
für Luft- und Raumfahrt 
DFVLR 
8031 Oberpfaffenhofen 
Federal Republic of Germany 
ABSTRACT 
The fact that future microwave sensors like the ERS-1 Synthetic Apertur Radar 
(SAR) will be operational systems, requires a processor system design which 
is significantly different from existing SAR correlators. In addition more 
attention must be paid to the user community needs in terms of various pro- 
duct levels and adequate production and organization schemes. 
This paper deals with the classification of an expert system as a part of 
DFVLR's Intelligent SAR Processor (ISAR) which is identified by a distributed 
architecture using a high speed array processor, enhanced by a two-dimen- 
sional accessable memory, a front-end processor and a knowledge engineering 
workstation. In order to ensure consistency and correctness during the 
development phase an expert system will support the program designer in 
accumulating the knowledge base. As a result the system will be accessible 
and comprehensive for both experts and system operators. 
1. INTRODUCTION 
During the last years the user requirements for SAR products have been 
evolving intensively. Among them are the requirements for multifrequency, 
multipolarization, high radiometric and geometric precision, varying inci- 
dence angles and related parameters. Furthermore a lot of experiments are 
demanding multiseasonal data acquisition or the inclusion of calibration 
experiments. This will result in a raised number of different products. 
However, the biggest problem beside the product variety is the requirement 
for high throughput together with the demand for high precision. 
A simple recipy would be to speed up the correlation process employing faster 
array processors based on a software oriented implementation approach. 
Another possibility would be to use pure hardware boxes instead. But what 
you would gain is- throughput and the processing time can easily be pushed 
down below the product set up time. Moreover the hardware approach is of 
disadvantage because the programming of such a system usually involves 
microcoding and even nanocoding. 
381