arxiv: 2605.10264 · v1 · submitted 2026-05-11 · 💻 cs.IT · cs.SY· eess.SY· math.IT

Recognition: 2 theorem links

· Lean Theorem

Low-Cost GNSS Anti-Jamming Through 2-Bit Phase Shift Beamforming with Machine Learning

Burak Soner , Ekin Uzun , Can Aksoy

Authors on Pith no claims yet

Pith reviewed 2026-05-12 05:16 UTC · model grok-4.3

classification 💻 cs.IT cs.SYeess.SYmath.IT

keywords GNSS anti-jamming2-bit phase shiftersbeamformingQPSK weightsmachine learninginterference suppressiondiscrete optimization

0 comments

The pith

Optimizing 2-bit QPSK phase weights with machine learning lets GNSS arrays null strong jammers while keeping satellite gain.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper shows how inexpensive 2-bit phase shifters can perform beamforming for GNSS anti-jamming by restricting each array weight to one of four QPSK phase states. This quantization shrinks the possible beampatterns, so the authors set up a discrete optimization that balances jammer suppression against satellite-direction gain and test several solvers. Simulations across array sizes reveal that exhaustive search reaches 34 dB nulling but is too slow, while a gradient-boosted decision tree followed by local search matches that performance at fixed low latency. Hardware emulation experiments confirm the practical payoff: at 70 dB jamming-to-signal ratio the protected receiver holds 20.8 dB-Hz average C/N0 while an unprotected one falls to 9.3 dB-Hz.

Core claim

By solving a discrete optimization over QPSK phase states, a 2-bit beamformer can trade interference suppression for satellite gain; an oracle solver achieves up to 34 dB null depth, and a gradient-boosted decision tree plus local search approximates oracle performance at constant latency. Experiments with a digital QPSK oracle show that at 70 dB J/S the beamformed receiver sustains 20.8 dB-Hz C/N0 while a standard receiver drops near tracking limits at 9.3 dB-Hz.

What carries the argument

Discrete optimization of four-state QPSK phase weights that maximizes satellite gain while minimizing jammer gain, solved either by combinatorial search or by a gradient-boosted decision tree predictor followed by local refinement.

If this is right

Performance scales upward with array size for both oracle and ML-aided solvers.
The gradient-boosted tree plus local search delivers near-oracle C/N0 at fixed low latency suitable for real-time use.
At 62-70 dB J/S the protected receiver sustains tracking while unprotected receivers fall below usable thresholds.
Hardware emulation reproduces the simulated gains, confirming 4.2 dB C/N0 improvement at 44 dB J/S and much larger gaps at stronger jamming.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

If reliable direction finding can be added at low cost, the same 2-bit hardware could protect consumer GNSS units in vehicles or drones.
The ML predictor might be retrained for other phased-array interference problems beyond GNSS.
Extending the method to joint direction estimation and beamforming would remove the main assumption and broaden applicability.

Load-bearing premise

Satellite and jammer directions of arrival are known or estimated accurately enough to solve the discrete weight optimization in real time.

What would settle it

A test in which direction-of-arrival estimates contain a few degrees of error and the resulting 2-bit beamformer loses enough C/N0 that tracking fails at 70 dB J/S would refute the practical claim.

Figures

Figures reproduced from arXiv: 2605.10264 by Burak Soner, Can Aksoy, Ekin Uzun.

**Figure 2.** Figure 2: Experimental GNSS C/N0 results comparing the proposed QPSK oracle beamformer against a GNSS receiver without beamforming under mild, moderate and high jamming levels (J/S ≈ 44, 62, and 70 dB). Average C/N0 under no jamming for this location / orientation is ≈ 37.6 dB-Hz, indicating a challenging reception condition due to setup isolation. DECLARATION Generative artificial intelligence tools were used only … view at source ↗

read the original abstract

We investigate low-cost GNSS anti-jamming using beamforming with inexpensive 2-bit phase shifters, constraining each complex array weight to one of four QPSK phase states (real/imaginary = -1 or +1). This severe quantization sharply limits the beampattern solution space, making conventional real-valued beamforming and naive weight quantization highly suboptimal. We formulate a discrete optimization that trades interference suppression against satellite-direction gain, and benchmark known combinatorial optimization methods across array sizes and interference conditions. Simulations show that performance improves with array size, with oracle and greedy search achieving up to 34 dB nulling, but oracle incurs exponential latency and greedy sampling is stochastic. To obtain deterministic low-latency performance, we propose an ML-aided method based on gradient-boosted decision trees followed by local search, which performs similar to the oracle for larger arrays at fixed latency. We further validate the approach experimentally using a fully digital emulation of the QPSK oracle beamformer and compare against a GNSS receiver without beamforming capability. Under mild jamming (J/S approximately 44 dB) both receivers maintain adequate tracking, with QPSK yielding a 4.2 dB higher average C/N0 (37.3 vs. 33.1 dB-Hz). Under moderate and strong jamming (J/S approximately 62-70 dB) the benefit is substantial. At J/S = 70 dB the unprotected receiver degrades to near tracking limits (avg C/N0 = 9.3 dB-Hz) while the QPSK oracle sustains an average C/N0 of 20.8 dB-Hz. These results confirm that 2-bit phase-shift beamforming provides considerable anti-jamming benefit over a standard GNSS receiver, motivating further research on oracle-level practical methods.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The work offers a practical ML-based solver for 2-bit GNSS beamforming that delivers clear anti-jamming gains in tests, though it relies on known directions of arrival.

read the letter

The paper shows that 2-bit phase shift beamforming with a machine learning solver can maintain usable GNSS signal quality under strong jamming in both simulation and hardware emulation. The gains are practical, but they depend on knowing the directions of arrival for the satellites and jammer in advance. What stands out is the concrete application to GNSS. They set up a discrete optimization over the four possible QPSK states per element to balance jammer nulling against satellite gain. They benchmark combinatorial methods, then introduce gradient-boosted trees followed by local search to get deterministic, low-latency results close to the oracle. For larger arrays this works well. The emulation part is also direct: they implement the QPSK oracle digitally and compare C/N0 against a standard receiver. At 70 dB J/S the difference is clear, from 9.3 dB-Hz unprotected to 20.8 dB-Hz with the beamformer. The soft spot is the DOA assumption. The optimization uses known angles, and the stress test is right that at high J/S the signals are too weak for easy direction finding. The paper does not show results with estimated directions or discuss how to get them in jammed conditions. This is a real gap for field use, though the simulation results still hold under the stated conditions. The ML training details are light in the abstract, and there are no error bars mentioned, but the overall pattern of improvement across conditions looks consistent. The citation pattern seems standard for the area. This is useful for GNSS hardware designers and engineers looking at low-cost anti-jamming solutions. Readers interested in applied array processing will find the solver comparison and the emulation data helpful. It is grounded enough in reproducible simulations and emulation to merit serious peer review. I would send it to referees with a note to examine the DOA estimation issue and request more on the training procedure.

Referee Report

1 major / 3 minor

Summary. The paper proposes low-cost GNSS anti-jamming via 2-bit phase-shift beamforming, where each array weight is restricted to one of four QPSK states. It formulates a discrete optimization trading jammer null depth against known satellite directions, benchmarks combinatorial solvers, and introduces a gradient-boosted decision-tree model followed by local search to achieve near-oracle performance at fixed low latency. Simulations report up to 34 dB nulling that scales with array size; a fully digital hardware emulation shows the QPSK oracle maintaining 20.8 dB-Hz average C/N0 at J/S ≈ 70 dB versus 9.3 dB-Hz for an unprotected receiver.

Significance. If the reported C/N0 gains hold under realistic conditions, the work demonstrates that inexpensive 2-bit phase shifters can deliver substantial anti-jamming benefit for GNSS, addressing a practical vulnerability. The ML approximation for deterministic low-latency execution and the inclusion of hardware emulation are concrete strengths that distinguish the contribution from purely theoretical discrete beamforming studies.

major comments (1)

[Abstract] Abstract and the optimization formulation: the headline result (QPSK oracle sustaining 20.8 dB-Hz C/N0 at J/S ≈ 70 dB) is obtained by solving the discrete weight optimization that explicitly requires exact satellite and jammer DOA vectors. No section evaluates performance when these directions must be estimated from the jammed snapshots; at the cited J/S levels the GNSS signals lie below the noise floor, rendering conventional estimators unreliable without additional side information or assumptions.

minor comments (3)

The abstract and method sections provide only high-level descriptions of the decision-tree training; adding dataset size, feature definitions, hyperparameter selection, and any cross-validation protocol would improve reproducibility.
Reported C/N0 figures lack error bars or trial-to-trial standard deviations, making it difficult to judge the statistical reliability of the 4.2 dB and 11.5 dB gains cited for mild and strong jamming.
Clarify the precise array sizes and number of satellites used in both the simulation benchmarks and the emulation experiment to allow direct comparison with prior array-processing literature.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the constructive feedback and for acknowledging the practical relevance of 2-bit phase-shift beamforming for GNSS anti-jamming. We address the major comment below and will revise the manuscript to better contextualize our assumptions and results.

read point-by-point responses

Referee: [Abstract] Abstract and the optimization formulation: the headline result (QPSK oracle sustaining 20.8 dB-Hz C/N0 at J/S ≈ 70 dB) is obtained by solving the discrete weight optimization that explicitly requires exact satellite and jammer DOA vectors. No section evaluates performance when these directions must be estimated from the jammed snapshots; at the cited J/S levels the GNSS signals lie below the noise floor, rendering conventional estimators unreliable without additional side information or assumptions.

Authors: We agree that the optimization formulation and all reported results, including the headline C/N0 values from both simulation and hardware emulation, assume exact knowledge of satellite and jammer directions of arrival (DOAs). The manuscript does not evaluate performance when DOAs are estimated from the received snapshots, which is a valid and important limitation given that GNSS signals are below the noise floor at the cited J/S levels. Our work is positioned as an investigation of the anti-jamming capability achievable with inexpensive 2-bit phase shifters under ideal DOA information (e.g., from auxiliary sensors or prior knowledge), with the ML model providing a deterministic, low-latency approximation to the oracle. We will revise the abstract to explicitly note the known-DOA assumption and add a short discussion paragraph on the challenges of DOA estimation under strong jamming, along with suggestions for future integration with robust estimators or side information. This revision will clarify the scope without altering the core technical contributions. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical C/N0 results obtained from independent simulations and emulation

full rationale

The paper reports C/N0 performance from running GNSS receiver simulations and a fully digital hardware emulation under controlled jamming scenarios. These measurements apply the proposed discrete optimization (or ML approximation) to generate weights and then evaluate tracking metrics on the resulting signals; the reported values (e.g., 20.8 dB-Hz vs. 9.3 dB-Hz at J/S=70 dB) are not algebraically or statistically forced by reusing the same data that trained the trees or solved the oracle. No equations, self-citations, or uniqueness theorems are invoked that reduce the central claims to tautologies or fitted inputs. The DOA-knowledge assumption affects applicability but does not create a definitional loop inside the derivation chain.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The approach rests on standard array signal processing models and off-the-shelf ML tools; no new physical entities or ad-hoc constants are introduced.

axioms (1)

domain assumption Satellite and jammer directions of arrival are known or estimable to sufficient accuracy for weight optimization.
The discrete optimization and ML predictor both require these directions as inputs.

pith-pipeline@v0.9.0 · 5643 in / 1228 out tokens · 54130 ms · 2026-05-12T05:16:38.099363+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

We formulate a discrete optimization that trades interference suppression against satellite-direction gain... max α|w^H a_g|^2 − (1−α)w^H R̂ w, w ∈ Q^N / √(2N)
IndisputableMonolith/Foundation/AlexanderDuality.lean alexander_duality_circle_linking unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

constraining each complex array weight to one of four QPSK phase states... exhaustive oracle... GBDT + Local Refinement

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

11 extracted references · 11 canonical work pages

[1]

VLSI systolic arrays for adaptive nulling [radar]

Rader, Charles M. “VLSI systolic arrays for adaptive nulling [radar].” IEEE Signal Processing Magazine 13.4 (2002): 29-49

work page 2002
[2]

A comparison of efficient beamforming algorithms

Mucci, Ronald. “A comparison of efficient beamforming algorithms.” IEEE Transactions on Acoustics, Speech, and Signal Processing 32.3 (2003): 548-558

work page 2003
[3]

Hybrid precoder and combiner design with low- resolution phase shifters in mmWave MIMO systems

Wang, Zihuan, et al. “Hybrid precoder and combiner design with low- resolution phase shifters in mmWave MIMO systems.” IEEE Journal of Selected Topics in Signal Processing 12.2 (2018): 256-269

work page 2018
[4]

”Efficient algorithms for constant-modulus analog beamforming

Arora, Aakash, et al. “”Efficient algorithms for constant-modulus analog beamforming.” IEEE Transactions on Signal Processing 70 (2021): 756- 771

work page 2021
[5]

Beamforming Gain with Nonideal Phase Shifters

Do, Heedong, and Angel Lozano. “Beamforming Gain with Nonideal Phase Shifters.” arXiv preprint arXiv:2601.09426, 2026

work page arXiv 2026
[6]

A Hardware-Efficient Hybrid Approach for Suppression of Multiple Jammers in GNSS Receivers

Drenkhahn, Kevin E., Markus Landmann, Ali Rashidifar, Christoph W. Wagner, and Renato Zea Vintimilla. “A Hardware-Efficient Hybrid Approach for Suppression of Multiple Jammers in GNSS Receivers.” preprint, 2024

work page 2024
[7]

MUSE-a systolic array for adaptive nulling with 64 degrees of freedom, using Givens transformations and wafer scale integration

Rader, Charles M., et al. “MUSE-a systolic array for adaptive nulling with 64 degrees of freedom, using Givens transformations and wafer scale integration”. No. TR886. 1990

work page 1990
[8]

A linearly convergent doubly stochastic Gauss–Seidel algorithm for solving linear equations and a certain class of over-parameterized optimization problems

Razaviyayn, Meisam, et al. “A linearly convergent doubly stochastic Gauss–Seidel algorithm for solving linear equations and a certain class of over-parameterized optimization problems.” Mathematical Program- ming 176.1 (2019): 465-496

work page 2019
[9]

Hybrid beamforming/combining for millimeter wave MIMO: A machine learning approach

Chen, Jienan, et al. “Hybrid beamforming/combining for millimeter wave MIMO: A machine learning approach.” IEEE Transactions on Vehicular Technology 69.10 (2020): 11353-11368

work page 2020
[10]

Xgboost: A scalable tree boosting system

Chen, Tianqi, and Carlos Guestrin. “Xgboost: A scalable tree boosting system.” Proceedings of the 22nd ACM SIGKDD international confer- ence on knowledge discovery and data mining. 2016

work page 2016
[11]

GPS-SDR-SIM,

OSQZSS, “GPS-SDR-SIM,” 2023. [Online]. Available: https://github.com/osqzss/gps-sdr-sim

work page 2023