GNSS Reception at the Moon: First Results of the Lunar GNSS Receiver Experiment (LuGRE)

3.1 Payload

The LuGRE hardware was designed to support the top-level scientific requirements described in Section 2 while also satisfying the do-no-harm requirement as a payload hosted on the BGM1 lander. The design prioritized robustness, adaptability, minimal mass, and power efficiency, which allowed LuGRE to withstand the severe challenges of the deep space environment (Fantinato et al., 2022, 2025; Pulliero et al., 2023). The LuGRE payload development process resulted in a flight model with a total mass of less than 5 kg and a maximum operational power of 14 W.

The LuGRE payload combines several custom hardware components. These components and their locations on the BGM1 lander are shown in Figure 2 and are described in detail in the following sections.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 2

LuGRE hardware components and their relative locations on the BGM1 lander. (a) LuGRE HGA, (b) LuGRE LNA, (c) LuGRE receiver, (d) Location of the LuGRE HGA on the BGM1 lander. The NGLR payload is just above the LuGRE HGA on the same panel. Image: Firefly Aerospace, (e) Location of the LuGRE receiver and LNA below the top deck of the BGM1 lander. Image: Firefly Aerospace.

3.1.1 High-Gain Antenna

The LuGRE HGA was designed to operate in both the L1/E1 and L5/E5a GNSS bands. It was designed as a 3×3 planar right-hand circularly polarized passive L-band antenna with a high gain and directive main-lobe. The resulting antenna design achieved a peak boresight gain of 15.35 dBic and 12° half-power beamwidth in L1/E1 and 14.56 dBic and 16° half-power beamwidth in L5/E5. Figure 3 depicts the antenna gain patterns measured in the NASA Goddard Space Flight Center (GSFC) anechoic chamber as a standalone component, mounted on a mockup panel with an integrated model of the neighboring Next-Generation Lunar Retroreflector (NGLR) payload. Additionally, the HGA contained an integrated notch filter to limit the contributions of out-of-band interference. The HGA was mounted on a gimballed platform on the top deck of the lander, adjacent to the lander’s X-band communications antenna and the NGLR payload on the same platform, as shown in Figure 2(d).

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 3

Measured HGA gain patterns in LuGRE receive bands. (a) L1/E1, (b) L5/E5a. Patterns are shown as gain (radial) over elevation (angular) at 10° azimuthal cuts (individual traces).

3.1.3 QN400-SPACE GNSS Receiver

The heart of the LuGRE payload was Qascom’s multi-constellation and multi-frequency QN400-SPACE GNSS receiver. The receiver hardware and software were informed by designs previously tested on several flights such as the UP Aerospace SL-14 and SL-15 sounding rockets (Guzzi et al., 2025) and the Bobcat-1 CubeSat (Croissant et al., 2021) and were further refined for high-sensitivity acquisition in deep space (Fantinato et al., 2022; Pulliero et al., 2023). The resulting receiver properties flown for the LuGRE mission are summarized in Table 2.

View this table:

• View inline
• View popup

TABLE 2 Summary of LuGRE Receiver Design Properties

The receiver was designed in a dual cold-redundant configuration managed by an integrated supervisory board. The supervisory board controlled the selection of two identical GNSS receiver modules (RXA as the primary and RXB as the secondary), providing robustness against single-event-latchup events as well as monitoring and mitigation of potential over-current scenarios. The receiver contained a Xilinx Zynq 7000 SoC baseband processor and an Analog Devices AD9361 front-end to capture and digitally process radio frequency (RF) signals. Additionally, software-defined radio technology allowed for software re-configurability and a high degree of signal processing flexibility. This included the ability to configure receiver settings and perform full or partial software updates through telecommands generated and uploaded from the ground.

The high-sensitivity GNSS performance of the receiver was based on Qascom’s proprietary acquisition and tracking logic specifically designed for the mission, combining coherent and non-coherent accumulation strategies (Fantinato et al., 2025; Tedesco et al., 2023). Moreover, the receiver used a warm-start approach, exploiting aiding data uploaded from the ground and received by the lander. The real-time processing (RTP) mode produced instantaneous real-time PVT navigation solutions (when a sufficient number of signals was available, referred to as NAV telemetry) via a weighted least-squares estimator initialized with a priori predicted lander trajectory information. Similarly, navigation filter solutions (referred to as POD telemetry) were obtained using a custom extended Kalman filter implementation. An additional receiver mode, SC mode, allowed for capture of snapshots of raw signals as I/Q samples (IQSs) for post-processing on the ground, referred to as IQS telemetry. The parameters of the captured IQS packets were configurable via telecommand.

The receiver architecture distributed a frequency reference from an integrated internal oscillator to the FEA and field programmable gate array. The oscillator used was a commercial-grade voltage-controlled temperature-compensated crystal oscillator (VCTCXO), with stability properties measured at ambient conditions shown in Figure 4. Notably, the oscillator was observed to exhibit significant transient clock drift behavior upon power-on. An example of the receiver clock drift start-up transient, typically lasting 10-15 min, is shown in Figure 5, as estimated by the receiver during a representative commissioning operation.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 4

Measured Allan deviation for the internal VCTCXO at ambient temperature

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 5

Example of start-up clock drift transient observed in-flight, estimated as part of the receiver PVT solutions

3.2 Ground System and Operations

The LuGRE project involved the development of both flight and ground segments. The LuGRE ground system was designed as a geographically distributed architecture involving NASA GSFC in Greenbelt, Maryland, USA; ASI in Rome, Italy; Qascom in Bassano del Grappa, Italy; and Politecnico di Torino in Turin, Italy. The components of the ground system and its external interfaces are shown in Figure 6 and are described in the following paragraphs.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 6

LuGRE ground system components (solid lines) and external interfaces (dashed lines)

Payload Operations Center (POC) (NASA GSFC) was the LuGRE payload operational authority and was responsible for monitoring the LuGRE payload, ensuring its safety, and delivering telecommands to the Mission Control Center (MCC). It also provided the point of contact for Firefly and NASA CLPS. The POC processed Firefly predictive data provided by the MCC as detailed in the “Aiding” paragraph below, performed the final step of the telecommand verification process, and led the payload operations planning cycle. The POC contained the Science Processing Center, which led scientific activities during operations and continued post-mission for data analysis and public product generation.

Remote Troubleshooting Center (RTC) (Qascom) was the LuGRE payload technical authority. The RTC generated, tested, and verified the telecommands to be provided to the receiver and processed downlinked receiver telemetry to verify consistency and to evaluate performance. The RTC developed all major software used for command validation simulation and troubleshooting and to support the operational activities through all phases. It also acted as a redundant control center for the POC, implementing all of the critical functions needed to support payload operations if needed.

Remote Science Processing Center (RSPC) (Politecnico di Torino (PoliTo), Department of Electronics and Telecommunications) was responsible for performing quick-look scientific evaluations of the payload telemetry, supporting the science team in the POC with payload telemetry analysis, and supporting the payload operations planning cycle through analysis to maximize the performance of upcoming operations. The RSPC also developed the LuGRE quick Navigation Analysis and Reporting Tool (LuNART-q) to deliver early assessments of the scientific integrity of received payload data to the POC via an Operation Summary Report and to evaluate POC-provided simulations via a Prediction Evaluation Report (Minetto et al., 2025).

Firefly Mission Control Center (MCC) (Cedar Park, Texas, USA) operated the BGM1 lander and was the primary external interface for LuGRE payload operations. The MCC was responsible for delivering payload commands to the lander for execution, returning payload and lander telemetry, managing the overall operation schedule, and providing voice communications between LuGRE and the BGM1 operators.

The following main activities were performed before, during, and after each operation, as illustrated in Figure 7:

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 7

Flow diagram of LuGRE operations

Colored arrows indicate the flow of preparatory data (black), commands (blue), and telemetry (red).

Planning The operation planning process was active in the days prior to each operation. LuGRE and Firefly utilized a baseline operation schedule developed pre-launch as the basis for scheduling the payload activities. As the flight developed, Firefly readily accommodated changes to the baseline plan as needed to respond to lander-level activities, e.g., schedule keep-outs, and to improve scientific value to LuGRE within the constraints of the overall mission-level operations. The LuGRE ground system retrieved GNSS broadcast ephemeris information from the Crustal Dynamics Data Information System (CDDIS) for GPS and the Time and Geodetic Validation Facility (TGVF) for Galileo.

To support the scientific side of the planning process, the LuGRE team simulated upcoming operations using the Goddard Enhanced Onboard Navigation System (GEONS) and its GEONS Ground Matlab Simulation (GGMS) front-end to predict the availability and signal strength (C/N₀). GEONS has an extensive history of use by multiple missions as an in-flight navigation filter and via GGMS as a ground tool for visibility simulation and navigation analysis (Ashman et al., 2024; Winternitz et al., 2019). Starting several days in advance, the POC produced simulations of signal availability and C/N₀ for upcoming operations. At daily planning meetings, the team selected the operation start time and the start times and durations of the SC and RTP modes to maximize (i) the likelihood of acquisition of strong L1/E1 signals early in the window, (ii) the total data volume and/ or likelihood of tracking multiple unique signals, and (iii) the signal content of IQS batches. These selections were then coordinated with Firefly for scheduling.

Aiding The LuGRE receiver was used in warm start, exploiting three types of aiding information at start-up: time (provided by the lander within approximately 100 ms of Coordinated Universal Time [UTC]), GNSS broadcast ephemerides, and POD filter aiding information including (i) predictive trajectory, (ii) Earth orientation parameters, and (iii) lander physical parameters. Firefly delivered the predicted trajectory and lander information at 24 h prior to each operation, and the POC processed this information into an aiding input delivery to the RTC for inclusion in payload commands. The broadcast ephemerides, from CDDIS for GPS and TGVF for Galileo, were pulled within 2 h of the operation start time to maximize the accuracy of the ephemeris data.

Command Generation and Validation After delivery of the aiding data from the POC, the RTC validated the data using internal tools and then generated receiver telecommand files consisting of the aiding information and the receiver configuration necessary to collect the desired scientific information. The RTC validated the generated telecommands in a testbed using a Spirent GNSS simulator and the LuGRE flight-spare receiver unit to verify accuracy and consistency with respect to the desired operation. Once validated, the RTC delivered the telecommand files to the POC for a final verification and validity check using a Firefly-provided lander interface simulator. Once finalized, the POC delivered the command files and associated execution commands to the MCC, before a Firefly-dictated cutoff time. For long-duration operations during the surface phase, updated broadcast ephemeris commands were additionally generated every 2 h and provided to Firefly for upload at the next opportunity.

Execution In the minutes prior to an operation, the MCC uplinked the payload command files to the lander for staging and commanded the lander to slew (in transit) or gimbal (during surface operations) the antenna platform to an attitude that pointed the LuGRE HGA to the Earth. During the execution of the operation, the MCC downlinked telemetry from the lander for retrieval by the POC and RTC. During each operation, the POC monitored the status in real time using a Firefly-provided telemetry dashboard showing health and status metrics (e.g., current, temperature, data received), whereas the payload telemetry was ingested into a custom Grafana dashboard developed for the mission and accessible to all LuGRE ground sites.

Quick-Look Science After completion of the operation, the RSPC used the purpose-built LuNART-q quick-look science tool to perform a preliminary evaluation of the receiver telemetry, including C/N₀, GNSS observables, PVT and precise orbit determination (POD) solutions, and evaluation of any IQS collections. The RSPC augmented LuNART-q with other tools to perform deeper analysis and to assist with the evaluation of predictions and specific observations as needed. The RTC independently produced a technical evaluation report of receiver status and performance; both evaluations were presented at the daily planning meeting to evaluate the performance of the prior operation and to assist in the planning of upcoming operations.

This flow of activities was necessarily flexible, with exceptions and additional iterations occurring as lander events required, particularly during longer-duration operations on the lunar surface.

4.1 Characteristics of the Operational Environment Relevant to a GNSS User

Several properties of the environment in which LuGRE operated necessarily influenced either the receiver’s ability to acquire and track signals or the characteristics of the resulting observables and state estimates. This section discusses two of the most influential factors: (1) the highly dynamic thermal environment associated with cislunar space and the lunar surface (a fact of physics) and (2) the capability of the lander to achieve precise Earth-pointing for signal reception (an operational service provided by a host vehicle). Both of these factors, being generally outside of the direct control of the LuGRE team, may reasonably be expected to also impact most other lunar GNSS users and are therefore informative to highlight for contextualizing the data collected by LuGRE as well as for planning future missions.

4.1.1 Temperature

The BGM1 mission experienced a wide range of thermal conditions, which are important to understand because of their impact on the LuGRE payload, especially the receiver clock. Figure 10 illustrates an example of the impact of fluctuating temperature on the clock drift estimated by the receiver during the first 24 h of surface operations. The strong correlation between the clock stability and the temperature experienced by the hardware is immediately clear.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 10

Receiver temperature and resulting clock drift behavior during early surface operations (as reported in the receiver POD telemetry)

The overall rising trend is due to increasing solar illumination, with smaller fluctuations due to the thermal output of other hardware components on the lander. It is clear that these thermal phenomena had direct impacts on the clock drift of the LuGRE receiver.

Firefly placed three LuGRE-specific temperature sensors to monitor thermal conditions for payload safety. The sensors for the receiver and LNA were placed adjacent to these components, whereas the sensor for the HGA was placed on the back side of the antenna panel, centered on the HGA. Therefore, the values reported by these sensors, while not necessarily the precise absolute temperatures experienced by each component, offer a proxy for the relative levels and overall trends experienced by the hardware.

Figure 11 shows the three temperature profiles reported by these sensors during the transit phase. The temperature ranges of the receiver and LNA are similar owing to their common location within the upper chamber of the lander and were generally stable around the expected nominal temperature of 20 °C. The HGA was typically hotter in the Earth-phasing loops and colder while in lunar orbit as compared with the receiver and LNA because of the radiative heating and cooling in its exposed position on the lander’s pointing platform.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 11

Temperatures observed during transit operations for the receiver, LNA, and HGA hardware

The mean value for each operation is depicted by a circle, with ranges above and below depicted by arrows.

Figure 12 shows the temperature profiles reported by the same sensors during the surface phase. This phase was highly thermally dynamic, correlating closely with solar illumination as well as the thermal output of other components on the lander, particularly those closest to the LuGRE receiver and LNA. The temperature of all LuGRE components rose steadily throughout the first three days of surface operations until just before 00:00 UTC on March 5, at which point LuGRE was powered off during OP40_0. During the power-off period (not shown in this figure), temperatures continued to rise. The temperatures of the receiver and LNA were primarily influenced by the thermal output of other components of the lander, peaking at 72 °C on March 7 and 68 °C on March 6, respectively. The sun-exposed HGA was the hottest component by far, ultimately peaking at 115 °C on March 10. Peak temperatures were all within yellow limits for the hardware, meaning that components were expected to survive with no significant damage as long as they remained powered off. During OP73_0, temperatures were briefly low, owing to the preceding eclipse, but quickly rose back to pre-eclipse levels, similar to those observed at the beginning of OP38_0, before declining steadily as the sun set. The final operations spanning lunar sunset on March 16 experienced a swift temperature decrease of 35 °C over a 5-h span for the internal components, with a greater decrease observed for the HGA.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 12

Temperatures observed during surface operations for the receiver, LNA, and HGA hardware

Temperatures during the power-off period between March 5 and March 14, 2025, are not shown.

4.1.2 Antenna Pointing

Receive antennas for high-altitude GNSS users are typically designed with relatively high gain compared with terrestrial users, often resulting in consequently narrow gain patterns; for example, see the LuGRE HGA gain patterns in Figure 3. Because of this design type, small errors in pointing the receive antenna toward the Earth can result in increasingly large losses in the received C/N₀. LuGRE required that the BGM1 lander point the LuGRE HGA to within 1° of the Earth’s center for the duration of all data collection to guarantee visibility of GNSS signals within the peak gain region of the HGA gain pattern without significant gain roll-off due to mis-pointing. This requirement was met for all operations, with a margin. The statistics of the actual achieved off-pointing error are shown in Figure 13. Off-pointing was calculated using the Firefly-provided definitive lander attitude for transit and antenna gimbal angles for the surface phase.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 13

Achieved Earth-center off-pointing angle statistics during the transit and surface phases of the mission. (a) Transit off-pointing: Probability distribution, (b) Surface off-pointing: Probability distribution.

Throughout transit phase operations, an off-pointing angle of less than 0.73° from the Earth’s center was achieved by the lander, with a median value of 0.35°. While on the lunar surface, the achieved off-pointing accuracy was better than 0.21° from the Earth’s center, with a median value of 0.05°. The difference in accuracy between the phases is likely due to the methods used by the lander to obtain Earth-pointing attitude; because the antenna panel was stowed during transit, antenna pointing could only be achieved with a full vehicle re-orientation, as opposed to finer gimbal control while on the surface. As discussed in Section 5.1.2, the as-flown lander attitude was incorporated into signal visibility simulations so that the precise receive gain can be accounted for in the link model, considering any off-pointing.

5.1 Received Signal Strength

The GNSS received signal strength was analyzed in terms of C/N₀ estimates collected at a rate of 1 Hz during RTP mode operations of the LuGRE payload. The C/N₀ values were estimated at the receiver tracking stage via pilot channels, i.e., GPS L1 C/A and L5Q and Galileo E1C and E5a-Q; therefore, this analysis reflects the signals that the LuGRE payload was actually capable of tracking during the assigned operations (Parker et al., 2022).

5.1.1 Results and Characteristics

Aggregating the statistics of the observed C/N₀ as histograms of normalized C/N₀ occurrences enables a distinction between signal contributions originating from the antenna main- and side-lobes of the GNSS satellites. Unlike terrestrial observations, where only main-lobe signals are tracked, this representation provides key insights into the unique signal environment experienced by the LuGRE payload in cislunar space and on the lunar surface.

Representative results are shown in Figure 14 to familiarize the reader with the different tracking scenarios encountered throughout the mission. On the left side of each subfigure, a space skyplot characterizes the relative geometry between the spacecraft and the tracked GNSS satellites during the operation, assuming a nadir-pointing attitude (i.e., the LuGRE HGA is considered to be perfectly aligned with Earth’s center). The center-of-mass positions and velocities of the GNSS satellites—originally broadcast in the constellation-specific Earth-fixed rotating frame—are transformed into the spacecraft’s local-vertical, local-horizontal frame defined by the Firefly BGM1 definitive ephemerides, capturing their apparent motion from the receiver’s perspective. The Earth’s disk and surrounding ionosphere layer are overlaid as a shaded region. Each satellite trajectory is plotted as a gray solid line, overlaid with a heatmap that encodes the C/N₀ values for each tracked signal component, separately for the L1/E1 (top diagram) and L5/E5a (bottom diagram) frequency bands. Such an augmented visualization allows for the identification of regions in which main-lobe and side-lobe tracking occurs. Stronger signals typically cluster near the Earth’s limb, suggesting main-lobe reception, whereas weaker signals are observed at lower elevations, often associated with side-lobe contributions. To complement the geometric and time series perspectives, C/N₀ values are statistically aggregated from the time series and presented as histograms of normalized occurrences on the right side of each subfigure, providing a quantitative view of the signal strength distribution across the LuGRE receive bands.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 14

Space skyplots (left) and corresponding C/N₀ time series with normalized histograms (right) for three representative LuGRE operations from different mission phases. (a) OP1_0: Space skyplot; L1 C/A/E1C (top) and L5/E5a-Q (bottom). (b) OP1_0: C/N₀ time series and histogram for 15-RE altitude. The group of tracked signals below approximately 33 dB-Hz arises from transmitter side-lobes. (c) OP17_0: Space skyplot; L1 C/A/E1C (top) and L5/E5a-Q (bottom). (d) OP17_0: C/N₀ time series and histogram for 44.9-RE altitude. The group of tracked signals below approximately 27 dB-Hz are from transmitter side-lobes. (e) Extract of OP38_0 (initial 72 min): Space skyplot; L1 C/A/E1C (top) and L5/E5a-Q (bottom). (f) OP38_0 (initial 72 min): C/N₀ time series and histogram at the lunar surface (55.8-RE altitude). All tracked signals shown are from transmitter main-lobes.

Histograms are grouped by GNSS signal component. The C/N₀ time series legend refers to RINEX observation codes.

The first effect observed in the time series of Figure 14(b) is the presence of specific satellite signal components whose C/N₀ estimates exhibit high variance, particularly at the beginning of their tracking window (e.g., PRN 15 on E1, PRN 8 on both E1 and E5a, PRN 32 on both L1 and L5). This behavior can be interpreted by jointly analyzing the corresponding skyplot in Figure 14(a), which shows that these fluctuations are associated with satellites whose trajectories graze the Earth’s limb and intersect the ionospheric layer. In particular, these components are often observed in transition phases, where GNSS signals may still reach the receiver even when the direct path is geometrically obstructed, owing to refraction in the ionosphere. The oblique propagation path introduces additional ionospheric delay and potential phase distortion, while spatial inhomogeneities along the signal path give rise to scattering. These effects collectively contribute to increased variability in the estimated C/N₀ values due to unstable signal tracking.

Another key observation characterizing this low-altitude LuGRE operation (see Table 8) is the emergence of a threshold—at approximately 33 dB-Hz—that serves as a qualitative boundary between main-lobe and side-lobe tracking. This distinction is evident both in the skyplot, where side-lobe signals are predominantly concentrated at lower elevation angles, and in the multimodal C/N₀ distributions, which indicate that a substantial portion of the occurrences are below 33 dB-Hz.

In contrast, Figures 14(c) and 14(d) illustrate primarily main-lobe tracking. For example, both PRN 6 on L1/L5 and PRN 34 on E1/E5a are tracked down to the receiver’s sensitivity threshold while remaining within the main-lobe regime. Conversely, PRN 15 on E1/E5a is initially tracked at low C/N₀, with the signal strength gradually increasing until signals become occluded by the Earth. On the other hand, PRNs 13 and 20 on L1 appear to be side-lobe receptions occurring above the receiver’s sensitivity threshold, as also confirmed by the satellites’ positions in the polar diagram.

Lastly, Figures 14(e) and 14(f), which correspond to the initial portion of the first surface operation, show a reduced number of tracked signals, all attributed to main-lobe reception. Among them, PRNs 2 and 34 on E1 are initially tracked closer to the receiver sensitivity limit—approximately 5 dB lower than their E5a counterparts—and their signal strength progressively increases as the satellite track moves toward the Earth’s limb. Conversely, GPS channels start being tracked at higher C/N₀ and fade toward the sensitivity threshold as Earth’s occultation occurs.

The trends shown in Figures 14(a)-14(f) are further confirmed by the statistical patterns in the corresponding C/N₀ histograms. High-intensity modes typically reflect strong main-lobe contributions (e.g., Figure 14(b)), whereas lower modes are associated with side-lobe reception (e.g., Figure 14(d)). In some cases, long tails extending from high or near-high modes indicate that main-lobe signals were tracked down to the receiver’s sensitivity threshold (e.g., Figure 14(f)), whereas side-lobe signals more often fall below trackable levels.

Figures 15 and 16 show the normalized occurrences of the C/N₀ estimates throughout all transit and surface operations for GPS and Galileo signals, respectively. The histograms, sorted by initial operation altitude, present the normalized distribution of C/N₀ estimates obtained during each operation and collected by 2-dB bins in the interval of 18-48 dB-Hz. The y-axis indicates the C/N₀ dynamic range whereas the x-axis shows the percentage of occurrences included in each bin. As the number of C/N₀ estimates differs for each operation, the representativeness of the corresponding histograms also varies. For this reason, the total number of estimates is reported at the bottom of each histogram as an indicator of statistical reliability. Different colors are used to distinguish mission phases.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 15

Normalized C/N₀ occurrences for GPS signals throughout mission operations. (a) GPS L1 C/A, (b) GPS L5Q.

Operation IDs and associated altitudes are reported at the top of each histogram. The total number of C/N₀ estimates is reported at the bottom of each subplot. Operations are sorted from lowest altitude on the left to highest on the right.

As shown in Figures 15 and 16, the altitude and, consequently, the increase in free space path loss produce a decreasing C/N₀ trend and a reduction in the number of visible side-lobes. Note that the observed variability of C/N₀ ranges depends not only on altitude but also on mission phase. For instance, although OP27_0 is performed at a similar altitude as OP78_1, the visibility conditions and dynamics during the low lunar orbit (LLO) phase yielded different behavior in the C/N₀ distribution.

5.1.2 Analysis of C/N₀ Levels

The C/N₀ results shown in Figures 15 and 16 are significantly lower than the levels predicted both by the LuGRE GEONS planning simulation (see Section 3.2) and in the literature for systems with similar properties (see, e.g., the work by Hassouneh et al. (2023)). See Figure 17 for an example of the discrepancy observed during a lunar surface operation. To quantify and characterize this result, the LuGRE team further developed its planning simulation into an analysis tool. This section details the configuration and resulting C/N₀ discrepancy characterization.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 16

Normalized C/N₀ occurrences for Galileo signals throughout mission operations. (a) Galileo E1C, (b) Galileo E5a-Q.

Operations IDs and associated altitudes are reported at the top of each histogram. The total number of C/N₀ estimates are reported at the bottom of each subplot. Operations are sorted from lowest altitude on the left to highest on the right.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 17

Example of C/A C/N₀ comparison between (a) simulation and (b) LuGRE flight data (lunar surface case)

Colors distinguish between different PRNs belonging to GPS blocks IIR-M and IIIA.

GEONS and its GGMS front-end simulate C/N₀ using the model described in Section 12.5.4 of the GEONS Mathematical Specification (Long & Stacey, 2025). The inputs to this model in the simulations implemented by the LuGRE team are described in Table 3; link budget parameters specific to the LuGRE payload are shown in Table 4. Any additional terms or inaccuracies in the values shown in these tables, causing either lower-than-expected power (i.e., C) or higher-than-expected noise (i.e., N₀), are unknown to the simulation and are therefore contained in the C/N₀ discrepancy observed in-flight.

View this table:

• View inline
• View popup

TABLE 3 C/N₀ Simulation Parameters

View this table:

• View inline
• View popup

TABLE 4 LuGRE Link Budget Parameters

For the specific purpose of simulating C/N₀ to quantify the observed in-flight discrepancy, the analysis focused on only GPS L1 signals from blocks IIR-M and IIIA, as these blocks have well-documented transmit antenna gain patterns and transmit powers that can be validated using multiple sources. The relevant transmit antenna gain patterns are publicly available. The transmit power was estimated using public products from the Cyclone Global Navigation Satellite System (CYGNSS) mission over the timespan of the LuGRE flight data (CYGNSS, 2024). CYGNSS data provide the measured effective isotropic radiated power (EIRP), from which the transmit power can be derived by removing the transmit antenna gain at the elevation angle of the measurement (Wang et al., 2019). These values have also been corroborated within 0.5 dB by work analyzing the total received EIRP measured by ground receivers (Steigenberger et al., 2024; Steigenberger et al., 2018).

The validity of the simulation was evaluated by using it to simulate C/N₀ as observed by the MMS. By applying MMS link budget parameters (Winternitz et al., 2019) and comparing the results to MMS C/N₀ flight data over a period in 2017 at an altitude of 12 RE, we obtained the comparison shown in Figure 18 (Winternitz et al., 2017). The results are highly consistent, with an overall difference that is zero-mean with variation on the order of 1-2 dB, validating the configuration of the underlying models.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 18

Comparison between GGMS simulation results and MMS flight data. (a) Direct C/N₀ arc comparison between MMS flight data (dots) and simulation (circles), (b) ∆C/N₀ (MMS flight minus simulation).

The focus is on GPS L1 C/A signals from block IIR-M. Colors distinguish between different PRNs belonging to the considered block.

The validated simulation was then used to predict the LuGRE C/N₀, focusing on the well-known GPS L1 signal from block IIR-M and block IIIA satellites. Comparing LuGRE flight data against this simulation shows that the C/N₀ estimated in-flight is generally 7–12 dB lower than expected. Furthermore, this discrepancy is present across the full LuGRE data set, including at lower altitudes covered by the MMS validation data, and does not appear to correlate with external factors such as temperature, altitude, or mission phase (i.e., transit or surface; see Figure 19).

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 19

C/N₀ difference from simulation (dB, flight minus simulation) for GPS L1 blocks IIR-M and IIIA, summarized by operation

Operations are sorted from lowest altitude on the left to highest on the right; see Table 8 for details. Operations numbered OP38_0 and higher are surface operations. The plot is defined as in the MATLAB boxplot function. The box width is proportional to the square root of the number of data points used.

The root causes of this discrepancy are under investigation. However, the lack of clear correlation to a coincident property such as altitude or temperature and the fact that the discrepancy is not present in the MMS data at a similar altitude suggest causes within the overall integrated lander/payload flight system, not systemic transmission or environmental factors. In other words, the lower-than-expected C/N₀ observed by LuGRE is a peculiarity of this mission and should not be interpreted as an inherent characteristic of lunar-vicinity GNSS reception in general.

It is important to note that the in-flight results described elsewhere in this section are necessarily influenced by the observed C/N₀ discrepancy. The most obvious effect is on signal availability, as the overall lower C/N₀ level results in the majority of the GNSS side-lobe signals falling below the receiver acquisition threshold, as shown in Figure 17. Prior work has consistently shown that side-lobes are expected to be abundant at C/N₀ levels that can be reliably acquired and tracked at lunar distance (Ashman et al., 2024; Delépaut et al., 2020). Likewise, the properties of the GNSS observables are subject to thermal noise theory, such that the pseudorange, Doppler shift, and carrier-phase measurements collected by LuGRE inevitably have additional noise beyond what would otherwise be expected at higher C/N₀ levels (Kaplan & Hegarty, 2017).

5.2 Measurement Quality

To assess the quality of the raw GNSS measurements collected during LuGRE, the standard deviation of the pseudorange (σ^τ) and carrier-phase (σ^ϕ) observables were computed by detrending the data using third-order finite differences with fourth-order accuracy. This approach allowed for the isolation of short-term fluctuations that are primarily attributable to receiver noise without long-term variations induced by the dynamics of the scenario, enabling a direct assessment of the intrinsic precision of the measurements for the different signal components across the mission phases.

Note that during a single operational window, the same signal component can be acquired and tracked multiple times with different quality levels owing to potentially different propagation conditions, especially for satellites tracked right before being occluded by the Earth or just rising from an occlusion window, compared with time intervals during which they are in full visibility. For example, Figure 20 shows the behavior of the third-order finite difference computed over the time series of code pseudorange measurements for Galileo PRN 15 (E15 E1 OS) in two distinct timespans, characterized by different standard deviations. The code measurements were not corrected for the estimated receiver clock bias before computation of the finite differences. Additionally, the C/N₀ value may vary during each operational window, as highlighted in Figure 14(b). For these reasons, the detrended time series for each satellite signal component (i.e., for each active receiver channel) was analyzed by identifying multiple continuous, uninterrupted intervals of tracking for that channel. To account for this variability, the standard deviation was estimated separately for each tracking interval and then combined using a weighted average, with weights equal to the ratio between the number of samples in each partial interval and the total number of non-null samples of the time series. For each LuGRE operation, this methodology yields a unique value per satellite signal component. Although this method enables comparability across channels, the statistical precision may vary, as channels tracked more extensively and under more stable conditions are statistically more significant.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 20

Examples of third-order finite difference of code pseudorange measurements from OP1_0

Figure 21 presents the estimated standard deviation of code pseudorange (Figure 21(a)) and carrier-phase (Figure 21(b)) measurements across LuGRE operations, grouped by GNSS signal type (i.e., GPS L1 C/A, GPS L5, Galileo E1 OS, Galileo E5a). The analysis spans all mission phases, from flight to surface operations (see Table 8), with results sorted by their initial altitude. The standard deviation is calculated over the full operation span for each tracked satellite, and correlation to C/N₀ or other factors is left for future work.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 21

Estimated standard deviation of GNSS code and carrier-phase measurement noise across LuGRE operations, grouped by signal component. (a) Code measurements, (b) Carrier-phase measurements.

Each marker corresponds to a tracked signal component from a specific satellite. Operations are ordered according to the spacecraft’s altitude at the start of each operation. Operations are sorted from lowest altitude on the left to highest on the right.

The results show good quality of the receiver measurements in all mission phases, comparable to what is typically achieved on Earth for the considered front-end bandwidth (Ribot et al., 2016). As expected, code measurements exhibit estimated standard deviations approximately two orders of magnitude higher than those of carrier-phase measurements, consistently reflecting the well-known difference in precision between code and phase tracking. Beyond this general trend, the results highlight meaningful differences across frequency bands and signal modulations. In particular, channels in the L5/E5a band consistently show lower code measurement noise than their L1/E1 counterparts. This behavior is due to a combination of factors: wider signal bandwidth, lower path loss, and higher EIRP such as in the case of Galileo E5a compared with E1 OS (Menzione et al., 2024). These characteristics typically result in higher observed C/N₀ levels (see Figure 16), improving the stability of code tracking loops and reducing thermal-noise-induced jitter, which represents the dominant source of residual noise on code measurements (Won & Pany, 2017). Even within the same bandwidth, differences are visible. For example, Galileo E1 OS observables exhibit lower noise than GPS L1 C/A, potentially because of the composite binary offset carrier modulation, which theoretically offers better performance. This modulation narrows the main-lobe of the autocorrelation function and improves the sensitivity of the code discriminator to small timing errors (i.e., a steeper discriminator slope around the prompt correlation), thus improving the receiver’s ability to suppress thermal-noise-induced jitter in the code tracking loop (Betz, 2015; Betz & Kolodziejski, 2000).

In contrast, carrier-phase measurements show less overall variation across signal modulations. Because the precision of phase tracking is also influenced by the signal wavelength, L5/E5a signals—despite their favorable C/N₀—do not always yield reduced noise variance compared with their higher-frequency counterparts. Although most satellite signal components fall within a consistent range of estimated standard deviation values, a few outliers are clearly visible in specific operations. These outliers are the result of signal degradations not yet sufficient to trigger a loss of lock but capable of impairing the loop stability. These degradations have a significant impact on the statistics owing to the short durations of such tracking intervals. In general, these results, including the relative behavior of different modulations, align well with theoretical expectations based on the Cramér–Rao bound for the maximum likelihood estimation of code delay and carrier-phase offset (Das et al., 2020; Medina et al., 2020).

5.3 Signal Availability

Signal availability in a GNSS generally refers to whether the GNSS signals of interest are present, trackable, and usable at a given receiver location and time (Betz, 2015). Signal availability is impacted by geometric and radiometric visibility, which, in turn, depends on receiver sensitivity. The following analysis additionally applies a further condition: that the GNSS signals are strong enough to be acquired and tracked above the receiver’s minimum C/N₀ threshold.

PVT availability extends beyond just having signals and is more restrictive, referring to whether the available signals and associated measurements (pseudorange, carrier phase, Doppler shift, etc.) are sufficient to solve for a multilateration solution with no a priori information of the receiver state (Kaplan & Hegarty, 2017). Thus, PVT availability is highly dependent on the specific time frame at which the operation took place, and even operations at similar altitudes can have different relative PVT availabilities. The following analysis provides an indication of both signal and PVT availability by counting distinct satellite signals at each epoch in RTP mode.

Figure 22 shows bar charts depicting overall GNSS signal availability throughout transit and surface operations, sorted according to the initial altitude of each operation. Each bar corresponds to an RTP operation and is color-segmented to represent the percentage of time, relative to the nominal duration of each RTP operation, during which exactly n signals from as many distinct n GNSS satellites were simultaneously tracked. To highlight PVT availability, cool-toned colors denote periods with n ≥ 4 distinct tracked signals, indicating sufficient signal availability to compute a PVT fix through a well-determined least-squares algorithm, while warm tones highlight degraded visibility conditions (n < 4), where standalone positioning would not be feasible unless underdetermined multilateration problems are addressed leveraging some a priori knowledge about the receiver state (Vouch et al., 2025).

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 22

Multi-constellation, multi-frequency availability: Percentage of time, relative to the nominal duration of each RTP operation, during which n signals from as many distinct GNSS satellites were simultaneously tracked

Results are shown as a function of operation ID and associated altitude in RE. Percentage values are reported for n ≥ 4.

A higher saturation of cool colors in a bar, such as in the case of OP1_0 and OP2_0, represents abundant signal availability throughout the operations. A few operations such as OP12_0 and OP37_0 were instead impacted by scarce signal availability due to poor geometric visibility of satellites.

The overall availability throughout the mission displayed in Figure 22 illustrates all tracked signals, for each frequency and constellation, reflecting the receiver’s full tracking capability during the mission, with only one signal per satellite vehicle considered. Data have also been aggregated by constellation, as shown in Figure 23, where Figure 23(a) details GPS-only signals (L1/CA and L5 in-phase) and Figure 23(b) presents Galileo-only signals (E1 and E5a). A comparison of these results against Figure 22 clearly shows the expected benefit of using a multi-constellation receiver; this analysis shows how a single-constellation receiver would have limited opportunities to achieve a PVT fix, even during operations relatively closer to the Earth. Despite tracking multiple frequencies within each constellation, the number of unique satellites is limited by geometry, thus reducing overall availability.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 23

Single-constellation, multi-frequency availability: Percentage of time, relative to the nominal duration of each RTP operation, during which n signals from as many distinct GNSS satellites were simultaneously tracked. (a) GPS L1+L5, (b) Galileo E1+E5a.

Results are shown as a function of operation ID and associated altitude in RE.

Results are also shown by frequency band in Figure 24, where Figures 24(a) and 24(b) report signal availability over the L1/E1 and L5/E5a bands, respectively, aggregating the GPS and Galileo satellites. Table 5 summarizes the benefit of leveraging a multi-frequency receiver in the LuGRE mission by comparing the availability achievable using the L1/E1 signals only of Figure 24(a), compared with the full operational configuration illustrated in Figure 22. Although adding the second frequency does not change the geometry of the problem or the number of independent signals, the higher L5/E5a power and the wider beam of the transmit antenna pattern (Anthony, 2022a, 2022b; European Union, 2023; Menzione et al., 2024) increase the availability of signals. This trend is also reflected in the cumulative amount of data collected for each frequency, as shown in Figure 9.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 24

Multi-constellation, single-frequency availability: Percentage of time, relative to the nominal duration of each RTP operation, during which n signals from as many distinct GNSS satellites were simultaneously tracked. (a) GPS/Galileo L1/E1, (b) GPS/Galileo L5/E5a.

Results are shown as a function of operation ID and associated altitude in RE.

5.4 Receiver State Estimation

The LuGRE receiver featured both a POD extended Kalman filter and a weighted least-squares PVT algorithm for navigation, as described in Section 3.1. This section focuses on the PVT solutions; the characteristics of these results are less implementation-specific than sequential navigation filter solutions and thus extend more generally to future lunar users. Navigation filter analysis is left for future work. The LuGRE receiver produced instantaneous real-time PVT solutions during periods in which simultaneous visibility of signals was achieved for four or more unique satellites. Overall, LuGRE calculated PVT solutions in both the transit and surface phases of the mission at altitudes ranging from 15 RE to 62 RE from the Earth’s surface.

Table 9 in Appendix A provides a catalog of PVT windows throughout the LuGRE data set. Each row represents a contiguous time window in which a PVT solution was calculated by the receiver for a total time period comprised of at least 30 individual epochs without any gaps greater than 30 continuous seconds. The column denoted as “% NAV” refers to the percentage of time within each window in which a navigation solution was obtained by the onboard receiver, accounting for brief periods in which visibility falls below four unique satellites.

In nearly all cases, PVT solutions were achieved via a combination of constellations and frequencies. The column in Table 9 denoted as “% Multi-Const.” refers to the percentage of time within each window in which at least one satellite from both GPS and Galileo was visible. In these cases, PVT availability was enabled by the GPS–Galileo time offset provided in the Galileo navigation message. Moreover, the overall signal availability was also impacted by the presence of the C/N₀ discrepancy discussed in Section 5.1.2; simulations suggest that without this discrepancy, the PVT availability would be greater.

As in any weak-signal GNSS application, the accuracy of PVT solutions is limited by the constellation/receiver geometry, as represented by the dilution of precision (DOP); Table 9 characterizes this specifically through the geometric DOP (GDOP). Over the LuGRE mission, the GDOP varied from a minimum below 50 in the lowest-altitude operation, to values averaging from 1,000 to 1,000,000 on the lunar surface, depending on the specific geometry.

Figures 25 and 26 show example error statistics from specific PVT windows, computed relative to the Firefly definitive trajectory characterized as in Table 3. Under poor geometric conditions, it is instructive to examine the range errors and root sum squared (RSS) lateral errors separately, as the range error becomes highly correlated with clock error and dominates the total solution accuracy at high altitudes (Konitzer et al., 2022).

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 25

OP1_0 (15.03 RE) onboard PVT range and RSS lateral state error from the Firefly definitive trajectory and GDOP value. (a) Range and lateral position error time series, (b) Range and lateral position error probability density, (c) Range and lateral velocity error time series, (d) Range and lateral velocity error probability density.

OP1_0 at 15-RE altitude, shown in Figure 25, is the most comparable to typical GNSS use, although the GDOP is at a steady-state value of approximately 50. Here, the PVT error is generally between 100 m and 1 km in the range component and just over 1 km in the lateral component. Similarly, the velocity is under 10 m/s in both the range and lateral directions.

Figure 26 shows lunar surface PVT errors from OP38_0 during two representative time windows with different GDOP environments. The window with PVT error in blue shows a more dramatic variation around an asymptotic discontinuity in the GDOP. Here, the range errors are 1–100 km in position, and the velocity error is especially high, on the order of 1000–10,000 m/s. Lateral errors are more controlled, under 1 km in position and mostly under 10 m/s in velocity. The cusps observed in the GDOP are caused by the evolving relative geometry of the satellites with respect to the payload antenna, causing gradual degradation of observability in the range direction. As the relative geometry evolves, the progressive alignment of the tracked satellites leads to a near-degeneracy of the observability matrix, whereas their subsequent separation restores observability. This phenomenon produces transients lasting up to a few minutes and is seen more commonly and dramatically at high altitudes as the satellite geometry tends to be more clustered.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 26

OP38_0 (lunar surface) PVT range and RSS lateral state error and GDOP within two different GDOP environments. (a) Range position error time series, (b) Range position error probability density, (c) Lateral position error time series, (d) Lateral position error probability density, (e) Range velocity error time series, (f) Range velocity error probability density, (g) Lateral velocity error time series, (h) Lateral velocity error probability density.

The first window (blue) spans a dynamically changing GDOP environment on the lunar surface, and the second window (black) spans a stable GDOP environment; note the different time scales for each case, which correspond to the duration of each PVT period.

By contrast, in another time window for the same operation spanning March 3, 2025, 11:22–11:30 UTC (shown in black), PVT solutions are stable during a nearly constant-GDOP timespan, with relatively lower steady-state PVT errors on the order of 0.1–1 km in position and 1–1000 m/s in velocity. The range error continues to dominate the overall solution error, as in OP1_0.

In general, these results show the feasibility of calculating GNSS least-squares navigation solutions even from the lunar surface, but reveal the inherent limitations to accuracy caused by the geometry of the solution at such a distance. Use for operational navigation should instead focus on sequential navigation solutions, e.g., calculated via a Kalman filter architecture. Furthermore, for high-precision navigation, a user must break the high degree of correlation between the range and clock error. As discussed by Winternitz et al. (2019), this can be done by either using a high-stability clock or including independent measurements that improve the observability of the range component.

5.5 Characterization of In-Phase/Quadrature Signal Batches

The unique LuGRE IQS data sets, collected on the Moon in both the L1/E1 and L5/E5a bands, provide an unprecedented record of the raw GNSS signal as received in the lunar environment, enabling high-fidelity post-processing for signal characterization, performance verification, environmental assessment, and testing of advanced acquisition and tracking algorithms. Despite their short duration owing to limited data volume and downlink capacity, these batches allow for a spectral analysis of GNSS signals received throughout mission operations.

To provide an overview of the LuGRE IQS data sets, Figure 27 shows the estimated power spectral density (PSD) under different payload configurations. The PSD is estimated using Welch’s method, aggregated by operations using similar SC configurations. The estimation used a 4096-point fast Fourier transform, a rectangular window, and an overlap of 128 samples between successive windows. Because of the low power of GNSS signals, their cumulative spectral density is barely distinguishable and decreases with altitude. The differences in the in-band density levels observed among the PSDs are primarily due to the automatic gain control, which maintains an approximately constant signal power at the input of the analog-to-digital converter.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 27

Aggregated PSD plots for L1/E1 and L5/E5a IQS batches captured with different payload configurations. (a) L1/E1 signal at 8 Msps and 4 bits; filter A, (b) L1/E1 signal at 8 Msps and 4 bits; filter B, (c) L5/E5a signal at 24 Msps and 4 bits.

Depending on the specific settings for the SC mode, different front-end filters resulted in different spectral behavior of the collected samples. In general, one filter (filter A) was used for dual-frequency SCs and for the OP78_0 single-frequency capture. Most single-frequency captures used filter B. As shown in Figure 27(a), captures performed with filter A exhibit pronounced ripples in frequency response induced by the anti-aliasing filter used prior to sample decimation. In fact, all rippling L1/E1 data sets were obtained by decimating 24-Msps SCs by a factor of three, in order to reduce storage requirements and downlink data volume. In contrast, filter B shows a comparatively flat and more uniform response, as these data sets were natively recorded at 8 Msps with no further decimation and no anti-aliasing filtering.

Figure 27 also reveals the presence of spurious signal power near the central frequency of each band, likely caused by EMI. Therefore, to assess the integrity of the received files as well as the potential impact on available GNSS signals due to EMI, post-processing of the received batches was performed. As an example, signal acquisition was performed in post-procesing using IQSs of GPS L1 for PRN 18 and PRN 24 during OP2_0, with 5 non-coherent sums, 8 ms of coherent integration time, a Doppler search space of ±20 kHz, and a Doppler step of 50 Hz. The acquisition was considered to be effective when comparison of the estimated C/N₀ and Doppler shift with values obtained through the onboard acquisition (ACQ) performed by the receiver in the subsequent RTP operation showed similar results. As shown in Table 6, the resulting values are considered reasonably close for the target PRNs, excluding the possibility of false acquisition.

Two cascading notch filters were then applied in post-processing to filter out the EMI components; a comparison between pre-filter and post-filter signal strength is shown in the “Filtered IQS” column of Table 6. The resulting C/N₀ increase of 0.09 dB and 0.16 dB without an impact on Doppler shift suggests a negligible impact of EMI during mission operations. Similar preliminary integrity checks were performed throughout the mission for different PRNs and constellations, with no significant issues reported. A deeper analysis is left for future work.

6.1 Signals

• As predicted by many prior studies and now demonstrated by LuGRE, open signals from GPS and Galileo in the L1 and L5 bands can be readily received, acquired, and tracked at lunar distance and on the Moon’s surface when in view of the Earth. This includes a robust quantity of main-lobe signals as well as a substantial quantity of side-lobe signals, which LuGRE was able to acquire only at their peaks, given the C/N₀ discrepancy observed in-flight. Side-lobe signals are especially important for (a) increasing the overall measurement quantity for navigation and (b) improving measurement continuity through gaps in main-lobe signals.

• The overall availability discussed in Section 5.3 benefits from signals originating from the side-lobes of the GNSS transmit antennas during both transit and surface operations. Figure 28 provides a representative example from surface operations during OP74_0 (61.90-RE altitude and above), during which several side-lobe signals were tracked for both systems and both frequencies. As noted in the figure, side-lobe signals were directly acquired and tracked from GPS PRN 1 (G1) and PRN 24 (G24) on L1 C/A and were tracked from Galileo PRN 27 (E27) on both E1C and E5a-Q. GPS L5 side-lobe signals were tracked at lower altitudes. These observations confirm that side-lobe signals can be effectively acquired and tracked at the lunar surface, improving overall availability.

• The signals that were acquired and tracked are of high quality, with code and phase measurement standard deviations consistent with that of measurements from Earth-vicinity receivers. In particular, the L5 and E5a signals show improved measurement quality over their L1 C/A and E1 OS counterparts. These results are true regardless of altitude or location, whether in transit or on the Moon’s surface.

• In general, the L5/E5a band offers substantial benefits in the lunar environment compared with the corresponding L1/E1 band. Improved C/N₀ and a broader main-lobe lead to increased visibility and acquisition opportunities, as demonstrated in Figure 29 for GPS PRN 6 and Galileo PRN 34 for OP17_0. Improved quality leads to a higher likelihood of successful acquisition, stable tracking, and positive contribution to the navigation solution.

• As highlighted by Delépaut et al. (2020), the specific geometry of each constellation affects its visibility from the Moon. LuGRE provided experimental evidence of this phenomenon. For Galileo, which features three highly inclined orbit planes, LuGRE observed a long-term trend in visibility as each plane aligned with the Moon line of sight; Figure 30 shows the Galileo tracked C/N₀ over a nearly 24-h period during OP38_0 on the lunar surface. In this case, main-lobe signals from Galileo PRNs 34, 15, 13, 26, 33, 10, and 36 are all tracked, but at progressively lower signal strength as their common orbit plane aligns with the LuGRE line of sight, eventually falling below the acquisition threshold of the receiver (as affected by the present C/N₀ discrepancy). Other Galileo signals in other orbit planes are not affected. GPS does not show this long-term behavior as markedly, as its satellites are distributed across six orbit planes. This result implies that GNSS signal visibility, especially for Galileo, will show a long-term oscillation, especially if primarily main-lobe signals are used.

• Interactions between GNSS signals and the ionosphere are well known, to the extent that these interactions form the basis for the field of GNSS radio occultation (Angling et al., 2021; Wu et al., 2024). These effects are visible from lunar distances as well and must be considered by lunar GNSS applications. Figure 31 shows an example of degraded C/N₀ measurements as the signals (in this case, from Galileo PRN 36) travel through the ionosphere at a grazing altitude as low as 50 km above the Earth’s surface. The observed C/N₀ shows a distinct signature, marked by pronounced fluctuations immediately after signal ingress and just before signal egress. Although this pattern remains recognizable in other operational contexts, it is most often interrupted by Earth occultation. An example of such a discontinuity is shown for Galileo PRN 34 (E34 E1 and E5a) in Figure 29, where both E1 and E5a signals were first tracked only after signal egress.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 28

Examples of side-lobe signals tracked from the lunar surface during OP74_0 (61.90 RE)

GPS L1 C/A and Galileo E1C are shown on the top; GPS L5Q and Galileo E5a-Q are shown on the bottom.

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 29

L1/E1 and L5/E5a C/N₀ tracking arcs for GPS PRN6 and Galileo PRN 34, as received during OP17_0

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 30

Galileo main-lobe signals tracked from the lunar surface during OP38_0 over approximately 24 h

Dashed lines show the linearly interpolated trend of C/N₀ maxima for the different orbital planes (i.e., A and B).

• Download figure
• Open in new tab
• Download PowerPoint

FIGURE 31

Galileo main-lobe signals showing effects on C/N₀ measurements when traversing the Earth ionosphere during OP76_0. (a) E36 E1/E5a space skyplot, (b) E36 elevation, (c) Signature of ionospheric occultation on E36 E1/E5 pilot signals.

6.2 Payload Design

• In general, multi-constellation and multi-frequency capabilities greatly enhance performance in this environment, primarily by increasing signal availability and measurement quantity. Increasing the capability of lunar receivers to support additional GNSS constellations and signals can be shown by analysis to be beneficial, and LuGRE demonstrates this benefit in practice. For example, during a 24-h period on the lunar surface (OP38_0), multi-fold (2+) signal availability increased from 25% for a dual-frequency GPS-only scenario to nearly 70% with both GPS and Galileo included (see Figures 22–24). Similarly, Table 5 shows gains in 4+ signal availability with the inclusion of L5/E5a in addition to L1/E1.

• In any weak-signal environment, especially at lunar distances, the C/N₀ margin is low, and signal visibility and thus navigation availability are greatly impacted by marginal link budget effects. The large C/N₀ discrepancy observed by LuGRE is under investigation, but highlights the need for a robust link budget margin, careful calibration of the payload link budget before flight, and root cause analysis for any observed discrepancies.

• Because of the inherent benefits of the L5/E5a signals, receivers can benefit by acquiring these signals directly, rather than using L1/E1 as an acquisition aid. This approach will improve the ease of acquisition as each PRN comes into view and will improve the overall measurement quantity, as the L5/E5a signal tends to be available above a given receiver C/N₀ threshold longer and earlier than the corresponding L1/E1 signal.

• If demanding navigation performance requirements are levied, a high-stability clock is likely necessary. LuGRE saw successful acquisition and tracking using a commercial-grade VCTCXO, but did not have specific requirements for navigation accuracy. Previous studies point to the need for high-stability clocks for high-precision real-time navigation (Winternitz et al., 2019).

• A robust aiding approach is beneficial in numerous scenarios, including during non-continuous operations, as in the LuGRE transit phase, or during continuous operations in lunar orbit, when occultations may periodically interrupt the line of sight. As described in Section 3.2, LuGRE used aiding of time, its own ephemeris, and GNSS constellation ephemerides and was thus enabled to quickly acquire the signal when started. Accepting some or all of these aiding inputs adds flexibility to the system. At a minimum, accepting an uploaded constellation ephemeris is likely the most important, as the ability to obtain ephemeris demodulated from the signal itself is degraded with a lower C/N₀. In the LuGRE case, lander ephemeris aiding added a high degree of operational complexity, and time aiding imposed additional requirements on the host vehicle that may not be reasonable in other cases.

6.3 Operations

• Continuous, long-duration operations are an obvious advantage of GNSS navigation, as it allows time for cold-start acquisition; allows for stabilization of payload hardware, especially clock drift; enables long-duration tracking arcs to be maintained for each acquired PRN; and allows for full convergence of a navigation filter. This approach depends on the availability of a GNSS antenna that can be pointed at will and to the necessary accuracy of the user’s system.

• If continuous operations cannot be achieved, powering the payload components well in advance of the start of processing is useful, as it allows for warm-up and stabilization of the components, especially the clock. For example, Figure 5 shows the 10- to 15-min clock drift transient observed during the LuGRE receiver power-on. Situations requiring rapid navigation convergence may especially benefit from this practice.

6.4 Navigation

• Use of a navigation filter is critical for robust navigation in the lunar environment, considering the relative sparsity of instantaneous PVT solutions and the poor geometry associated with lunar-distance signal reception. Such a filter should implement best practices, including the option to exclude measurements from signals for which the height of the ray path (or grazing altitude) is below a threshold and to automatically edit measurements for outliers and visibility constraints (Carpenter & D’Souza, 2025).

• Utilizing current Earth orientation parameter data is critical for high-accuracy navigation at lunar distances, as errors can increase to the level of hundreds of meters in coordinate frame rotations if incorrect or outdated values are used. These values can be provided via telecommand, as was the case for LuGRE, or by supporting broadcast ephemeris messages containing these data, such as the GPS Civil Navigation (CNAV) message on its L5 signal.