|Abstract:||Planet is a vertically integrated aerospace and data analytics company that operates the world’s largest commercial fleet of satellites. Our mission is to image the whole world everyday, and make change visible, accessible, and actionable. We have launched over 300 satellites and built up an automated mission control and ground station infrastructure to monitor and control the satellites, and download the imagery data. The majority of these satellites are 3U cubesats known as “Doves”. Designing, building, and operating small satellites is replete with challenges, but one of particular interest is orbit determination. Independent orbit determination is an essential capability for responsibly operating small satellites. Public sources of ephemerides are known to be unreliable, especially during the launch and early operations phase of the mission as it is common for a given launch vehicle to deploy dozens of similar satellites in nearly identical orbits. Planet has invested in two independent, cooperative schemes for orbit determination: a time-of-flight ranging system using UHF radios and a GNSS-based system that uses the GPS L1 C/A signal. Unlike primary radar systems, both of these schemes are able to unambiguously identify each Dove. The Dove spacecraft is size, weight and power constrained. In the launch configuration, each satellite is roughly 30 cm x 10 cm x 10 cm with a mass of 5 kg. Volume constraints in particular made commercially available GNSS receiver modules unsuitable for this application. Since the Dove design already incorporated onboard compute and communication capabilities, we saw an opportunity to reuse those resources to provide a GNSS receiver solution. The only additional components that were needed for this receiver were a patch antenna, low-noise amplifier, SAW filter, and a GNSS RF front-end chipset. The volume, weight and power of these added components is 1 cm3 , 1.5 g and 115 mW, respectively. Unlike most GNSS receivers, the Dove receiver does not provide real-time results. Rather, it works by way of taking short (32 ms) snapshots of the GPS L1 channel. These recordings are downlinked and post-processed using published GPS ephemerides and prior estimates of the satellite’s orbits to obtain state vectors. GPS satellite geometry and received power are excellent from the Dove’s nominal 500 km orbit (inside the Terrestrial Service Volume) which allows use of a small patch antenna. State vectors are ingested by our orbit-fitting service which in turn generates ephemerides for each of our satellites. An orbit model is fit to the GNSS state vectors via a non-linear least squares. The orbit model consists of a Cartesian state vector and ballistic coefficient numerically propagated under a force model with a GGM03C gravity field and the NRL-MSIS00 atmosphere model. The orbit fit process is run periodically against the most recent GNSS-derived state vectors. This entire processing flow is automated and runs on a scalable cloud compute solution. Moving GNSS sample processing off the satellite comes at the price of added latency to the orbit determination process. This is acceptable for our medium resolution (3.7 m Ground Sample Distance) imaging mission, however, it may not be applicable for systems with higher precision requirements or for missions with greater dynamics (e.g., satellites with propulsion capabilities). Nevertheless, it has proven to be a size, weight and power efficient approach for implementing GNSS reception on a small satellite. To date this solution has been used operationally on over 185 Dove satellites.|
Proceedings of the 32nd International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2019)
September 16 - 20, 2019
Hyatt Regency Miami
|Pages:||1157 - 1163|
|Cite this article:||
Kingsbury, Ryan W., Self, Matthew Cullen, Vittaldev, Vivek, Foster, Cyrus, "Dove GPS: An Unconventional Approach to CubeSat Orbit Determination," Proceedings of the 32nd International Technical Meeting of the Satellite Division of The Institute of Navigation (ION GNSS+ 2019), Miami, Florida, September 2019, pp. 1157-1163.
ION Members/Non-Members: 1 Download Credit