Novel Algorithm for Precision Time Synchronization of Distributed Defense Systems
James P. Vogel, Evan D. Walsh, and David A. Friedman, Systems & Technology Research
Location: Ballroom E
Alternate Number 1
The DoD continues to move away from single, large defense assets toward systems of numerous small assets, such as unmanned space, aerial, ground, and/or underwater vehicles. With this push, the need is arising for robust, efficient relative PNT solutions to ensure these distributed systems can achieve their operational goals on tactical timescales. In space, specifically, precision relative PNT is an enabling technology for large constellations of satellites to execute unique distributed sensing modalities or to operate as alternate forms of PNT in GPS-denied environments. We present the STR-LIGHTT (Systems & Technology Research Loop-Inclusive Graphical Hierarchical Time Transfer) synchronization algorithm, a novel new algorithm developed to synchronize a constellation of satellites via time transfer over crosslink communication channels. Originally developed to increase the accuracy of GPS satellite clock synchronization while decreasing the system’s reliance on ground updates, STR-LIGHTT has significant potential to be extended to other important PNT problems in space, other domains, as well as cross-domain environments. The STR-LIGHTT algorithm poses constellation synchronization as “loopy belief propagation.” Previous literature on loopy belief propagation focuses on either isolating classes of problems in which loops are not necessary or are benign. STR-LIGHTT’s approach, however, seeks to maintain an optimal subset of loops to deliver highly desirable complexity, fault tolerance, and convergence properties. The algorithm was mathematically proven to converge to an accurate solution, be robust to satellite dropouts, and perform near the limit of optimal computational efficiency, all necessary properties for the algorithm to find practical use in operational systems. For the case of synchronization of the GPS satellites over crosslink channels, STR-LIGHTT converged to a sub-100-ps solution on a time scale twice as fast as a conventional solution and always converged if less than 5 out of 24 satellites dropped out. The algorithm has been further generalized to allow for more diverse geometries in time synchronization applications and subsequently applied in simulation to LEO constellations. As currently implemented, STR-LIGHTT requires either that the connectivity of the network be fixed or, such as in the case of the GPS constellation, that the connectivity is periodic with periodicity known prior to synchronization. This property makes the algorithm suitable for synchronization of fixed distributed defense systems, such as fixed ground or underwater stations, and for any constellation of satellites whose topology does not often change, such as the GPS constellation. With multiple planned LEO mega-constellations, modifications to network topologies may occur on accelerated timelines. To account for such situations, we are extending STR-LIGHTT to include more difficult constellation geometries and ad hoc networks of satellites. As the extensions are made, the scope of possible applications quickly grows much larger than the GPS time transfer problem for which STR-LIGHTT was originally developed. Beyond LEO mega-constellations, an ad hoc version of STR-LIGHTT would open opportunities for precision synchronization of unmanned aerial, ground, and underwater vehicles. The mathematical framework employed by STR-LIGHTT further lends itself to solving for joint positioning and timing – future variants of STR-LIGHTT could be used as a unified framework for PNT in highly complex and scalable cross-domain distributed systems. This work was funded by AFRL/RV under contract FA9453-19-P-0529 and Program Manager Dr. Khanh Pham. Approved for public release; distribution is unlimited. Public Affairs release approval #AFRL-2021-0225.