Three projects on GNSS-free navigation using Phased-array Radio System

See this post on the NTNU TekNat blog for more context.

Background

Today, Global Navigation Satellite System (GNSS) is used in a myriad of navigation applications, such as drones, but also manned aircraft. The last years have also shown an increase in jamming/spoofing of GNSS signals, with the ultimate consequence of loss of vehicle and personnel. Examples include the Iran-U.S. RQ-170 incident where Iranian forces sucessfully captuered a U.S. drone, spoofing near Russian military bases and jammers causing problems for Norsk luftambulanse. Before unmanned aerial vehicles can be used for safety critical applications, this vulnerability must be addressed. One way avoid the GNSS vulnerabilities is to rely on position measurements from another source, such as a Phased Array Radio System (PARS). A PARS base antenna consists of multiple small antennas that are positioned in a two-dimensional array. Through timing and estimation of the antenna characteristics, an estimate of the spherical position (elevation angle, azimuth angle and range) of the vehicle can be found, illustrated below.

Phased-array antennas

The figure below illustrates how the differences in the phase of the received signals at the different antenna locations can be used to determine the direction of the incoming signal, in one dimension. This leads to an electronically steerable highly directional antenna system. Unlike conventional mechanically steerable directional antennas, the PARS system is small, powerful and agile, and can send and receive in multiple directions at the same time in order to serve multiple moving nodes. The inherent beam-forming in the PARS can be used to locate the radio nodes onboard UAVs with a high accuracy.

Presently, a navigation solution capable of providing position, velocity and attitude estimates that is used by a UAV in closed loop flight, has been developed. Here estimated azimuth angle, elevation angle and range from a “black-box” Phased-array radio system is used to aid inertial navigation.

Scope project 1: Multiple-hypothesis GNSS-free navigation using phased-array radio measurement

In a way, the phased array radio system produce a “heat map” of where the UAV is located. If this heat map has one distinct peak, it is straight-forward to find the position, as the Gaussian distribution of the Kalman filter is well suited to explain the uncertainty in the position measurement. However, in situations with multiple peaks (in other words; there is more than one position measurements that could correspond to the true location of the UAV), one has to proceed with caution! The goal with this project is to further improve this system, by utilizing the low-level signal strengths from each antenna in the array to determine the quality/uncertainty of the measurements and to improve performance in multipath conditions.

Proposed tasks, project 1

1. Investigate different approaches/algorithms that can mitigate effects of multipath when the phased array system produce multiple, distinct position measurements. Some suggestions:
   • the naïve approach: give all the position measurements to the Kalman filter, and “gamble” that a simple outlier-rejection will reject the multipath measurements and accept the measurement corresponding to the true position.
   • Weighted-fix, ala probabilistic data association filter (PDAF), (Groves 2013, chap 3.4.5): use all the measurements, but weigh them according to their likelihood and use these to update a single state estimate.
   • particle filter (Gustafsson 2012), which represent the posterior distributions by a set of particles with an associated weight, and thus inherently can represent multi-modal distributions.
   • Multiple-hypothesis Kalman filter (MHKF): use a bank of filters that maintain a set of l state vectors and covariance matrix hypothesis that are propagated independently.
   • probabilistic hypothesis density filter
   • RANSAC (Niedfeldt and Beard 2014)
2. Familiarize with basic principles behind phased-array radio
3. Hypothesize what algorithm(s) are best suited, and what simplifications that can be made for PARS, compared to other multi-hypothesis tracking problems such as radar.
4. Implement the algorithm(s) in matlab
5. Test on exisiting datasets or in a simulator
6. Conclude the work in a written report

Possible extensions for the master thesis include

• Conduct experiments to demonstrate the effectiveness of the algorithm. The UAVlab has a large variety of different appropriate UAVs, as well as pilots and facilities to conduct the experiments
• Based on the results from the project thesis, it could be of interest to include other algorithms in the comparison

Scope project 2: Direction-of-Arrival estimation for PARS

The goal of this project is to investigate current state-of-the-art algorithms for estimating the elevation angle, azimuth angle and range, known as direction-of-arrival estimation. It is of particular interest to investigate how these algorithms can be extended to use knowledge of the motion of the UAV, from e.g. an onboard accelerometer.

Proposed tasks, project 2

1. Investigate existing algorithms for Direction-of-Arrival (DoA) estimation
2. Comparison of a selection of these algorithms in simulations small simulations
3. Demonstrate the effectiveness of the algorithm on existing data from flight experiments.
4. Conclude the work in a written report

This work can be extended into a master thesis, focusing on integrating the DoA estimates with other sensors in e.g. a Kalman filter.

Scope project 3: Robust navigation using PARS and GNSS

Luckily, GNSS is available and can be trusted most of the time. Therefore it is of interest to use GNSS when it is available, and deemed reliable, both in the navigation itself, but also to improve the quality of the PARS-based navigation. Then, in the event of GNSS interference, the PARS-based estimates will be of high quality and will be able to detect the reduced quality of the GNSS measurements. As the PARS-based navigation is relative to the ground antenna, the quality of the global position estimates hinges on precise knowledge of the position and orientation of the base antenna (Gryte et al. 2019) and (Gryte, Bryne, and Johansen 2021 ). It is therefore of interest to investigate methods for determining the position and orientation of the base antenna, both with and without GNSS/compass, and before and after takeoff.

Proposed tasks, project 3

1. Investigate methods for determining the position and orientation of the base antenna, both with and without GNSS/compass, and before and after takeoff.
2. Implementation and proof-of-concept testing based on simulations or recorded data
3. Conclude the work in a written report

This work can be extended into a master thesis, focusing on methods for determining when GNSS can be trusted and not. Also, the methods developed during the project thesis could be extended and demonstrated in a flight demonstration. The UAVlab has a large variety of different appropriate UAVs, as well as pilots and facilities to conduct the experiments.

Prerequisites

This is a list of recommended prerequisites, more to signal what it will involve than to be used as a filter on candidates.

• Project 1: Strong theoretical background and interest in statistics for estimation/sensor fusion, through e.g. TTK4250 Sensor Fusion (which also could be taken in parallel with the project, instead of the two 3.75 credit specialization courses).
• Project 2: Strong theoretical background in signal processing, such as TTT4120 - Digital Signal Processing
• Project 3: Hardware know-how, Kalman filter

Contact

Contact supervisor Kristoffer Gryte for more information.

References

For 1: Blackman, S. S. Multiple hypothesis tracking for multiple target tracking. AES 2004. For 2: Schmidt, R. Multiple emitter location and signal parameter estimation, 1986

Groves, P. D. 2013. Principles of GNSS, Inertial, and Multisensor Integrated Navigation Systems. Artech House.

Gryte, Kristoffer, Torleiv Håland Bryne, and Tor Arne Johansen. 2021. “Unmanned Aircraft Flight Control Aided by Phased-Array Radio Navigation.” Journal of Field Robotics.

Gryte, Kristoffer, Torleiv H. Bryne, Sigurd M. Albrektsen, and Tor A. Johansen. 2019. “Field Test Results of GNSS-Denied Inertial Navigation Aided by Phased-Array Radio Systems for UAVs.” In 2019 International Conference on Unmanned Aircraft Systems (ICUAS), 1398–1406. https://doi.org/10.1109/ICUAS.2019.8798057.

Gustafsson, Fredrik. 2012. Statistical Sensor Fusion. 2nd ed. Studentlitteratur.

Niedfeldt, P. C., and R. W. Beard. 2014. “Multiple Target Tracking Using Recursive RANSAC.” In 2014 American Control Conference, 3393–8. https://doi.org/10.1109/ACC.2014.6859273.