Real-time autonomous orbit determination of LEO satellites using GPS

This paper describes an investigation, carried out for the UK Defence and Evaluation Research Agency (DERA), into the use of GPS as a real-time autonomous (ie no ground tracking) Orbit Determination system for use on-board LEO satellites. The investigation concentrated on firstly, the development and testing of a processing technique which could be implemented in a real-time system, and secondly, on the impact of this processing technique on the processing and storage requirements for a LEO satellite, leading to a high level OD module design. The target real-time accuracy set by the DERA was 1 m, using Precise Positioning Service (PPS) pseudoranges. The study has developed the Kalman filter solution, referred to as the reduced dynamic technique, which is ideally suited to use in a real-time environment, as it produces an estimate of the satellite's position whenever a new set of measurements is acquired. The results presented in this paper are a combination of real data and simulated data. The real data studies were restricted to Standard Positioning Service (SPS) data sets, specifically from the TOPEX/POSEIDON mission which carries a JPL-designed GPS receiver. The study also produced simulated SPS measurements, using TOPEX/POSEIDON as the model. The close agreement between the results from real and simulated data provided a validation of the simulation, so that simulated PPS data could then be used to draw the required conclusions of the study. Memory size and the relatively complex processing requirements dictate the choice of a state-of-the-art on-board processor. Two processors, the ESTEC developed ERC32 and the R3000, are based on well established RISC instruction sets and offer a high speed, high power but expensive solution. The third option, the military standard 1750A microprocessor, despite being the least powerful of the three candidates, is more than capable of generating the required single orbit prediction each minute, having a margin of at least a factor of 10. This space-proven microprocessor is physically more resilient, requires much less power, is available off the shelf and is well supported. It is therefore the preferred option. In summary, it has been established that an orbit determination processing system with real-time one metre accuracy can easily be implemented using current off-the-shelf technology. Such a system would have mass, power, volume and cost characteristics which could prove attractive for a wide range of LEO missions.