Track-Consistency-Based GNSS RFI Monitoring Using Crowdsourced ADS-B Sensor Networks

arXiv:2607.09700 · eess.SP · Submitted 2026-06-21 · Read on arXiv

Listen

Transcript

Introduction to the show: ident: Electrical Engineering. The day in electrical engineering: signals, systems and the hardware that carries them.

Rosa: We're looking at the paper: "Track-Consistency-Based GNSS RFI Monitoring Using Crowdsourced ADS-B Sensor Networks".

Dev: Here's the gist of the paper. Growing reports of global navigation satellite system (GNSS) radio-frequency interference (RFI) highlight the need for scalable wide-area sensing for situational awareness <ref:2607.09700#pg2>.

Rosa: The proposed framework addresses this by developing a three-stage process to screen position-jump candidates, verify local track consistency to suppress timing artifacts, and group confirmed anomalies into traffic-adaptive multi-aircraft events <ref:2607.09700#pg4>.

Dev: Framework Overview The proposed framework processes raw ADS-B reports into regional anomaly events through three stages <ref:2607.09700#pg2>.

Rosa: Each ADS-B report from aircraft j is denoted by mjk = (tj,k, φj,k, λj,k, hj,k, NICj,k, NACpj/NACp), where k indexes successive reports and the reported latitude longitude and altitude are included <ref:2607.09700#pg3>.

Title and authors: Rosa: So we're talking about how this paper, "Track-Consistency-Based GNSS RFI Monitoring Using Crowdsourced ADS-B Sensor Networks," addresses the issue of GNSS radio frequency interference by using crowdsourced ADS-B receiver networks as a large scale opportunistic sensor network.

Dev: The core idea is that you take raw ADS-B reports and process them through this three stage framework to identify real anomalies that aren't just noise or timing errors.

Taro: So, the first stage identifies position jump candidates, which are any reports where the estimated speed exceeds a certain threshold vth.

Rosa: Right, but those candidates might be caused by actual spatial inconsistencies or they could just be receiver timing artifacts, so that's why they move to stage two for verification.

Dev: Stage two is the local track consistency check, which verifies if each candidate is spatially consistent with neighboring reports from the same aircraft.

Taro: I see. It sounds like they test this by fitting a two dimensional cubic smoothing spline to the local Cartesian positions and checking how far off the candidate position is from that fitted reference track.

Rosa: And they also add a predecessor gap test, which measures the displacement between the candidate and its immediately preceding report to check for sequential continuity.

Dev: So, if both of those checks pass, then you confirm it as an anomaly in set A, meaning you have a position jump that is spatially consistent and temporally continuous.

Taro: That means they're separating actual displacement from receiver timing errors that create an apparent speed spike without any real spatial inconsistency.

Rosa: And that leads us into the next part of what this paper is proposing: how they group these confirmed anomalies into regional multi aircraft events using spatio temporal density based spatial clustering of applications with noise, or ST-DBSCAN.

The paper's summary: Dev: Moving on to the summary, the paper outlines how stage three takes those confirmed anomalies and groups them into regional multi aircraft events defined by a spatiotemporal neighborhood B(an).

Rosa: The neighborhood is defined by spatial radius εs, vertical radius εv, and temporal radius εt. An anomaly an becomes a core point if its neighborhood B(an) has at least MinPts points.

Taro: And they make that minimum point count adaptive for traffic density by modeling MinPts using a Poisson tail based on the hourly background aircraft count Di to estimate how many background supported anomalies you might expect in that search window, μi.

Dev: So, it’s not just a fixed number of points; it dynamically adapts based on local traffic conditions in terms of the expected number of anomalies within the search window.

Rosa: This means they can define what constitutes a significant event even if the area is typically busy or if it's quieter than usual.

Taro: That seems like a robust way to define an event because it accounts for how busy traffic is, which is important when you're dealing with crowdsourced data.

Dev: And the summary emphasizes that this combination of tracking verification and adaptive clustering allows them to exhibit stronger temporal consistency with the validity period of GNSS related Notice to Airmen RKRR Z1401/twenty-five compared to prior methods <ref:2607.09700#pg2,stronger temporal consistency with the validity period of GNSS related Notice to>.

Rosa: That is a key finding, showing that this method is better at maintaining consistency during periods when timing artifacts might be more likely to occur.

The paper's improvements: Taro: Looking specifically at the proposed improvements, the paper points out that their framework provides a way to screen position jump candidates by applying a local track consistency check before clustering.

Dev: This is different from earlier work because they are explicitly verifying whether a candidate remains spatially compatible with the surrounding track despite its apparent speed spike.

Rosa: That’s what suppresses those receiver timing artifacts, which is something previous methods struggled with when they were just looking at quality indicators like NIC or NACp.

Taro: And then there's the traffic adaptive clustering improvement, where they model MinPtsi using a Poisson tail based on the hourly background aircraft count Di to adapt for varying traffic density.

Dev: So, they are addressing two major blind spots identified in related approaches: the lack of trajectory consistency studies and the failure to combine NIC and NACp screening with position jump clustering.

Rosa: They show that this combination is what allows them to exhibit stronger temporal consistency during a NOTAM period, which is a direct improvement over prior position jump clustering methods.

Conclusion: Dev: To wrap up, the paper proposes this three stage framework for identifying GNSS RFI consistent anomaly events and grouping them using ST-DBSCAN with adaptive density modeling.

Rosa: The implication is that track consistency verification can provide a complementary sensing criterion beyond ADS-B quality indicators like NIC and NACp.

Taro: And they show that this method demonstrated stronger temporal consistency with the validity period of NOTAM RKRR Z1401/twenty-five compared to the prior position jump clustering method <ref:2607.09700#pg2,stronger temporal consistency with the validity period of>.

Dev: Plus, they found that ninety-nine point five four percent of confirmed position anomalies during the NOTAM valid period remained in NIC minus NACp high regimes, which suggests that this track consistency verification is a useful addition.

Rosa: So, they conclude that independent GNSS monitoring measurements could further corroborate event timing to make the system even more reliable.

Taro: And for someone just listening to the show, it means we've got a method that filters out timing noise so you can trust the reported position jumps more when you're trying to build a picture of movement.

Department of Electronics Engineering, Chungbuk National University · Department of Aeronautics and Astronautics, Stanford University

eess.SP

Submitted: 2026-06-21

Updated: 2026-08-13

Journal ref: IEEE Sensors Letters, 2026

DOI: 10.1109/LSENS.2026.3743890

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 82/100

The gist: Growing reports of global navigation satellite system (GNSS) radio-frequency interference (RFI) highlight the need for scalable wide-area sensing for situational awareness <ref:2607.09700#pg2>.

Key concepts

Position-Jump Detection: This stage uses a constant-acceleration Kalman Filter on aircraft position data to estimate apparent speed. Reports exceeding a set threshold indicate potential position jumps, which could be real spatial inconsistencies or timing artifacts from receiver errors.
Local Track-Consistency Check: This verifies if a candidate report is spatially consistent with neighboring aircraft tracks. It uses cubic smoothing splines and distance metrics to determine if the candidate deviates too much from the surrounding flight path, filtering out speed spikes caused by timing issues.
Traffic-Adaptive Clustering: The system groups confirmed anomalies using Spatio-Temporal DBSCAN. Crucially, it adjusts the required density for forming an event based on local background traffic levels to ensure detection remains sensitive regardless of how busy the airspace is.
Complementary Sensing Criterion: The framework proves that verifying track consistency provides a useful check beyond standard ADS-B quality indicators like NIC and NACp. High NIC/NACp values during confirmed anomalies suggest the anomaly is real, even if timing errors are present.

Terminology

Consistency — 35× in this paper · explained in 3 episodes
Clustering — 21× in this paper · explained in 2 episodes
Verification — 13× in this paper · explained in 2 episodes
Temporal Consistency — 10× in this paper · explained in 3 episodes
Core — 5× in this paper · explained in 2 episodes
Accuracy — 2× in this paper · explained in 1 episode
Korea Agency for Infrastructure Technology Advancement — 2× in this paper
Localization — 2× in this paper · explained in 2 episodes
Ministry of Education — 2× in this paper
Calibration — 1× in this paper · explained in 9 episodes
Incheon Flight Information Region — 1× in this paper
Navigation Accuracy Category for Position — 1× in this paper
Navigation Integrity Category — 1× in this paper
Open Performance Data Initiative — 1× in this paper

Summary

Growing reports of global navigation satellite system (GNSS) radio-frequency interference (RFI) highlight the need for scalable wide-area sensing for situational awareness <ref:2607.09700#pg2>. The proposed framework addresses this by developing a three-stage process to screen position-jump candidates, verify local track consistency to suppress timing artifacts, and group confirmed anomalies into traffic-adaptive multi-aircraft events <ref:2607.09700#pg4>.

Framework Overview

The proposed framework processes raw ADS-B reports into regional anomaly events through three stages <ref:2607.09700#pg2>. Each ADS-B report from aircraft j is denoted by mjk = (tj,k, φj,k, λj,k, hj,k, NICj,k, NACpj/NACp), where k indexes successive reports and the reported latitude longitude and altitude are included <ref:2607.09700#pg3>. The framework first forms a position-jump candidate set C based on an estimated speed exceeding a threshold vth <ref:2607.09700#pg4>.

Position-Jump Detection

Stage 1 involves applying a constant-acceleration Kalman Filter (CA-KF) to the time-ordered position sequence of each aircraft to yield an apparent horizontal speed estimate v̂j,k <ref:2607.09700#pg4>. Reports whose estimated speed exceeds vth form the position-jump candidate set C = mjk: v̂j,k > vth <ref:2607.09700#pg4>. These candidates may include receiver-timing artifacts as well as actual spatial inconsistencies and are passed to Stage 2 for verification <ref:2607.09700#pg4>.

Local Track-Consistency Check

Stage 2 verifies whether each candidate in C is inconsistent with neighboring reports from the same aircraft <ref:2607.09700#pg4>. This stage applies two sequential tests to separate actual displacement from receiver-timing errors that create an apparent speed spike without spatial inconsistency <ref:2607.09700#pg4>.

A local context Nj,k of N neighboring reports is restricted to a continuous track segment so that reports from different flight phases are not compared <ref:2607.09700#pg4>.
A two-dimensional cubic smoothing spline sjk(-k)(u) is fitted to the local Cartesian positions in Nj,k excluding the candidate mjk to evaluate geometric consistency with the surrounding track.
The local-track residual djk is calculated as the shortest distance from the candidate position to the fitted local reference track. A candidate is treated as a local-track outlier when djk > τtrackj,k.
The predecessor-gap test measures the displacement between the candidate and its immediate preceding report gjk = pj,k - pj,k-1. A candidate satisfies the condition when gjk > τgapj,k.

The set of confirmed anomalies is then defined as A = mjk ∈ C: djk > τtrackj,k ∧ gjk > τgapj,k.

Traffic-Adaptive Multi-Aircraft Clustering

Stage 3 groups the retained anomalies in A into regional multi-aircraft events using Spatio-Temporal Density-Based Spatial Clustering of Applications with Noise (ST-DBSCAN). The spatiotemporal neighborhood B(an) is defined by spatial, vertical, and temporal radii εs, εv, and εt.

The anomaly an is treated as a core point when B(an) ≥ MinPts. To adapt for varying traffic density, MinPts i is modeled using a Poisson tail based on the hourly background aircraft count Di. The expected number of background-supported anomalies within the search window is modeled as μi = Di · S · c.

Experimental Validation

The experimental validation used 606,922,091 reports collected over Northeast Asia from December 2025 through February 2026 to test the framework against a Notice to Airmen (NOTAM) RKRR Z1401/25 validity period. The proposed framework produced 250 event clusters involving 817 unique aircraft during the NOTAM-valid period and no clusters in the pre-NOTAM period. Furthermore, 99.54% of confirmed position anomalies during the NOTAM-valid period remained in NIC-/NACp-high regimes, suggesting that track-consistency verification provides a complementary sensing criterion beyond ADS-B quality indicators.

Conclusion

This letter has proposed a three-stage framework for identifying GNSS-RFI-consistent anomaly events that manifest as abrupt inconsistencies in crowdsourced 1090ES ADS-B reported positions. The framework demonstrated stronger temporal consistency with the validity period of NOTAM RKRR Z1401/25 than the prior position-jump clustering method. Because 99.54%/99.68% of the confirmed anomalies during the NOTAM-valid period remained in NIC-/NACp-high regimes, track-consistency verification can provide a complementary sensing criterion beyond ADS-B quality indicators. Independent GNSS-monitoring measurements could further corroborate event timing.

Acknowledgment

This work was supported in part by the Korea Agency for Infrastructure Technology Advancement (KAIA) grant funded by the Ministry of Land, Infrastructure and Transport (Grant No. RS-2026-25527966) and in part by the Regional Innovation System & Education (RISE) program through the Chungbuk Regional Innovation System & Education Center, funded by the Ministry of Education (MOE) and Chungcheongbuk-do, Republic of Korea (Grant No. 2025-RISE-11-014-03). The work was supported in part by Chungbuk National University Glocal30 project (2025).

References

[1] Trýb J, Hospodka J (2025), “GNSS interference and security: Impacts on critical infrastructure and mitigation strategies,” Procedia Comput. Sci., vol. 253, pp. 2635–2644, doi: 10.1016/j.procs.2025.01.323

[8] Liu Z, Lo S, Blanch J, Walter T (2024), “GNSS spoofing detection and localization using ADS-B data,” in Proc. 37th Int. Tech. Meeting Satell. Division Inst. Navig. (ION GNSS+ 2024), Baltimore, MD, USA, pp. 796–803, doi: 10.33012/2024.19749

[15] 14 C.F.R. § 91.227, “Automatic Dependent Surveillance-Broadcast (ADS-B) Out equipment performance requirements,” Electronic Code of Federal Regulations

[6] Pleninger S, Hospodka J, Steiner J, Lukeš P, Topková T, Pileningová T, Kraus J (2025), “Aircraft resilience to GNSS jamming and impact on ADS-B quality indicators,” Aeronaut. J., vol. 129, no. 1336, pp. 1629–1646, doi: 10.1017/aer.2024.165

[9] Schäfer M, Strohmeier M, Lenders V, Martinovic I, Wilhelm M (2014), “Bringing up OpenSky: A large-scale ADS-B sensor network for research,” in Proc. 13th IEEE/ACM Int. Symp. Inf. Process. Sensor Netw., Berlin, Germany, pp. 83–94, doi: 10.1109/IPSN.2014.

Improvements for AI systems

Track-Consistency Verification for GNSS RFI Monitoring: The improved system can screen position-jump candidates by applying a local track-consistency check to verify whether a candidate remains spatially compatible with the surrounding track despite its apparent speed spike. This suppresses receiver-timing artifacts that cause false speed spikes, thus filtering out timing inconsistencies before event clustering.
Traffic-Adaptive Multi-Aircraft Event Clustering: The system can group confirmed anomalies into events using ST-DBSCAN, adapting the minimum point count based on local background traffic density. This allows the system to define core points based on MinPtsi = max[MinPtsbase, FPoisson−1(1 − α; μi) + 1], ensuring that anomaly detection is sensitive to whether a region is typically busy or quiet.
Complementary Anomaly Screening: The improved framework provides a complementary sensing criterion beyond ADS-B quality indicators like NIC and NACp. This allows the system to identify anomalies where NIC and NACp remained high throughout the event but still confirmed position anomalies, as shown in Fig 4 of the paper.