A High Performance Conjunction Analysis Technique for Cluster and Multi-Core Computers

Satellite collision avoidance analysis has received increased emphasis since the Iridium-COSMOS collision of February 2009, including serious consideration of regular “all vs. all” collision avoidance analysis of the entire space catalog by the US Air Force. This paper describes a technique for the very fast screening of large conjunction analysis problems. This technique makes no assumptions about orbit shape or motion and is suitable for the analysis of powered or hyperbolic flight with no special consideration required. Implementations of this technique (CSieve) on both cluster computers and large multi-core computers are presented including discussion of analysis scaling. Results are presented for all vs. all analysis of up to 93K objects. Discussion of batch vs. continuous analysis is also presented 1. Nomenclature Δt Analysis time step X, Y, Z Component directions of a Cartesian coordinate frame. Xlow/Xhigh Minimum and Maximum values the X position of a satellite within the current time step Ylow/Yhigh Minimum and Maximum values the Y position of a satellite within the current time step Zlow/Zhigh Minimum and Maximum values the X position of a satellite within the current time step A.Xlow The Xlow value for object A N Number of satellites in the analysis catalog PCA Point of Closest Approach TCA Time of Closest Approach R Range Screening Threshold C The set of objects whose trajectory circumscribing volumes intersect the volume of a specified object. Cx The set of objects whose trajectory circumscribing volumes intersect the volume of a specified object in the X direction. r Relative position vector r Relative velocity vector