Performance Comparison of Microprocessors for Space-based Navigation Applications

Kurzfassung

In this paper the absolute and relative run-time performance of different space-borne microprocessors are compared. A realistic test scenario which comprises the propagation of a state vector and its transition matrix for a satellite in Low Earth Orbit (LEO) has been used to perform the tests. In addition, the computation time required for a Kalman filter time update with respect to the order and degree of the Earth’s gravity field has been evaluated. The microprocessors considered are typical platforms for space-based applications. In particular detailed benchmark testing is conducted on the ARM7TDMI (Phoenix GPS receiver on the X-SAT mission), on the PowerPC823e (BIRD spacecraft onboard computer), LEON (PRISMA spacecraft onboard computer) and NiosII (NAMURU board for GPS applications). The benchmark software has been integrated as sole application into the respective operating and software system of the microprocessors. The analysis has been performed in the framework of the Swedish led PRISMA formation flying mission in which the German Aerospace Center (DLR) provides various key contributions which comprise a redundant GPS hardware architecture for the two spacecraft, a real-time navigation software to support formation flying during all phases, and dedicated experiments for absolute and relative orbit control.