Jamming Source Localization Using Augmented Physics-Based Model

Andrea Nardin, Tales Imbiriba, Pau Closas

Abstract

Monitoring interferences to satellite-based navigation systems is of paramount importance in order to reliably operate critical infrastructures, navigation systems, and a variety of applications relying on satellite-based positioning. This paper investigates the use of crowdsourced data to achieve such detection and monitoring at a central node that receives the data from an arbitrary number of agents in an area of interest. Under ideal conditions, the pathloss model is used to compute the Cramér-Rao Bound of accuracy as well as the corresponding maximum likelihood estimator. However, in real scenarios where obstructions and reflections are common, the signal propagation is far more complex than the pathloss model can explain. We propose to augment the pathloss model with a data-driven component, able to explain the complexities of the propagation channel. The paper shows a general methodology to jointly estimate the interference location and the parameters of the augmented model, showing superior performances in complex scenarios such as those encountered in urban environments.

Introduction

Global Navigation Satellite Systems (GNSS) like GPS are the backbone of modern positioning and timing, yet they are incredibly fragile. GNSS signals arrive at Earth with very low power, making them easily drowned out by "jammers"—malicious or accidental transmitters operating on the same frequency. To mitigate these threats, accurate localization is required.

A common approach is crowdsourcing: using received signal strength (RSS) measurements from many agents, such as smartphones or connected cars. Traditionally, researchers use a physics-based pathloss model to estimate the source location. However, in urban environments, buildings cause reflections (multipath) and blockages (shadowing) that simple physics equations cannot account for. Conversely, pure "black-box" neural networks require massive amounts of data to learn these complexities. This paper proposes the Augmented Physics-Based Model (APBM) to provide a "best of both worlds" solution.

Methodology

The APBM models the received signal power $y_n$ as a combination of a physics-based term and a data-driven correction: $$y_n = f(x_n; \theta) + g(x_n; \phi) + \xi_n$$

  • $f(x_n; \theta)$ (The Physics): A modified pathloss model providing the general structure of signal decay based on the jammer's location $\theta$.
  • $g(x_n; \phi)$ (The Data): A feed-forward neural network (NN) acting as a nonlinear correction term for urban complexities.
  • $\xi_n$: Measurement noise.

Key innovations include Joint Optimization, where the jammer's location and the neural network's weights are estimated simultaneously, and Singularity Avoidance, which uses a "far-field distance" constraint to stabilize the optimization landscape near the source. Additionally, $\ell_2$ regularization is applied to the NN parameters to ensure the physics model remains the primary driver of the estimation.

Experimental Setup

The study utilizes high-fidelity simulations to validate the model, specifically focusing on a complex urban scenario generated using ray-tracing techniques. This environment accounts for 3D geometry and up to four signal reflections, creating a highly irregular signal power surface.

Signal power distribution in a simulated urban environment

To mimic a realistic, bandwidth-constrained system, the simulation uses 10,000 observations over a 1 $km^2$ area, but only the 15 most powerful observations are utilized for the actual localization process.

Results and Discussion

In ideal "open-sky" conditions, the APBM matches the performance of the Maximum Likelihood Estimator (MLE) and approaches the theoretical Cramér-Rao Bound (CRB). The model's advantage is most pronounced in the urban scenario, where standard models fail due to environmental mismatch.

RMSE performance comparison in an urban scenario

As shown in the performance metrics, the APBM significantly outperforms both the standard pathloss model and pure neural networks. While the physics-only model's error remains high due to model mismatch, the APBM achieves much lower Root Mean Square Error (RMSE), approaching 10-20m as the Interference-to-Noise Ratio (INR) increases. Furthermore, the "$P_0$-blind" variant demonstrates that the APBM can accurately localize a source even when the jammer's transmit power is unknown, as the neural network component learns to compensate for the missing power offset.

Conclusion

The Augmented Physics-Based Model successfully hybridizes physical principles with data-driven flexibility. By using physics as a global prior and a neural network for local environmental corrections, the system achieves high localization accuracy in challenging urban environments with very limited data. Future research will focus on adapting the model for moving jammers and incorporating privacy-preserving techniques like Federated Learning.

Summary updated

Abstract

Monitoring interferences to satellite-based navigation systems is of paramount importance in order to reliably operate critical infrastructures, navigation systems, and a variety of applications relying on satellite-based positioning. This paper investigates the use of crowdsourced data to achieve such detection and monitoring at a central node that receives the data from an arbitrary number of agents in an area of interest. Under ideal conditions, the pathloss model is used to compute the Cramér-Rao Bound of accuracy as well as the corresponding maximum likelihood estimator. However, in real scenarios where obstructions and reflections are common, the signal propagation is far more complex than the pathloss model can explain. We propose to augment the pathloss model with a data-driven component, able to explain the complexities of the propagation channel. The paper shows a general methodology to jointly estimate the interference location and the parameters of the augmented model, showing superior performances in complex scenarios such as those encountered in urban environments.
BibTeX
@inproceedings{nardin2023jamming,
  author = {Andrea Nardin and Tales Imbiriba and Pau Closas},
  title = {{Jamming Source Localization Using Augmented Physics-Based Model}},
  booktitle = {IEEE International Conference on Acoustics, Speech and Signal Processing},
  year = {2023},
  doi = {10.1109/icassp49357.2023.10095731},
}
Figure