Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

Angle of Arrival Estimation Using SRS in 5G NR Uplink Scenarios

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read This paper claims that a single 5G base station with a three-element antenna array can estimate the angle of arrival of an uplink user's Sounding Reference Signal with less than 2 degrees of error in open-field line-of-sight conditions…

desk verdict A useful testbed paper with honest limitations; the <2° accuracy claim needs phase-calibration validation and proper error statistics. read the letter →

arxiv 2411.16501 v1 pith:5OSNNNWO submitted 2024-11-25 eess.SP

Thodoris Spanos , Fran Fabra , José A. López-Salcedo , Gonzalo Seco-Granados , Nikos Kanistras , Ivan Lapin , Vassilis Paliouras This is my paper · ORCID
classification eess.SP
keywords AngleofArrival5GNRSoundingReferenceSignalMUSICESPRITJADE-ESPRITUniformLinearArraySoftwareDefinedRadio
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper claims that a single 5G base station, equipped with a three-element uniform linear array, can estimate the angle of arrival of a user's uplink Sounding Reference Signal with an error below 2 degrees in open-field line-of-sight conditions. The claim is supported by ray-tracing simulations of the test environment and by field trials at 15 and 50 meters in the 2.4 GHz and 3.5 GHz bands. The authors compare three super-resolution algorithms, MUSIC, ESPRIT, and JADE-ESPRIT, and find comparable performance across tested angles from 0 to 45 degrees. If accurate, this makes single-node positioning in GNSS-denied areas a practical option with commercial software-defined radios rather than specialized hardware.

What carries the argument

The load-bearing machinery is the three-element uniform linear array phase-aligned through a two-stage calibration: offline per-pair phase offsets from a splitter-injected tone, and real-time inter-pair alignment using a common signal, as described in Algorithm 1. On this aligned array, the SRS pilots are extracted after cross-correlation timing synchronization, and each of MUSIC, ESPRIT, and JADE-ESPRIT is applied to the frequency-domain snapshot to produce one angle estimate per second. The steering-vector relation in Eq. (3) is what connects the measured inter-antenna phase differences to the azimuth angle.

What would settle it

Run a static 0-degree test at 15 meters while time-stamping the calibration tone's differential phase every second for 60 seconds; if the angle estimate moves by more than 2 degrees when the correction is held fixed, or if the residual phase changes by more than the equivalent of 2 degrees of angle of arrival, the accuracy claim collapses. Alternatively, compare field estimates against a surveyed ground-truth angle from a theodolite or RTK position.

Watch

Extended reading notes

Core claim

The paper's central discovery is that a phase-calibrated software-defined radio receiver, using only three of its four channels, can estimate the angle of arrival from 5G SRS pilots with consistent sub-2-degree accuracy over a 60-second static test. The authors attribute this to compensating for the receiver's two daughterboard local oscillators by injecting a common signal and measuring the differential phase in real time, combined with the known steering-vector model in Eq. (3). Simulations with a ray-tracing model of the same field, including a single ground reflection and a nearby building, show all three algorithms behaving similarly, and the field tests confirm this pattern.

Load-bearing premise

The whole measurement chain assumes the phase offsets measured during calibration stay fixed for the 60 seconds of each test; if either local oscillator drifts, the steering-vector model is wrong and every angle estimate becomes biased.

Editorial extensions

If this is right

  • A 5G base station with just three antenna elements can localize an uplink user in angle well enough for single-node positioning in open environments.
  • The existing 5G Sounding Reference Signal can serve as the positioning signal, so operators do not need a separate reference signal or dedicated hardware.
  • The comparable performance of MUSIC, ESPRIT, and JADE-ESPRIT means a simpler closed-form estimator such as ESPRIT may be sufficient in line-of-sight conditions.
  • The same testbed can be extended to higher bands such as 5.8 GHz, as the authors state is their ongoing work.
  • Real-time phase calibration between daughterboard channel pairs is a practical solution for coherent processing on multi-channel software-defined radios.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The 2-degree figure is reported for static, open-field line-of-sight with a single dominant path; in dense multipath or with moving users the accuracy would likely degrade, since three antennas can resolve at most two sources.
  • A fair test of the calibration assumption would be to repeat a 0-degree run while logging the estimated phase offset every second; if the offset drifts, the central accuracy claim would need to be re-baselined.
  • The paper's alignment uncertainty from manual transmitter placement could be separated from estimator error by using a surveyed ground-truth position, such as from RTK GPS, as reference instead of the nominal angle.
  • The ray-tracing model with one-bounce reflections could be extended to include diffuse scattering or vegetation to see how robust the sub-2-degree performance is in less ideal open fields.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. The paper describes a software-defined-radio testbed for Angle of Arrival (AoA) estimation using the Sounding Reference Signal (SRS) in 5G NR uplink scenarios. The system uses an Ettus E312 transmitter and an Ettus N310 receiver with a three-element ULA, and the study covers MUSIC, ESPRIT, and JADE-ESPRIT algorithms. The authors present MATLAB ray-tracing simulations for the field environment and report field trials at 2.4 GHz and 3.5 GHz in LOS open-field conditions at 15 m and 50 m. The central claim is that the testbed achieves an accuracy of less than 2 degrees of error, with consistent results over 60-second tests.

Significance. If the performance claim is rigorously established, the paper documents a useful practical platform for single-base-station 5G positioning, including a full SRS-based uplink AoA processing chain and a phase-calibration strategy for a multi-channel USRP. The strengths are the detailed system description, the use of realistic 5G waveforms, and the availability of field measurements rather than only simulations. However, the significance is presently limited by the lack of error bars, statistical aggregation, and calibration validation, which prevents the reader from assessing whether the claimed sub-2-degree accuracy is actually demonstrated.

major comments (4)
  1. [Section 4.2, Figures 6–8] The central claim of 'less than 2 degrees of error' is not supported by the presented evidence. The paper shows only time-series plots over 60 seconds, without reporting per-test summary statistics such as mean error, standard deviation, RMSE, or bias for each angle, distance, and algorithm. Since each test provides 300 AoA estimates (5 per second over 60 seconds), the authors should provide a statistical table or error bars to substantiate the claimed accuracy and to compare the three algorithms quantitatively.
  2. [Section 3.1 and Algorithm 1] The phase calibration procedure is load-bearing for the measured accuracy, but no calibration validation is reported. The offline 1-to-4 splitter calibration and the real-time differential-LO phase correction assume that the phase misalignment remains constant over the 60-second test and that the calibration tone path matches the antenna path. Without a hold-time test, a residual-phase-error measurement, or a sensitivity analysis, a small inter-daughterboard phase bias can directly bias MUSIC/ESPRIT estimates, especially with only M=3 antennas. The authors should quantify the residual phase error after the calibration procedure.
  3. [Section 4.1] The simulation setup is under-specified. The paper does not report the SNR values used in the Monte-Carlo simulations, the number of sources D assumed in MUSIC/ESPRIT/JADE-ESPRIT, or the exact ray-tracing parameters. Without this information, the simulation results in Figures 4 and 5 cannot be reproduced, and the discrepancy with the field results noted in Section 4.2 cannot be evaluated. The authors should state the chosen D and SNR values and provide a direct simulation-measurement comparison under comparable operating conditions.
  4. [Section 2.3.2 and Algorithm 1] The choice of the number of sources D and the outlier removal rule (3 scaled MAD) are user-defined and not specified for the field tests. The reported accuracy depends on these choices; for instance, an incorrect D in MUSIC can produce false peaks. The authors should state the values used for D across the tests and discuss the sensitivity of the reported accuracy to the outlier threshold.
minor comments (5)
  1. [Eq. (6)] The covariance matrix should be E[YY^H] rather than E[YY^T], since the received signal matrix is complex and the Hermitian transpose is required for a valid covariance matrix.
  2. [Eq. (5)] The summation index in the cross-correlation runs from i=1 to Lseq-1; please verify whether the index should start at i=0 or whether the upper limit is intended to be Lseq.
  3. [Figure 6 caption] The legend text '2D ESPRITMUSICESPRIT' lacks spacing between the algorithm names; it should read '2D ESPRIT, MUSIC, ESPRIT'.
  4. [Section 4.2] The text states that the angle of arrival for the 2.4 GHz tests was fixed at 0 degrees, but Figures 7 and 8 show estimates around 1.5 to 2 degrees and -1 to -2 degrees, respectively; the authors should quantify the precision of the ground-truth angle setup and the positional alignment uncertainty.
  5. [Section 4.2] The statement that 'the discrepancy between the simulation results and actual measurements may seem significant' is vague; please provide a quantitative comparison of simulated versus measured error statistics to make the discussion concrete.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the AoA algorithms and ground-truth angles are external, and the phase-calibration loop is an engineering calibration, not a self-derived prediction.

full rationale

The paper's central claim is an empirical accuracy figure for standard MUSIC/ESPRIT/JADE-ESPRIT AoA estimation on an SDR testbed. The algorithms are external, well-known methods (Schmidt 1986; Roy and Kailath 1989; van der Veen et al. 1998), and the ground-truth angles are set physically (0° through 45° at 15 m and 50 m), not derived from the algorithms or from a fitted model. The phase-calibration procedure in Section 3.1 and Algorithm 1 estimates hardware phase offsets using injected tones; these offsets are input corrections to the steering-vector model of Eq. (3), not the AoA output itself. The cited [13] is a self-citation by overlapping authors regarding N310 phase misalignment, but it concerns externally falsifiable hardware behavior and is not used to prove the reported sub-2-degree accuracy. The paper explicitly separates simulation, which assumes perfect calibration (Section 4.1), from measurements, and it reports residuals rather than fitting the model to the data. No equation is defined in terms of the quantity it predicts, and no fitted parameter is renamed as a prediction. The absence of a residual-phase-error or hold-time test is a correctness and robustness limitation, not a circularity. Thus the derivation chain is self-contained with respect to circularity.

Assumptions & free parameters 3 free parameters · 8 assumptions · 0 invented entities

The central claim rests on standard array-processing mathematics, the correctness of the SRS waveform, a calibrated three-element ULA, and a ray tracing model of the field. No new physical entities are introduced. The most fragile unstated inputs are the source number D, the simulation SNR, and the long-term validity of the phase calibration.

free parameters (3)
  • Number of sources D = not reported
    MUSIC and ESPRIT need the signal-subspace dimension D (Section 2.3.2); the paper never says how D is chosen for the three-element array, and the angle estimates depend on it.
  • Outlier removal threshold = 3 scaled median absolute deviations
    Algorithm 1 step 13 discards estimates more than 3 scaled MADs from the median before averaging; this hand-set threshold directly affects the reported accuracy.
  • Simulation SNR = not stated
    Section 4.1 says AWGN was added for given SNR values but the SNR values are not reported, so the simulated RMSE curves in Figs. 4 and 5 cannot be reproduced or compared with field conditions.
assumptions (8)
  • standard math Eigenvalue decomposition and subspace orthogonality underpin MUSIC and ESPRIT.
    Used in Section 2.3.2; standard linear algebra, no independent evidence needed.
  • domain assumption Uniform linear array with identical elements and known spacing d.
    The steering vector in Eq. (3) assumes an ideal ULA; mutual coupling, gain and phase ripple, and the actual element spacing are not characterized.
  • domain assumption SRS waveform follows the 3GPP TS 38.211 mapping.
    Waveform generation and subcarrier allocation follow reference [6]; deviation in the testbed would affect cross-correlation timing and channel estimates.
  • domain assumption Phase calibration of the N310 channels remains valid during the measurement.
    Section 3.1 and Algorithm 1 step 7 assume the differential LO phase measured from a common injected signal stays constant across a 60 second test; no hold-time or residual-phase evaluation is given.
  • domain assumption Ray tracing model with at most one bounce captures the field environment.
    Section 2.2 uses SBR with one bounce and asserts the building reflection is negligible outside 20 to 45 m; field distances of 15 and 50 m were selected based on this model.
  • domain assumption The number of sources D is known or correctly estimated.
    The algorithms in Section 2.3.2 separate D signal eigenvectors from the noise subspace; all three-element-array results assume a valid D.
  • domain assumption Simulation assumes perfect antenna calibration.
    Section 4.1 explicitly assumes perfect calibration and notes this is not true in reality; the simulated RMSE is an upper bound on ideal performance.
  • ad hoc to paper Outlier removal with 3 scaled MAD is a valid cleaning rule.
    Algorithm 1 applies this filter without justification or sensitivity analysis; it is a data-processing choice specific to this paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Angle of Arrival Estimation Using SRS in 5G NR Uplink Scenarios." pith.science (2026). https://pith.science/paper/5OSNNNWO

@misc{pith2026241116501,
  author       = {Pith},
  title        = {Pith review of: Angle of Arrival Estimation Using SRS in 5G NR Uplink Scenarios},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5OSNNNWO}},
  note         = {Machine review of arXiv:2411.16501}
}
read the original abstract

This paper presents a comprehensive exploration of Angle of Arrival (AoA) estimation techniques in 5G environments, using the Sounding Reference Signal (SRS) in Uplink scenarios both in simulations and with actual measurements. Leveraging 5G capabilities, we investigate AoA algorithms for single-base station positioning. The study includes simulations and practical tests on a developed dedicated testbed featuring a base station equipped with a three-element Uniform Linear Array (ULA), considering Line of Sight conditions in an open environment. The testbed, employing Ettus E312 as the transmitter and Ettus N310 as the receiver, details waveform structures and reception processes. Additionally, our study examines the performance of Angle of Arrival (AoA) estimation algorithms, such as Multiple Signal Classification (MUSIC), Estimation of Signal Parameters via Rotational Invariant Techniques (ESPRIT), and Joint Angle and Delay Estimation (JADE) ESPRIT. A MATLAB ray tracing propagation model of the environment where the measurements are conducted, has been constructed. Simulation results using this model are presented, along with the actual measurements. The obtained results affirm the effectiveness of our implementation.

Figures

Figures reproduced from arXiv: 2411.16501 by the authors.

Figure 1
Figure 1. Propagation environment and ray creation via MATLAB ray tracing model 2.2. Channel Model Simulations were carried out via the MATLAB [7] ray tracing propagation model [8], in the field trials environment, in every frequency band that was intended to be employed (2.4 GHz, 3.5 GHz). The shooting and bouncing (SBR) method was used for the creation of the rays, with a maximum of one bounce per ray. The simulation enviro… view at source ↗
Figure 2
Figure 2. User and Base Station setup during field trials 3.1. Testbed Equipment The testbed setup utilizes an Ettus E312 as the transmitter and an Ettus N310 as the receiver, presented in [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. User and Base Station setup during the calibration procedure Algorithm 1 AoA Estimation with Ettus N310 1: while remaining size of the IQ Samples is greater than or equal to the size of a snapshot do 2: Align the phase of the snapshot for the individual channels of the two pairs, using values computed in offline calibration. 3: Initialize a pointer at the first IQ sample. Load IQ samples corresponding to one snapsho… view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: RMSE in degrees for reception with three antennas, simulated with ray tracing model for the field trials area, at 2.4 GHz, with LOS at 0°. 10 20 30 40 50 0° 0.2° 0.4° 0.6° 0.8° Distance (m) RMSE (°) 2D ESPRIT MUSIC ESPRIT [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: RMSE in degrees for reception with three antennas, simulated utilizing a ray tracing model for the field trials area, at 3.5 GHz, with LOS at 0°. 4. Results and Discussion 4.1. Simulations Prior to conducting field trials, we utilized the MATLAB ray tracing propagation…
Figure 6
Figure 6. Figure 6: Performance evaluation of super-resolution algorithms in Licensed 3.5 GHz band for static tests for different Angles of Arrival at 15 m distance 10 20 30 40 50 60 1° 2° 3° Time (s) Angle (°) Configuration I - 2D ESPRIT Configuration I - MUSIC Configuration I - ESPRIT C…
Figure 7
Figure 7. Figure 7: Angle of Arrival Estimation at 15 m distance in ISM 2.4 GHz band, using Configurations I and II, with LOS at 0° fluctuations observed in the 3.5 GHz band test, particularly at 0°, 10°, and 15° angle of arrival, can be attributed to small channel fluctuations and potent…
Figure 8
Figure 8. Figure 8: Angle of Arrival Estimation at 50 m distance in ISM 2.4 GHz band, using Configurations I and II, with LOS at 0° E312 as the user and Ettus N310 as the base station, equipped with a three-element ULA. Calibration of N310 channels, compensating for phase offsets, was per…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Passive AoA Estimation of COTS 5G NR Handsets from Uplink SRS: A Practical USRP-B210 Implementation

    eess.SP 2026-07 conditional novelty 6.0 of 10

    Passive angle-of-arrival estimation of unmodified commercial 5G handsets from native uplink SRS is feasible with a two-element USRP B210 and a stock srsRAN gNB; accuracy is gated by SINR and multipath rather than range.

Reference graph

Works this paper leans on

13 extracted references · 8 canonical work pages · cited by 1 Pith paper

  1. [1]

    Blanco, N

    A. Blanco, N. Ludant, P. J. Mateo, Z. Shi, Y. Wang, J. Widmer, Performance Evaluation of Single Base Station ToA-AoA Localization in an LTE Testbed, in: 2019 IEEE 30th Annual International Symposium on Personal, Indoor and Mobile Radio Communications (PIMRC), 2019, pp. 1–6. doi:10.1109/PIMRC.2019.8904454

    work page arXiv 2019

  2. [2]

    B. Sun, B. Tan, W. Wang, E. S. Lohan, A Comparative Study of 3D UE Positioning in 5G New Radio with a Single Station, Sensors 21 (2021). URL: https://www.mdpi.com/1424-8220/21/4/1178. doi:10.3390/s21041178

    work page doi:10.3390/s21041178 2021

  3. [3]

    Y. Li, Z. Zhang, L. Wu, J. Dang, P. Liu, 5G Communication Signal Based Localization with a Single Base Station, in: 2020 IEEE 92nd Vehicular Technology Conference (VTC2020-Fall), 2020, pp. 1–5. doi:10.1109/VTC2020-Fall49728.2020.9348766

    arXiv 2020

  4. [4]

    H. B. Tuğrel, H. Alakoca, G. K. Kurt, C. Ayyıldız, Angle of Arrival (AoA) Estimation by using Software Defined Radios, in: 2016 24th Signal Processing and Communication Application Conference (SIU), 2016, pp. 1429–1432. doi:10.1109/SIU.2016.7496018

    work page arXiv 2016

  5. [5]

    Rares, C

    B. Rares, C. Codau, A. Pastrav, T. Palade, H. Hedesiu, B. Balauta, E. Puschita, Experimental Evaluation of AoA Algorithms using NI USRP Software Defined Radios, in: 2018 17th RoEduNet Conference: Networking in Education and Research (RoEduNet), 2018, pp. 1–6. doi: 10.1109/ ROEDUNET.2018.8514133

    arXiv 2018

  6. [6]

    Physical Channels and Modulation

    “Physical Channels and Modulation”, 3GPP TS 38.211, Version 17.4.0, 2023. Release 17

    work page 2023

  7. [7]

    URL: https://www.mathworks.com

    The MathWorks Inc., Matlab version: 23.2.0 (r2023b), 2023. URL: https://www.mathworks.com

    work page 2023

  8. [8]

    Z. Yun, M. F. Iskander, Ray Tracing for Radio Propagation Modeling: Principles and Applications, IEEE Access 3 (2015) 1089–1100. doi:10.1109/ACCESS.2015.2453991

    arXiv 2015

Show all 13 references
  1. [9]

    Schmidt, Multiple emitter location and signal parameter estimation, IEEE Transactions on Antennas and Propagation 34 (1986) 276–280

    R. Schmidt, Multiple emitter location and signal parameter estimation, IEEE Transactions on Antennas and Propagation 34 (1986) 276–280. doi:10.1109/TAP.1986.1143830

    1986

  2. [10]

    R. Roy, T. Kailath, ESPRIT-Estimation of Signal Parameters via Rotational Invariance Techniques, IEEE Transactions on Acoustics, Speech, and Signal Processing 37 (1989) 984–995. doi:10.1109/ 29.32276

    1989

  3. [11]

    van der Veen, M

    A.-J. van der Veen, M. Vanderveen, A. Paulraj, Joint angle and delay estimation using shift- invariance techniques, IEEE Transactions on Signal Processing 46 (1998) 405–418. doi: 10.1109/ 78.655425

    1998

  4. [12]

    Zoltowski, M

    M. Zoltowski, M. Haardt, C. Mathews, Closed-form 2-D angle estimation with rectangular arrays in element space or beamspace via unitary ESPRIT, IEEE Transactions on Signal Processing 44 (1996) 316–328. doi:10.1109/78.485927

    1996 doi

  5. [13]

    Xhafa, F

    A. Xhafa, F. Fabra, J. A. López-Salcedo, G. Seco-Granados, I. Lapin, Pathway to Coherent Phase Acquisition in Multi-Channel USRP SDRs for Direction of Arrival Estimation, in: 2023 Tenth Inter- national Conference on Software Defined Systems (SDS), 2023, pp. 1–8. doi:10.1109/SD...

    2023

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.