A Genetic Algorithm with Tournament Selection for Automated Testing of Satellite On-board Image Processing

In the satellite domain, on-board image processing technologies are subject to extremely strict requirements with respect to reliability and accuracy in hard real-time. In this paper, we address the problem of automatically selecting test cases from a huge input domain that are specifically tailored to provoke mission-critical behavior of satellite on-board image processing applications. Due to the large input domain of such applications, it is infeasible to exhaustively execute all possible test cases. Moreover, the high number of input parameters and complex computations make it difficult to find specific test cases that cause mission-critical behavior. To overcome this problem, we define a test approach that is based on a genetic algorithm combined with input parameter partitioning. We partition the input parameters into equivalence classes to automatically generate a reduced search space with complete coverage of the input domain. Based on the reduced search space, we run a genetic algorithm to automatically select test cases that provoke worst case execution times and inaccurate results of the satellite on-board image processing application. For this purpose, we define a two-criteria fitness function and evaluate two different selection methods with a case study from the satellite domain. We show the efficiency of our test approach on experimental results from the Fine Guidance System of the ESA medium-class mission PLATO.