Xona Pulsar Single-Satellite Positioning: System Perspective and Experimental Validation

Xona is deploying Pulsar, a low Earth orbit (LEO) commercial navigation system designed to deliver resilient positioning, navigation, and timing (PNT) where traditional solutions fall short. Pulsar satellites broadcast dedicated signals optimized for commercial users. This brings rapid geometry change, strong Doppler observability, and robust timing, enabling new approaches to positioning even when only one satellite is visible. Internet of Things (IoT) applications often prioritize availability over sub-meter accuracy in urban canyons, semi-indoor spaces, and other constrained environments. Many platforms are battery-powered, have strict size, weight, and power (SWaP) limits, and cannot support complex multi-sensor architectures. Leveraging LEO dynamics and signal strength, Pulsar can maintain navigation capability under these conditions without specialized user hardware. Here we present a single-satellite positioning (SSP) concept that uses available Pulsar measurements to estimate user position and receiver clock states without external aiding. Early in Pulsar deployment, only one or two satellites may be in view, yet this still benefits stationary or near-stationary users, including in semi-indoor and indoor settings. We discuss algorithmic details and system implications: SSP enables positioning with minimal satellite visibility, reduces reliance on dense constellations, and supports integration into resource-constrained platforms. We present simulation and live sky results. High-fidelity constellation simulations configured for Pulsar provide controlled performance assessment. We also present early findings from a Pulsar-enabled receiver using observations from the Pulsar-0 satellite on orbit. Preliminary tests demonstrate meter-level accuracy outdoors and indoors, highlighting potential under varied reception conditions.

💡 Research Summary

The paper presents a comprehensive study of single‑satellite positioning (SSP) using Xona’s Pulsar, a low‑Earth‑orbit (LEO) commercial navigation constellation. Pulsar’s design leverages the high received power and rapid geometry change inherent to LEO satellites to enable positioning even when only one satellite is visible. The authors first describe the Pulsar system: a planned 258‑satellite constellation at ~1080 km altitude, with the current in‑orbit validation satellite Pulsar‑0 operating at ~520 km. Pulsar transmits GNSS‑compatible L1 and L5 signals using enhanced QPSK and code‑shift keying, delivering roughly 10 dB higher signal strength than GPS at the user terminal.

The core of the SSP concept is the exploitation of the strong Doppler shift and Doppler rate produced by the fast‑moving satellite. The Doppler curve’s shape encodes the user’s ground track, and the zero‑crossing point (where Doppler changes sign) corresponds to the satellite’s closest approach, providing a latitude cue. By stacking carrier‑Doppler (converted to range‑rate) and code‑based pseudorange measurements over a portion of a pass, the algorithm converts temporal diversity into spatial diversity, making the user’s three‑dimensional position and receiver clock bias/drift observable from a single satellite with known ephemerides.

Mathematically, the state vector comprises position (x, y, z), clock bias (b) and drift (d). The geometric range, line‑of‑sight unit vector, pseudorange model, and range‑rate model are expressed in standard GNSS form. Residuals and Jacobians are derived analytically, and two nonlinear least‑squares solvers are employed: an un‑weighted Gauss‑Newton (GN) method for simulation studies and a weighted Levenberg‑Marquardt (LM) method for real‑world data, with adaptive damping.

Simulation results use a high‑fidelity Pulsar constellation simulator. For a static user, a 15‑minute pass sampled at 30 s intervals (100 Hz Doppler, 1 Hz pseudorange) yields sub‑meter RMS position error (≈0.8 m) when the signal‑to‑noise ratio exceeds 30 dB. Clock bias converges to ~10 ns. Sensitivity analyses show that ephemeris and satellite clock errors dominate the error budget, while shorter passes (<5 min) reduce observability and may prevent convergence.

Experimental validation is performed with a prototype receiver tracking Pulsar‑0. Outdoor tests achieve ~1.2 m RMS error; indoor tests (through walls and with multipath) still deliver ~2.3 m RMS error. These results demonstrate that even with only one visible satellite, the combined pseudorange‑Doppler solution provides reliable positioning for static or slowly moving users, meeting the availability‑centric requirements of many IoT and logistics applications.

The paper discusses system‑level implications: SSP reduces dependence on dense constellations, enables early‑deployment utility when only a few satellites are operational, and fits within stringent size‑weight‑power (SWaP) constraints of battery‑powered IoT devices because no auxiliary sensors or complex multi‑constellation processing are required. Limitations include sensitivity to satellite orbit and clock errors, reduced performance for dynamic users, and the need for longer observation windows to accumulate sufficient geometry change.

Future work outlined includes multi‑pass data fusion, integration with inertial or terrestrial radio‑based aiding, real‑time adaptive filtering to mitigate model mismatches, and the development of correction services (e.g., satellite‑based ranging corrections) to push accuracy toward decimeter levels. Overall, the study validates that LEO‑based single‑satellite navigation is a viable complement to traditional GNSS, offering robust positioning in environments where conventional signals are weak or obstructed.