Parallel sigma point and particle filters for orbit determination

Haijun Shen,C. Karlgaard,R. Russell,V. Vittaldev,E. Pellegrini

2013-01-01

Abstract:Advanced filters like the sigma point and particle filters are more accurate than the extended Kalman filter for nonlinear and non-Gaussian applications, but experience drawbacks such as being computationally expensive if implemented in serial. However, since the majority of the computation can be carried out simultaneously, these filters are inherently well suited for parallel computing. This research leverages inexpensive and personal-level parallel computing architectures, such as the NVIDIA Graphics Processing Units (GPUs) and multi-core CPUs to exploit such parallelism. In particular, parallelization of the Unscented Kalman filter (UKF) and the bootstrap particle filter applied to an orbital determination problem is the main objective of this work. This work demonstrates that the UKF and the particle filter can be applied to the orbital determination problem with great benefits in the presence of nonlinearities and non-Gaussian noises. An 8-time speedup is achieved for the UKF with an 8-thread CPU, and up to two orders of magnitude speedups are achieved using a M2090 GPU. The results also show that real-time implementations for both filters are feasible.