Abstract
This paper presents an experimental alternative positioning, navigation, and timing system for ground users, implemented at X-band, and demonstrated using a formation of four small unmanned aerial vehicles. The experimental system is designed to support stand-alone ground user positioning at the few-meter level within 30 s and to study the features and characteristics of X-band for this purpose. The payload that creates the X-band transmissions (8.52 GHz) and the user receiver are both based on commercial-off-the-shelf components and a software-defined radio architecture, enabling the study of various signal designs and receiver acquisition and tracking strategies. Horizontal positioning at a level of 1.5 m (root mean square) is demonstrated for both static and moving receivers. Examples of the tracking performance and multipath effects on individual signals highlight the effectiveness of the signal design and the payload and receiver implementations.
1 INTRODUCTION
The need for alternative sources for position, navigation, and timing (PNT) for situations in which basic global navigation satellite systems (GNSSs) are not available has been well established and widely publicized. An important reality is that no single alternative PNT option can serve as a full replacement for all existing GNSS applications. Rather, we are seeing the emergence of alternatives designed to efficiently address a subset of PNT needs. Currently available examples include ground-based regional systems such as Locata (Rizos & Lang, 2019) and NextNav’s Metropolitan Beacon system (Meiyappan et al., 2020), expanded two-way time and frequency transfer services for critical infrastructure by the National Institute of Standards and Technology (Sherman et al., 2021), and passive Doppler-based techniques for positioning in regions widely serviced by low-Earth-orbit (LEO) communication satellites (Psiaki, 2021; Kassas et al., 2019). Investments are also being made in existing and developing dedicated LEO-based systems proposed by commercial companies (Reid et al., 2018, 2020; Kassas, 2020).
Here, we describe X-PNT, an experimental alternative PNT system for ground users implemented at X-band, and demonstrate this system using a formation of small unmanned aerial vehicles (UAVs). The intent of this overall effort is to explore PNT performance at X-band, in anticipation of potential future integrated communications and navigation services from LEO, UAV, or fixed platforms. We designed the experimental demonstration system to achieve stand-alone ground user positioning at the few-meter level within 30 s and to study the features and characteristics of X-band for this purpose.
Both the payload that creates the transmissions and the user receiver are based on a software-defined radio (SDR) architecture. The experimental X-PNT demonstration system comprises the following key elements:
Signals: X-band carrier, pseudorandom noise (PRN) codes, data required for positioning and timing.
Payload: Ettus USRP B210 SDR, Microchip (2024) SA65 chip-scale atomic clock (CSAC)-based timing system, NovAtel (2024) OEM729 GNSS receiver, Blue Cubed (2024) Bluefin X-band transmitter, and custom X-band patch antenna.
Ground receiver: Antenna, front-end, and SDR intermediate-frequency (IF) data capture; software receiver for acquisition, tracking, position, and time solutions.
Small UAV-based testbed: Set of four custom-built hexacopters, ground control system with features to support smooth trajectories, stable flight operation, and truth referencing using GNSS real-time kinematic (RTK) components.
Figure 1 illustrates the configuration we used for live-sky testing. Each UAV is equipped with an X-PNT payload that receives Global Positioning System (GPS) signals and autonomously creates and transmits a PRN-coded signal on an experimentally licensed carrier at 8.52 GHz. In a future integrated communication/navigation system, the payload position and timing could be derived by means other than GPS or predicted based on a combination of past GPS data and accurate models for the vehicle trajectory and clock behavior. For the purpose of the X-band experiment with payloads on UAVs, an onboard GPS receiver was a convenient solution. Ground user receivers (not shown) capture the X-band signals to solve for their positions. An independent truth reference system based on GNSS RTK is installed on each UAV and ground receiver for validation purposes.
X-PNT payloads use an onboard GPS receiver for their position and time and then automatically generate and transmit X-band ranging signals with navigation and timing messages to support ground users without access to GPS (not to scale).
This initial demonstration system relies on GPS being available to the UAV host platforms to support flight operations and to their onboard X-PNT payloads to establish time and the transmitter locations included in the navigation messages sent to the ground users. The X-PNT ground user receiver only requires access to the X-PNT signals for its operation, but has the option to simultaneously capture raw GPS IF data for validation and comparison.
The remainder of this paper describes the design features of each element of the X-PNT system and then presents results from a live-sky experiment demonstrating positioning at a level of 1.5 m (root mean square [RMS]) for both static and moving receivers, based entirely on the X-PNT signals.
2 SIGNAL DESIGN
The motivation for exploring X-band as an alternative PNT signal option is based on its following advantages: significant frequency diversity from widely used L-band signals, small antennas that can support a wide field of view or that can be fabricated in arrays for higher directivity and interference rejection, potential for integration of PNT functions onto existing X-band communication links, and significantly reduced ionospheric effects when used in a space-to-ground configuration (not relevant for a UAV-based system). The potential utility of X-band PNT as compared with L-band is also negatively influenced by a larger link loss (~15 dB) due to the higher frequency, greater rain attenuation (evaluated to be less than ~5 dB under most conditions when tracking signals at elevations above 5°), and increased signal phase noise (not a significant influence on code tracking). For the specific small UAV-to-ground experiments conducted thus far, X-band signal paths are less than 1 km; thus, tropospheric delays and attenuation are negligible, and ionospheric effects are not present at all.
To carry out this study, we designed a minimum viable signal based on the target performance metrics described above and obtained an experimental license from the Federal Communications Commission to transmit at 8520 MHz from an altitude above ground level (AGL) of less than 130 m within a 50-km radius of Boulder, CO.
The qualitative signal design drivers are as follows:
The signal must be simple, modular, and operable on the extant Bluefin X-band transmitter to minimize development risk. To meet this requirement, the pilot and data channels were placed in quadrature at equal powers, and Weil-based memory PRN codes were utilized (Rushanan, 2007). One of the key advantages for memory codes in general is the ease with which they can be changed in transmit and receive signal processing chains.
The signal must be “easy” to acquire at the higher X-band carrier frequency (compared with GNSS L-band) and with a low-cost user clock. This requirement led to the use of short, length-1000, primary PRN codes on the pilot channel with a modest rate of 1 megachips per second (Mcps).
Multipath issues in a ground mobile environment can significantly influence navigation performance. Propagation at X-band is poorly characterized in a navigation context. Obtaining a better understanding of its characteristics was one of our major objectives. Faster code chipping rates are a powerful mitigant against long delay reflections, as any signal with a delay of more than 1.5 chips is highly attenuated in the correlation process. For this reason, the data channel employs a 10-Mcps chip rate and serves as the primary ranging channel.
Timely and reliable conveyance of the time, clock offset, and transmitter location data is essential in a navigation system. Our initial studies suggested that a 200-bit packet size is sufficient to provide adequate granularity without imposing too much overhead. Each block-encoded packet takes 2 s to transmit using a standardized forward error correction (FEC) format. With the selected content format, each packet is sufficient to provide a complete snapshot of time, transmitter position, and clock corrections.
Power-efficient FEC encoding and decoding techniques have been widely studied, especially in the cellular community. In addition to being power-efficient, FEC also provides improved tolerance to short signal outages. Early field testing showed that at X-band, short delay ground reflections can be quite strong, particularly at low elevation angles, and that these reflections could lead to rapid signal fluctuations in amplitude and phase. Concomitant tracking issues suggested a need for signal designs that are operable in the absence of a reliable phase lock; thus, a differential phase-shift keying (DPSK) symbol format was selected for the data messages. Of note, the DPSK format does require that the C/N0 be ~3 dB higher compared with that of BPSK under the same conditions, but this tradeoff is viewed as worthwhile in exchange for improved robustness against channel effects (Proakis, 2008).
All X-PNT signal components are based on a 10-MHz base frequency. The signal uses a quadrature phase-shift key format (QPSK), with equal amplitudes on in-phase “I-channel” and quadrature “Q-channel” components. The I and Q channels are intended to be jointly processed, with the I-channel primarily serving in a carrier pilot/acquisition role and the Q-channel channel supporting code tracking and navigation data conveyance.
The PRN code modulation is binary phase-shift keying (BPSK), with a length-1000-bit code at 1 Mcps on the (I) pilot channel and a length-10,000-bit code at 10 Mcps and 100-bps navigation data on the (Q) data channel. The length-1000-bit codes are based on length-997 Weil codes with insertions [0 1 1] to make codes of length 1000. The length-10,000-bit codes are based on length-10,007 Weil codes with deletions of 7 contiguous symbols (Rushanan, 2007). In both cases, a set of 200–500 candidate codes was developed based on measured autocorrelation properties. Candidate families were then developed by randomly selecting a family from the candidate codes and measuring cross-correlation properties. Ten to 100 million candidate families were tested, and the family with the best cross-correlation properties was selected. In the configuration we considered, only a small number of transmitters are expected to support a particular region; hence, codes are not assumed to be uniquely assigned to specific vehicles. Consequently, the X-PNT receivers will always search over the full set of possible PRNs for acquisition, and there is no need for GPS-almanac-type predictions of the transmitter trajectories.
FEC is implemented using standard 5G low-density parity-check (LDPC) encoding, as specified in 3GPP TS38.212 but adapted to our needs, and a DPSK symbol format is used to provide robust data demodulation without coherent carrier-phase tracking. TS38.212 allows for flexible rate adaption using puncture techniques. We use 5G NR DL-SCH (down-link shared channel) defaults. The navigation message consists of 200 bits covered by a 16-bit cyclic redundancy check. The message is conveyed via 399 DPSK symbols, resulting in a code rate of R=216/399. Length-10 Neuman–Hofman secondary codes are used to reduce vulnerability to interference associated with short code sequences, and a 400-bit tertiary code on the pilot establishes timing with a 4-s ambiguity.
Navigation messages are automatically constructed by the X-PNT payload, reporting the time of transmission, PRN assignment, UAV location, clock bias, drift, and time of applicability. Navigation messages are currently sent once every 2 s, with the location and clock data currently updated every 5 s.
3 PAYLOAD DESIGN
Hardware Components
The payload architecture developed for this project is a software-defined transmitter built on top of commercial-off-the-shelf (COTS) components. COTS components, including the GPS receiver, CSAC, and X-band transmitter, supply the key system functions. Figure 2 shows a top-level block diagram of the payload hardware and software components, with the primary hardware components described in Table 1 Primary Payload Hardware Components. Rather than implementing a GPS-disciplined clock onboard, we use the GPS receiver to monitor the frequency and bias of a free-running CSAC (Dobbin, et al., 2024 Feb, 2024) that serves as the common timing reference for all system components. The use of the CSAC provides the ability for holdover on the order of tens of minutes, to support continued transmission of the X-PNT signal in case of a GPS outage. This short-term extension of accurate timing in the absence of GPS is compatible with the operation of alternative PNT platforms that have predictable motion or access to other sources of positioning data.
X-PNT payload architecture
CPU: central processing unit; LPF: low-pass filter; PPS: pulse per second; TX: transmitter; UART: universal asynchronous receiver–transmitter
Primary signal generation functions and implementation LO: local oscillator
In addition to the COTS components, a substantial amount of custom hardware, digital logic, and software is required to achieve the desired functionality. The primary component of the payload is a Xilinx Zynq system-on-chip, which is a hybrid general-purpose processor and field-programmable gate array (FPGA). The Zynq hosts the X-PNT transmitter software application as well as custom digital logic that allows X-PNT spreading codes to be deterministically generated relative to the system clock. Figure 3 displays the primary signal generation functions of the payload and how they have been implemented in the system.
The payload’s digital electronics leverage the design of similar boards developed for the MAXWELL CubeSat project (Aboaf, 2020), which has similar size, weight, and power (SWaP) constraints as a small UAV platform. The hardware is organized on two stacked 5 in × 2.5 in printed circuit boards, as shown in Figure 4. The bottom board hosts the Zynq module and configuration memory. The top board hosts the GPS receiver, CSAC reference clock, clock fanout circuitry, and Bluefin transmitter interface. Figure 4 also shows the payload (without the antenna) inside the housing constructed for UAV testing.
Digital electronics board stack (left) and payload inside housing (right)
Payload Operation
Each payload operates autonomously, with key configuration parameters—such as the spreading code, modulation and chipping rate, and navigation data rate— stored on an SD card. Upon power-up, the transmitter software initializes, polls the CSAC for status, and waits for the CSAC to report “lock.” Once locked, the GPS receiver is commanded to use the external clock and computes a full position and time solution. The transmitter then sets its reference time using the GPS 1PPS signal, derived from the CSAC’s 10-MHz clock and aligned within one clock cycle (100 ns) of GPS time. After initialization, all components operate on the 10-MHz reference from the CSAC.
The payload GPS receiver monitors the CSAC clock offset and drift relative to GPS time while estimating the platform’s position and velocity. The transmitter application polls the GPS receiver at 1-s intervals, logging time, position, and clock data. These values are parsed and encoded in the compact X-PNT navigation message transmitted to the users. For diagnostics, the transmitter also records status messages from the CSAC.
Once the timing lock and required navigation data are established, the transmitter application sequences the digital modulation, with all timing derived from the CSAC. The transmitter then begins broadcasting and continues until powered off.
Laboratory testing using a software-defined RF simulator was conducted to calibrate and verify time alignment. A consistent digital delay of five clock cycles (500 ns) was observed in all payloads and was applied as a calibrated scheduling offset in the payload software. Early testing led to the removal of an unnecessary onboard temperature-compensated crystal oscillator (TCXO) that introduced additional noise. Early testing also informed the selection of a frequency plan ensuring integer-N phase-locked loop (Int-N PLL) operation in the generation of all required clocks in the payload. Integer-N division in frequency synthesis minimizes phase noise and avoids fractional spurs often seen with fractional-N synthesis (Barrett, 1999).
4 RECEIVER DESIGN
The demonstration system ground receiver is designed for maximum flexibility in evaluating the performance of the proposed system concept. Its design stems from the existing GNSS SDR architecture developed at the University of Colorado Boulder (Morton et al., 2020), which efficiently captures raw IF data in real time and then allows for signal processing and navigation solutions to be performed offline in a software receiver code.
Receiver Hardware
The X-PNT receiver architecture follows an approach similar to that used in the payload design in leveraging SDR methods with COTS hardware components. For both laboratory and field testing, we relied on the ability to capture and record IF data sets that are then post-processed on a general-purpose computer. The X-PNT user receiver hardware is designed to be transportable for outdoor testing, but it is not constructed as a rugged fieldable piece of user equipment. A photograph of the fabricated receiver is shown in Figure 5.
The key receiver hardware components and their configuration are presented in the block diagram in Figure 6. The X-band front-end is on the right, starting with the patch antenna tuned to our experimentally licensed frequency of 8.52 GHz. A 20-dB low-noise amplifier (LNA) and 10-dB attenuator follow. In the center (shown in blue) is the 10-MHz oscillator, which is distributed as a common clock reference to both the mixer that downconverts the X-band to 1120 MHz and to the Ettus USRP B210 SDR that performs the data capture. The data stream is logged to the laptop computer shown in gray. This computer is responsible for the real-time capture of digital samples as well as the subsequent signal processing and compu-tation of the navigation solution.
Receiver Software
The X-PNT signal processing and navigation solution software is written in Python, expanding upon existing GPS receiver code (Morton et al., 2020) to incorporate the unique features of the X-PNT signal structure. The key elements of the software receiver application are shown in Figure 7.
Fabricated X-PNT receiver and X-band patch antenna (not including the GPS antennas and LNAs, which are connected via cable)
The dimensions of the receiver assembly are approximately 6”× 6”× 6”.
Block diagram of the X-PNT ground receiver hardware RHCP: right-hand circularly polarized
Signal Acquisition and Tracking
The use of an SDR data capture and full software receiver implementation allows for implementation and evaluation of multiple acquisition and tracking strategies. Processing options are specified in a configuration file. The Python receiver application streams in recorded data samples from “collect” files that we typically limit to 120 s. Signal processing is performed with coherent integration times from 1 to 10 ms. The receiver software is currently being run on a rugged field laptop (11th Generation Intel® Core™ i5-1135G7 at 2.40 GHz).
Acquisition of the pilot signals from all four transmitters is performed in parallel, using a 20-ms block of samples and non-coherently combining two 10-ms coherent integrations. Correlations are implemented with an efficient fast Fourier transform-based circular correlation approach (Tsui, 2000) covering the full range of code phase values and expected Doppler. Tertiary code wipe-off improves acquisition performance when there is a large dynamic range between signals. The average time to perform the calculations for signal acquisition, as currently implemented on the field laptop, is just under 1 min.
The default tracking configuration is designed for accurate and robust code tracking of weak signals. On both the pilot and data channels, we implement a first-order delay-locked loop (DLL) (1.5-Hz bandwidth [BW]) for code tracking and a third-order PLL (18-Hz BW) aided with a second-order frequency-locked loop (FLL) (2-Hz BW) for carrier tracking, also known as an F-PLL. Correlations are computed with a 5-ms coherent integration period that includes tertiary code wipe-off on the pilot channel. The pilot and data channels can each be tracked independently, but our nominal tracking mode couples the two together to take advantage of their distinct features. In the coupled tracking method, carrier tracking for both pilot and data channels is based on the pilot signal with the F-PLL discriminator driving the feedback loop. The aligned local carrier replica generated by this loop is used as input for both pilot and data channel processing, with the latter applying a quarter-cycle phase shift to account for the signal transmission in quadrature. Code tracking is based on the DLL discriminator for the data channel with the higher chipping rate (10 Mcps), thus reducing code tracking noise and multipath effects. Data decoding is performed using the same AFF3CT library routines implemented on the payload for encoding purposes.
X-band software receiver application block diagram PVT: position, velocity, and time; RX: receiver; SNR: signal-to-noise ratio
The variance of the code tracking error due to noise is estimated as follows (Misra & Enge, 2011, Eq. (12.34)):
1
where Bτ,1 is the DLL loop bandwidth, d is the correlator spacing, TC is the chip width, TCO is the coherent integration period.
The current implementation of tracking and data decoding on the field laptop nearly achieves real-time operation, with an average time of 68 s to perform tracking of four signals on a 60-s data collect.
Positioning
Full three-dimensional position and time solutions are possible when signals from four transmitters are successfully acquired, tracked, and decoded. The receiver forms a pseudorange to each transmitter based on the time of reception for the observed signal, the transmitted reference time, and the payload clock offset reported in the navigation message. The position of each UAV in Earth-centered Earth-fixed coordinates is computed by interpolating the positions reported in the X-PNT navigation messages. The ground receiver position and clock solutions are solved using an iterated linear least-squares method. The position estimate is initialized to the projection of the average position of all of the transmitters onto Earth’s surface. The linear least-squares solution is computed at each time step and iterated until the change in the position solution is less than 0.1 m (typically 3–4 iterations). Of course, more sophisticated estimation methods incorporating dynamical models for the user position (stationary or moving) can be applied; however, for this initial system demonstration, we considered only independent point solutions.
5 UNMANNED AIRCRAFT SYSTEM TESTBED
To demonstrate the operation and performance of X-PNT, we developed a small unmanned aircraft system (UAS) testbed. This system is designed to enable flexible and accurate positioning of the transmitters across a range of flight dynamics while logging precise ground truth data for validation of the X-PNT system performance. The UAS is comprised of a set of four custom-built multirotor UAVs, a ground control station supporting multi-vehicle telemetry links, an RTK base station, and a stand-alone RTK GPS tracker, as shown in Figure 8.
The UAVs are primarily constructed from COTS components, based around the commercially available Pixhawk “Cube Blue” autopilot. The lift system components were selected to keep each transmitter airborne for roughly 30 min at a time, with enough surplus thrust to safeguard the payloads in the event of a lift system failure.
Photographs of the demonstration system (clockwise from upper left): one of four UAVs, X-PNT payload and receiver, X-band antenna, static user site, UAS ground system
Command and control of the UAVs are accomplished using a 915-MHz point-to-multipoint telemetry link connected to the UAS ground control station laptop. This link can simultaneously handle telemetry data from multiple aircraft and allows commands to be issued to the aircraft during flight. Flight paths are pre-planned using coordinated waypoint paths generated using MATLAB and are uploaded to each UAV before flight.
Onboard positioning is provided by a uBlox F9P multi-GNSS receiver equipped with a radio link for receiving RTK corrections. We emphasize that the onboard positioning of the UAVs used for flight control and truth referencing provides a precise system that operates completely independently of the X-PNT payload described in Section 3. RTK is used here to enhance the navigation accuracy and execution of the pre-planned UAV flight profiles and to provide precise trajectories of each transmitter and user platform as a truth reference for analysis of the X-PNT system performance. The RTK corrections are computed by an RTK base station based on the uBlox F9P multi-GNSS receiver, located at a reference point within our testing area. The base station is equipped with a radio that broadcasts correction data to all rover units within range. The RTK position solutions available onboard each UAV have decimeter-level accuracy with respect to the base station reference location. We record these data for comparison with real-time stand-alone GPS solutions from the X-PNT payloads and for diagnostics.
An additional RTK GPS tracker, based on the same hardware used onboard the UAVs, was assembled to log positioning truth data for the static and car-mounted moving X-PNT ground receivers. The X-PNT system payloads and receivers operate independently of the UAV testbed, with no data shared between the two during operation.
6 X-PNT DEMONSTRATION
The remainder of this paper details the UAS flight experiment conducted on April 22, 2024, to demonstrate the X-PNT system at the Arvada Associated Modelers club field south of Boulder, Colorado. The UAVs were operated under a University of Colorado Certificate of Authorization (COA) (2023-WSA-13256-COA) issued by the Federal Aviation Administration.
Four airborne transmitter payloads were used to establish an X-PNT signal coverage area surrounding the central parking area and access road. Two X-PNT ground-based receivers were operated in this area, with one stationary at the center of the signal coverage area and the other mounted on a car moving dynamically within the region. An overview of the test site, receiver locations, and planned UAV flight paths is shown in Figure 9.
The static receiver was set up in the center of a dirt parking lot at the Arvada test site. Its location was specified as the reference point for the pointing of all payload transmitting antennas. Using the RTK base station and tracker, the location of the static receiver patch antenna was surveyed for 15 min to generate a truth position estimate. Once surveyed, the RTK tracker antenna was replaced with the receive patch antenna to minimize error in our knowledge of the reference point.
The UAV flight plans were designed to achieve a suitable geometry for demonstrating positioning based on the X-PNT signals transmitted from moving platforms. To comply with operational safety requirements and simplify piloting, the paths were chosen to avoid overflight of the ground crew and crossing of the UAV paths. Each transmitter slowly sweeps back and forth through a unique profile of azimuths and elevations, at a relatively constant distance from the static receiver reference point. The trajectories illustrated in Figure 9 show the circular arc of transmitter A (TX A) at elevations of 30°–70° with respect to the reference point, TX B and C in opposite quadrants of the sky arcing over a range of elevations between 10° and 40°, and TX D traveling at a low elevation of 5°–7° through a 30° arc in azimuth. As is done in GNSSs, we use the dilution of precision (DOP) (see, e.g., the work by Misra and Enge (2011)) to describe the quality of the geometry for the X-PNT point solutions. For positioning of ground users, we are primarily interested in the DOP in the east and north directions or the overall horizontal (HDOP). Figure 10 shows the HDOP and vertical DOP (VDOP) values over the course of a 20-min flight plan, for both the static reference receiver location and points along the path taken by the moving vehicle.
Multi-transmitter flight plan overlaid on a Google Earth image of the Arvada Associated Modelers club field
The UAVs were deployed sequentially by dedicated pilots and held their positions at staging waypoints. When commanded into autonomous operational mode, each UAV executed its unique pre-programmed flight path, while automatically adjusting its heading to orient the payload X-band antenna toward a specified ground reference point. Raw IF data sets were captured by the static and moving receivers in 1- to 2-min blocks to facilitate file transfer and data analysis.
After testing, the autopilot and tracker logs were offloaded and post-processed to generate the truth reference data set. This involved converting the autopilot logs to a MATLAB-readable format and extracting the relevant data for each UAV or ground object. The altitude data were converted from above mean sea level (AMSL) to WGS84, which is necessary to refer the data logged by the autopilot to the same vertical datum as the outputs from the X-band system. Finally, truth data products were calculated and output, which includes a position log for each moving transmitter or receiver, coordinates for any static equipment, and transmitter geometry relative to each receiver (elevation angle, range, etc.).
HDOP and VDOP of the UAV system for static (top) and moving (bottom) receivers (RX)
(April 22, 2024 test)
TOW: GPS time of week
Static Receiver Results
The static receiver recorded nine data collects during this test flight, of which three have a duration of 60 s and six have a duration of 120 s.
Table 2 lists the collect ID, collect duration, percentage of successful position solutions produced out of the total possible for the given duration, and RMS errors with respect to the RTK position. In each collect, the first position solution was completed after processing only 4–8 s of recorded samples with no a priori knowledge of the user or transmitter locations, enabled by the compact navigation message structure, robust data recovery, and effective acquisition and tracking approaches. This result indicates that with a faster acquisition implementation (faster processor and/or code optimization), the system should be able to support a time-to-first-fix well below the minimum 30 s required for the GPS legacy navigation message. The 100% position solution success (Pos. % column) for each entry indicates that after the initial acquisition and first fix, the receiver was able to continuously track all four transmitters and form position solutions for the remainder of the 60-s or 120-s data collect.
Figure 11 shows the point positioning results for each collect, overlaid with the truth position surveyed using the RTK tracker and base station. Figure 12 shows the distribution of errors in the east, north, and up directions with respect to the truth established by the RTK station. The mean error and standard deviation over all solutions are included below each subplot.
Static receiver position results for the east, north, and up coordinates relative to the reference point
Static position errors (referenced to RTK truth position)
In summary, the following results were achieved with the static receiver:
Four transmitters were acquired and tracked for 100% of the test period.
The horizontal position solution was within 1.3 m (RMS) of the truth reference.
The vertical position solution was within 1.4 m (RMS) of the truth reference.
The first fix was achieved with a duration of 4–8 s of collected samples.
Moving Receiver Results
The moving receiver antenna was mounted to the top of a Toyota 4Runner SUV. The truth reference was established by an RTK tracker on the roof of the vehicle, located less than 1 m away from the X-band antenna along the length of the car. During the duration of the test, the RTK tracker recorded its position for comparison to the X-PNT position solution for each collect.
The moving receiver recorded a total of nine complete collects for this flight; five collects had a duration of 60 s, and four had a duration of 120 s. Table 3 lists the collect ID, collect duration, percentage of successful position solutions produced out of the total possible for the given duration, and RMS errors with respect to the RTK positions. Positioning was not possible for collect 5 because of difficulty in tracking Payloads C and D due to the range and geometry between the moving receiver vehicle and the payloads.
For each of the collects, the moving receiver vehicle primarily drove back and forth in the east–west direction along the main road in the testing area. For collects 5, 6, and 7, the vehicle added a north–south path through the dirt field, forming a large rectangle around the test area. Figure 13 shows the position solutions from the moving X-PNT receiver with the truth reference from the RTK tracker overlaid in red.
Figure 14 shows the distribution of errors in the east, north, and up directions with respect to the truth established by the RTK tracker on the moving vehicle. Although the RTK positions are considered as “truth” when error statistics are computed, we note that there may be small errors and biases within the RTK system. The average error and standard deviation are included below each subplot.
Moving receiver position solution in geodetic coordinates
Histogram of moving receiver position error (referenced to RTK truth)
In summary, the following results were achieved with the moving receiver:
Four transmitters were acquired and tracked for 100% of the test period (#1–4, 6–9).
The horizontal position solution was within 1.4 m (RMS) of the truth reference.
The vertical position solution was within 1.3 m (RMS) of the truth reference.
The first fix was achieved with a sample duration of 4–10 s.
Example Tracking Results
Examples of code tracking performance, with the coupled data and pilot channel tracking strategy, are presented in Figure 15 and Figure 16 for the static and moving receivers, respectively. Each example shows the measured carrier-to-noise ratio (C/N0), zero-mean code-minus-carrier (CMC) observable, and code discriminator values, along with the range and elevation of the transmitter with respect to the receiver, for all four transmitters. The range and elevation are included to provide context for the variations in C/N0 and resulting tracking effects. For typical situations in which carrier tracking is far more precise than code tracking, the CMC provides an indication of both multipath and noise in the pseudorange measurements. The discriminator plots illustrate the tracking noise and show theoretical 2-σ bounds computed using Equation (1), with the observed C/N0 values and default tracking parameters Bτ,1 = 1.5 Hz, d = 0.6 chip, TC = 10−7 s, and TCO = 0.005 s. Average values of the C/N0, CMC, and DLL discriminator are presented in Table 4 for each of the examples. The plots and values in the table are based on the tracking results from the time of first fix to the end of the collect.
Tracking performance: static receiver, slow transmitter motion; April 22, 2024
Tracking performance: moving receiver, slow transmitter motion; April 22, 2024
Starting with the static receiver tracking shown in Figure 15, we observe the highest C/N0 and lowest tracking noise on PRN0, transmitted by the platform at closest range and highest elevation (approximately 45°–65°). Both the CMC and DLL tracking noise are at the 10-cm level, decreasing slightly with the decrease in range and increase in elevation from the start to end of the collect. PRN2 was transmitted from a nearly constant range of 127 m and elevation of 20° and received at an average C/N0 of 38 dB-Hz, with the DLL tracking noise approximately three times larger than that of PRN0. PRN3, transmitted from a UAV flying less than 10° above the horizon relative to the static receiver, has a higher tracking noise of 0.46 m (1-σ) and CMC of 0.70 m (1-σ). PRN1, operating at approximately the same range and slightly higher elevation than PRN3, shows both DLL and CMC values at the 0.38-m (1-σ) level. In both PRN3 and PRN1 results, we can see time-correlated variations in the C/N0 and CMC observables, indicative of multipath near the receiving antenna. For PRN1, we observe sharper fades, which may be the result of partial signal blockages. For this static receiver collect, 99.8% of the CMC values for PRN 3 and all of the CMC values for PRNs 0, 1, and 2 were below 2 m.
The moving receiver results presented in Table 4 and Figure 16 show variations in C/N0, tracking noise, and CMC that are more directly related to changes in the range and elevation. Smaller features may be due to multipath or effects of the SUV driving over bumpy terrain; however, in this example, we observe stronger high-frequency noise and less evidence of correlated pseudorange errors than observed for the static receiver. For this moving receiver collect, 99.3% of the CMC values for PRNs 2 and 3 and all the CMC values for PRNs 0 and 1 were below 2 m.
Overall, the results show that the X-PNT signal structure supports code tracking at the desired level for both static and moving ground receivers. Successful generation of pseudorange measurements and positioning during this time confirm proper data recovery as well.
7 CONCLUSIONS
This paper has presented the motivation, design, and example test results for an experimental X-band alternative PNT system demonstrated using a small UAS. The signal design incorporates key lessons from GNSSs and is optimized for easy acquisition and rapid, PRN code-based positioning. The small payloads automatically produce stable signals on an 8.52-GHz carrier, aligned to GPS time with all required information for positioning included within a 2-s navigation message. Ground receivers capture raw IF data, allowing for flexibility in testing various acquisition, tracking, data decoding, and positioning strategies. The UAS provides GPS RTK-based truth reference data for validation of the X-PNT performance.
Test results with the small UAS show horizontal positioning at a level of 1.4 m (RMS) and the ability to form the first position fix with 4–8 s of captured IF data. Testing with an X-band carrier shows higher-frequency, more noise-like code multipath than is typical for GPS L-band tracking. The use of coupled pilot and wideband code (10 Mcps) tracking loops produces code measurements with more than 99% of tracking errors below 2 m.
We are eager to further explore the utility of this basic system design – for example, to provide alternative positioning and timing signals in areas where GPS is not available, to perform two-way ranging as an alternative to the one-way approach used in GNSSs, and to add the demonstrated PNT capabilities to existing X-band communication links.
FINANCIAL DISCLOSURE
This work was supported by the Air Force Research Laboratory, contract FA9453-20-C-2000.
HOW TO CITE THIS ARTICLE:
Axelrad, P., Dobbin, M., Akos, D., Morton, J., Palo, S., Scott, L., Kingsbury, R., Breitsch, B., Kenyon, A., Bourne, H., Taylor, S., Wallace, B., Crews, A., & Pham, K. (2026). UAV-Based Demonstration of Alternative PNT at X-Band. NAVIGATION, 73. https://doi.org/10.33012/navi.761
ACKNOWLEDGMENTS
The views expressed herein are those of the authors and do not reflect the official guidance or position of the United States Government, the Department of Defense, or the United States Air Force.
Manuscript cleared for public release, Case Number AFRL-2025-1666.
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
REFERENCES
- ↵Aboaf, A. (2020). Maxwell mission handbook [PDF]. NASA. Retrieved February 12, 2025, from https://s3vi.ndc.nasa.gov/ssri-kb/static/resources/maxwell_mission_handbook_released.pdf
- ↵Barrett, C. (1999). Fractional/integer-N PLL basics [Technical brief SWRA029]. Texas Instruments. Retrieved April 7, 2025, from https://www.ti.com/lit/an/swra029/swra029.pdf
- ↵Blue Cubed. (2024). Bluefin datasheet [PDF]. Retrieved February 12, 2025, from https://cdn.prod.website-files.com/61ca672af16cb0773cab77c8/65e14a2a8eed47acbc17b04b_Bluefin%20slick%2020220425.pdf
- ↵Dobbin, M., Colpaert, C., Krebs, C., & Axelrad, P. (2024, February). Characterizing CSAC performance in a simulated mission environment. In Proceedings of the 44th Annual American Astronautical Society Guidance, Navigation, and Control Conference, 2022 (pp. 19–36). Springer, Cham. https://doi.org/10.1007/978-3-031-51928-4_2
- ↵Dobbin, M., Hunter, A., Gravel, W., Parker, C., & Axelrad, P. (2024, July). Characterizing a chip scale atomic clock in low Earth orbit. In Proceedings of the Small Satellite Conference: Advanced Technologies I (Paper 124). Utah State University. http://digitalcommons.usu.edu/smallsat/2024/all2024/124/
- ↵Kassas, Z., Morales, J., & Khalife, J. (2019). New-age satellite-based navigation STAN: Simultaneous tracking and navigation with LEO satellite signals. Inside GNSS, 14(4), 56–65. https://insidegnss.com/new-age-satellite-based-navigation-stan-simultaneous-tracking-and-navigation-with-leo-satellite-signals/
- ↵Kassas, Z. Z. M. (2020). Navigation from low-Earth orbit. In Morton, Y. T. J., van Diggelen, F., Spilker, J. J., Jr., Parkinson, B. W., Lo, S., & Gao, G. (Eds.), Position, navigation, and timing technologies in the 21st century: Integrated satellite navigation, sensor systems, and civil applications (Vol. 2, Ch. 43, Part 2, pp. 1381–1412). Wiley. https://doi.org/10.1002/9781119458555.ch43b
- ↵Meiyappan, S., Raghupathy, A., & Pattabiraman, G. (2020). Position, navigation, and timing with dedicated metropolitan beacon systems. In Morton, Y. T. J., van Diggelen, F., Spilker, Jr., J. J., Parkinson, B. W., Lo, S., & Gao, G. (Eds.), Position, navigation, and timing technologies in the 21st century: Integrated satellite navigation, sensor systems, and civil applications (Vol. 2, Ch. 39, pp. 1225–1241). Wiley. https://doi.org/10.1002/9781119458555.ch39
- ↵Microchip. (2024). SA65 CSAC datasheet [PDF]. Retrieved December 20, 2024, from https://www.microchip.com/en-us/product/CSAC-SA65
- ↵Misra, P., & Enge, P. (2011). Global Positioning System: Signals, measurements, and performance (Revised 2nd ed.). Ganga-Jamuna Press.
- ↵Morton, Y., Yang, R., & Breitsch, B. (2020). GNSS receiver signal tracking. In Morton, Y. T. J., van Diggelen, F., Spilker Jr., J. J., Parkinson, B. W., Lo, S., & Gao, G. (Eds.), Position, navigation, and timing technologies in the 21st century: Integrated satellite navigation, sensor systems, and civil applications (Vol. 1, Ch. 15, pp. 339–375). Wiley. https://doi.org/10.1002/9781119458449.ch15
- ↵NovAtel. (2024). OEM729 GNSS/GPS receiver boards. Retrieved December 20, 2024, from https://novatel.com/products/receivers/gnss-gps-receiver-boards/oem729
- ↵Psiaki, M. (2021). Navigation using carrier Doppler shift from a LEO constellation: TRANSIT on steroids. NAVIGATION, 68(4), 621–641. https://doi.org/10.1002/navi.438
- ↵Reid, T. G., Neish, A. M., Walter, T., & Enge, P. K. (2018). Broadband LEO constellations for navigation. NAVIGATION, 65(2), 205–220. https://doi.org/10.1002/navi.234
- ↵Reid, T. G., Walter, T., Enge, P. K., Lawrence, D., Cobb, H. S., Gutt, G., O’Conner, M., & Whelan, D. (2020). Navigation from low Earth orbit. In Morton, Y. T. J., van Diggelen, F., Spilker, J. J., Jr., Parkinson, B. W., Lo, S., & Gao, G. (Eds.), Position, navigation, and timing technologies in the 21st century: Integrated satellite navigation, sensor systems, and civil applications (Vol. 2, Ch. 43, Part 1, 1359–1379). Wiley. https://doi.org/10.1002/9781119458555.ch43a
- ↵Rizos, C., & Yang, L. (2019). Background and recent advances in the Locata terrestrial positioning and timing technology. Sensors, 19(8), 1821. https://doi.org/10.3390/s19081821
- ↵Rushanan, J. (2007). The spreading and overlay codes for the L1C signal. NAVIGATION, 54(1), 43–51. https://doi.org/10.1002/j.2161-4296.2007.tb00394.x
- ↵Sherman, J., Arissian, L., Brown, R., Deutch, M., Donley, E., Gerginov, V., Levine, J., Nelson, G., Novick, A., Patla, B., Parker, T., Stuhl, B., Sutton, D., Yao, J., Yates, W., Zhang, V., & Lombardi, M. (2021). A resilient architecture for the realization and distribution of Coordinated Universal Time to critical infrastructure systems in the United States: Methodologies and recommendations from the National Institute of Standards and Technology (NIST) (NIST Technical Note 2187). https://doi.org/10.6028/NIST.TN.2187
