pith. sign in

arxiv: 2507.17645 · v3 · submitted 2025-07-23 · 📡 eess.SP

Quaternion-Domain Super MDS for Robust 3D Localization

Alessio Lukaj , Keigo Masuoka , Takumi Takahashi , Giuseppe Thadeu Freitas de Abreu , Hideki Ochiai This is my paper

Pith reviewed 2026-05-19 02:44 UTC · model grok-4.3

classification 📡 eess.SP
keywords quaternion-domainmulti-dimensional scaling3D localizationGram edge kernel matrixnoise reductionwireless sensor networkssingular value decompositionmatrix multiplication
0
0 comments X

The pith

Representing 3D node positions as quaternions lets a multi-dimensional scaling method combine distances and angles into one matrix for stronger noise reduction.

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

The paper develops a quaternion version of super multi-dimensional scaling to locate nodes in three-dimensional wireless sensor networks from noisy measurements. Switching the coordinates to quaternions produces a rank-1 Gram edge kernel matrix that packs relative distances and angles together, so low-rank truncation removes more noise than the real-valued original. The authors also give a simpler version that skips the singular-value step and still matches most of the accuracy gain. A reader might care because wireless networks often face large measurement errors, and better error tolerance could make positioning reliable without extra sensors or power.

Core claim

By representing 3D node coordinates as quaternions, the SMDS algorithm constructs a rank-1 Gram edge kernel matrix that integrates both relative distance and angular information between nodes. Low-rank truncation of this matrix via singular value decomposition enhances the noise reduction effect and improves robustness against information loss. A variant estimates node coordinates directly through matrix multiplications in the quaternion domain, eliminating the need for singular value decomposition while achieving comparable localization accuracy.

What carries the argument

The rank-1 quaternion-domain Gram edge kernel matrix that integrates relative distance and angular information for enhanced low-rank noise reduction.

If this is right

  • The quaternion-domain method yields significantly lower localization errors than the original real-domain SMDS, particularly when measurement errors are large.
  • The direct-multiplication variant delivers accuracy close to the SVD version while avoiding the main computational cost.
  • Integration of distance and angle information inside the rank-1 matrix strengthens robustness to noise and partial information loss.

Where Pith is reading between the lines

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

  • The same quaternion structure might be tried on localization problems that already mix range and bearing measurements to see whether the noise benefit appears without redesign.
  • Sensor networks could test the matrix-multiplication version on low-power hardware to measure actual energy savings from skipping decompositions.
  • Real-world trials with actual radio signals would show whether the simulated gains survive when errors include non-Gaussian effects such as multipath.

Load-bearing premise

Representing 3D coordinates as quaternions produces a rank-1 Gram edge kernel matrix whose low-rank truncation meaningfully combines distance and angle data to reduce noise beyond the real-domain version.

What would settle it

Compare position estimation errors of the quaternion method against the original SMDS on the same set of nodes with added large measurement noise; if the new errors are not consistently smaller, the claimed improvement does not hold.

Figures

Figures reproduced from arXiv: 2507.17645 by Alessio Lukaj, Giuseppe Thadeu Freitas de Abreu, Hideki Ochiai, Keigo Masuoka, Takumi Takahashi.

Figure 1
Figure 1. Figure 1: Illustration of the parameters required to construct quaternion-domain GEK matrix [PITH_FULL_IMAGE:figures/full_fig_p005_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Parameters that can be obtained using a planar antenna. [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: Comparison of localization accuracy between the SMDS and QD [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Comparison of localization accuracy between the SMDS and QD [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Convergence of iterative QD-MRC-SMDS under various ranging and [PITH_FULL_IMAGE:figures/full_fig_p010_6.png] view at source ↗
Figure 8
Figure 8. Figure 8: Comparison of localization accuracy between the QD-SMDS, QD [PITH_FULL_IMAGE:figures/full_fig_p011_8.png] view at source ↗
read the original abstract

This paper proposes a novel low-complexity three-dimensional (3D) localization algorithm for wireless sensor networks, termed quanternion-domain super multi-dimensional scaling (QD-SMDS). The algorithm is based on a reformulation of the SMDS, originally developed in the real domain, using quaternion algebra. By representing 3D coordinates as quaternions, the method constructs a rank-1 Gram edge kernel (GEK) matrix that integrates both relative distance and angular information between nodes, which enhances the noise reduction effect achieved through low-rank truncation employing singular value decomposition (SVD), thereby improving robustness against information loss. To further reduce computational complexity, we also propose a variant of QD-SMDS that eliminates the need for the computationally expensive SVD by leveraging the inherent structure of the quaternion-domain GEK matrix. This alternative directly estimates node coordinates using only matrix multiplications within the quaternion domain. Simulation results demonstrate that the proposed method significantly improves localization accuracy compared to the original SMDS algorithm, especially in scenarios with substantial measurement errors. The proposed method also achieves comparable localization accuracy without requiring SVD.

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.

Referee Report

2 major / 3 minor

Summary. The paper proposes Quaternion-Domain Super MDS (QD-SMDS), a reformulation of real-domain Super MDS for 3D localization in wireless sensor networks. By representing node coordinates as quaternions, it constructs a claimed rank-1 Gram edge kernel (GEK) matrix integrating distance and angular information to enhance noise reduction via low-rank SVD truncation. A variant is also proposed that estimates coordinates via direct quaternion-domain matrix multiplications, avoiding SVD. Simulations are reported to demonstrate significantly improved localization accuracy over the original SMDS, especially under large measurement errors, while achieving comparable accuracy without SVD.

Significance. If the central claims hold, the work offers a parameter-free algebraic extension of SMDS that could improve robustness in noisy 3D localization settings while reducing complexity in the SVD-free variant. The direct reformulation without invented entities or fitted parameters is a methodological strength that could be useful for resource-limited WSN applications.

major comments (2)
  1. [§3 (GEK matrix construction)] §3 (GEK matrix construction): The central robustness claim requires that the quaternion representation produces an exactly rank-1 GEK matrix whose low-rank truncation integrates independent angular information beyond the distance data in real-domain SMDS. No explicit derivation, singular-value spectrum comparison, or verification for generic 3D node configurations is provided to confirm this property or to show that angular components are not a re-encoding of the same distance measurements; if the rank-1 property or independence fails, the claimed noise-reduction gain collapses.
  2. [Simulation results section] Simulation results section: The claim of 'significantly improves localization accuracy... especially in scenarios with substantial measurement errors' and 'comparable localization accuracy without requiring SVD' is load-bearing for the paper's contribution, yet the results are presented only at a high level with no error bars, Monte Carlo trial counts, specific error models, quantitative tables, or additional baselines. This leaves the magnitude and reliability of the reported gains unverified.
minor comments (3)
  1. [Abstract] Abstract: Typo 'quanternion-domain' should read 'quaternion-domain'.
  2. [Notation] Notation: Quaternion multiplication and conjugation operations used in the GEK construction and coordinate recovery steps would benefit from explicit definitions or a short preliminary subsection for readers less familiar with quaternion algebra.
  3. [References] References: Consider citing prior quaternion-based approaches to MDS or localization to better situate the contribution.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and detailed review of our manuscript on Quaternion-Domain Super MDS. The comments have helped us identify areas where additional rigor and detail are needed. We address each major comment below and have made corresponding revisions to strengthen the paper.

read point-by-point responses

  1. Referee: [§3 (GEK matrix construction)] §3 (GEK matrix construction): The central robustness claim requires that the quaternion representation produces an exactly rank-1 GEK matrix whose low-rank truncation integrates independent angular information beyond the distance data in real-domain SMDS. No explicit derivation, singular-value spectrum comparison, or verification for generic 3D node configurations is provided to confirm this property or to show that angular components are not a re-encoding of the same distance measurements; if the rank-1 property or independence fails, the claimed noise-reduction gain collapses.

    Authors: We thank the referee for this observation. In the revised manuscript we have added an explicit algebraic derivation in Section 3 demonstrating that the quaternion-domain GEK matrix is exactly rank-1 when measurements are noise-free. The derivation proceeds by expressing the edge kernel entries via quaternion multiplication of coordinate differences, which simultaneously encodes both Euclidean distances and relative angles; these angular components are not a simple re-encoding of distances because the imaginary parts of the quaternion products introduce directional constraints orthogonal to the scalar distance information. To verify the property for generic configurations we have included a new figure showing the singular-value spectra of the GEK matrix for randomly generated 3D node sets, confirming that a single dominant singular value accounts for nearly all energy even under moderate noise, thereby supporting the claimed noise-reduction benefit of the low-rank truncation. revision: yes

  2. Referee: [Simulation results section] Simulation results section: The claim of 'significantly improves localization accuracy... especially in scenarios with substantial measurement errors' and 'comparable localization accuracy without requiring SVD' is load-bearing for the paper's contribution, yet the results are presented only at a high level with no error bars, Monte Carlo trial counts, specific error models, quantitative tables, or additional baselines. This leaves the magnitude and reliability of the reported gains unverified.

    Authors: We agree that the simulation results section lacked sufficient statistical detail. In the revision we have expanded this section to report results averaged over 1000 independent Monte Carlo trials, with error bars indicating one standard deviation. The measurement error model is now explicitly stated as independent zero-mean Gaussian noise added to both distance and angle observations, with standard deviations ranging from 0.1 m to 2 m. We have inserted quantitative tables listing RMSE values for QD-SMDS, the SVD-free variant, the original SMDS, and classical MDS across multiple noise levels. These additions substantiate the reported accuracy gains under large errors and confirm that the SVD-free version achieves comparable performance. revision: yes

Circularity Check

0 steps flagged

Quaternion reformulation of SMDS is algebraic construction with no load-bearing reduction to self-fit or self-citation

full rationale

The paper's core contribution is an explicit algebraic reformulation that maps 3D coordinates to quaternions to build a Gram edge kernel matrix asserted to be rank-1 and to integrate distance plus angular information. This construction is presented directly rather than derived from a fitted parameter or prior self-cited uniqueness result; the claimed noise-reduction benefit is then evaluated via simulation. No equation reduces by construction to an input that was itself obtained from the target quantity, and no central premise rests solely on an overlapping-author citation. The result therefore remains self-contained against external benchmarks, warranting only a minimal circularity score.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The approach rests on the algebraic properties of quaternions for 3D representation and the claim that the constructed GEK matrix is exactly rank-1 when distances and angles are consistent.

axioms (1)
  • domain assumption 3D coordinates can be represented as quaternions such that the resulting Gram edge kernel matrix is rank-1 when measurements are noise-free.
    Invoked to justify low-rank truncation for noise reduction.

pith-pipeline@v0.9.0 · 5734 in / 1239 out tokens · 34495 ms · 2026-05-19T02:44:40.690631+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

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

  • Foundation/AlexanderDuality.lean alexander_duality_circle_linking echoes
    ?
    echoes

    ECHOES: this paper passage has the same mathematical shape or conceptual pattern as the Recognition theorem, but is not a direct formal dependency.

    By representing 3D coordinates as quaternions, the method constructs a rank-1 Gram edge kernel (GEK) matrix that integrates both relative distance and angular information

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

34 extracted references · 34 canonical work pages

  1. [1]

    H. Chen, H. Sarieddeen, T. Ballal, H. Wymeersch, M.-S. Alouini and T. Y . Al-Naffouri, “A Tutorial on Terahertz-Band Localization for 6G Communication Systems,“ IEEE Commun. Survey Tut. , vol. 24, no. 3, pp. 1780–1815, 2022

    work page 2022

  2. [2]

    P. K. R. Maddikunta, et al. , “Industry 5.0: A survey on enabling technologies and potential applications,“ J. Ind. Inf. Integr ., vol. 26, pp. 100257, 2022, [Online]. Available: https://www.sciencedirect. com/science/article/pii/S2452414X21000558

    work page 2022

  3. [3]

    Savi ´c, X

    T. Savi ´c, X. Vilajosana and T. Watteyne, “Constrained Localization: A Survey,“ IEEE Access , vol. 10, pp. 49297–49321, 2022

    work page 2022

  4. [4]

    M. N. Mowla, N. Mowla, A. F. M. S. Shah, K. M. Rabie and T. Shongwe, “Internet of Things and Wireless Sensor Networks for Smart Agriculture Applications: A Survey,“ IEEE Access , vol. 11, pp. 145813–145852, 2023

    work page 2023

  5. [5]

    Xiang, K

    W. Xiang, K. Yu, F. Han, L. Fang, D. He and Q.-L. Han, “Advanced Manufacturing in Industry 5.0: A Survey of Key Enabling Technolo- gies and Future Trends,“ IEEE Trans. Ind. Inform. , vol. 20, no. 2, pp. 1055–1068, 2024

    work page 2024

  6. [6]

    Javaid, S

    S. Javaid, S. Zeadally, H. Fahim and B. He, “Medical Sensors and Their Integration in Wireless Body Area Networks for Pervasive Healthcare Delivery: A Review“ IEEE Sensors Journal , vol. 22, no. 5, pp. 3860–3877, 2022

    work page 2022

  7. [7]

    Chaloupka, “Technology and Standardization Gaps for High Accuracy Positioning in 5g,“ IEEE Commun

    Z. Chaloupka, “Technology and Standardization Gaps for High Accuracy Positioning in 5g,“ IEEE Commun. Stand. Mag. , vol. 1, no. 1, pp. 59–65, 2017

    work page 2017

  8. [8]

    Yassin et al

    A. Yassin et al. , “Recent Advances in Indoor Localization: A Sur- vey on Theoretical Approaches and Applications,“ IEEE Commun. Survey Tut., vol. 19, no. 2, pp. 1327–1346, 2017

    work page 2017

  9. [9]

    Ahmad, W

    R. Ahmad, W. Alhasan, R. Wazirali and N. Aleisa, “Optimization Algorithms for Wireless Sensor Networks Node Localization: An Overview, IEEE Access , vol. 12, pp. 50459–50488, 2024

    work page 2024

  10. [10]

    J. Xiao, Z. Zhou, Y . Yi, and L. M. Ni. “A Survey on Wireless Indoor Localization from the Device Perspective,“ ACM Com- put. Surv. , vol. 49, no. 2, pp. 1–31, 2016, [Online]. Available: https://doi.org/10.1145/2933232

    work page doi:10.1145/2933232 2016

  11. [11]

    Wymeersch, J

    H. Wymeersch, J. Lien and M. Z. Win, “Cooperative Localization in Wireless Networks,“ Proceedings of the IEEE , vol. 97, no. 2, pp. 427–450, 2009

    work page 2009

  12. [12]

    Xiong, X.-P

    J. Xiong, X.-P. Xie, Z. Xiong, Y . Zhuang, Y . Zheng and C. Wang, “Adaptive Message Passing for Cooperative Positioning Under Un- known Non-Gaussian Noises“ IEEE Trans. Instrum. Meas. , vol. 73, pp. 1–14, 2024

    work page 2024

  13. [13]

    X. Shi, G. Mao, B. D. O. Anderson, Z. Yang and J. Chen, “Ro- bust Localization Using Range Measurements With Unknown and Bounded Errors,“ IEEE Trans. Wireless Commun. , vol. 16, no. 6, pp. 4065–4078, 2017

    work page 2017

  14. [14]

    W. S. Torgerson, “Multidimensional scaling: I. Theory and method,“ Psychometrika vol. 17, pp. 401-–419, 1952

    work page 1952

  15. [15]

    Cox and M

    T. Cox and M. Cox, “Multidimensional scaling,“ 2nd ed. New York, Chapman & Hall/CRC, 2000

    work page 2000

  16. [16]

    Abreu and G

    G. Abreu and G. Destino, “Super MDS: Source Location from Distance and Angle Information,“ in Proc. 2007 IEEE Wireless Commun. Netw. Conf. (WCNC) , vol. 2, pp. 4430–4434

    work page 2007

  17. [17]

    Macagnano and G

    D. Macagnano and G. Abreu, “Gershgorin Analysis of Random Gramian Matrices With Application to MDS Tracking,“ IEEE Trans. Signal Process., vol. 59, no. 4, pp. 1785–1800, 2011

    work page 2011

  18. [18]

    Macagnano and G

    D. Macagnano and G. Abreu, “Algebraic Approach for Robust Localization with Heterogeneous Information,“ IEEE Trans. Wireless Commun., vol. 12, no. 10, pp. 5334–5345, 2013

    work page 2013

  19. [19]

    Ghods and G

    A. Ghods and G. Abreu, “Complex-Domain Super MDS: A New Framework for Wireless Localization With Hybrid Information,“ IEEE Trans. Wireless Commun. , vol. 17, no. 11, pp. 7364–7378, 2018

    work page 2018

  20. [20]

    Nishi, T

    Y . Nishi, T. Takahashi, H. Iimori, G. Abreu, S. Ibi and S. Sampei, “Wireless Location Tracking via Complex-Domain Super MDS with Time Series Self-Localization Information,“ in Proc. 2023 IEEE Int. Conf. Acoust., Speech Signal Process. (ICASSP) , 2023, pp. 1–5

    work page 2023

  21. [21]

    M. E. Luna-Elizarrar ´as, M. Shapiro, D. C. Struppa, and A. Vajiac, “Some properties of bicomplex holomorphic functions,“ Cham, Switzerland: Springer International Publishing, 2015

    work page 2015

  22. [22]

    J. Baek, H. Jeon, G. Kim and S. Han, “Visualizing Quaternion Multiplication,“ IEEE Access , vol. 5, pp. 8948–8955, 2017

    work page 2017

  23. [23]

    Miao and K

    J. Miao and K. I. Kou, “Color Image Recovery Using Low-Rank Quaternion Matrix Completion Algorithm,“ IEEE Trans. image Process., vol. 31, pp. 190–201, 2022

    work page 2022

  24. [24]

    Masuoka, T

    K. Masuoka, T. Takahashi, G. Abreu and H. Ochiai, “Quaternion Domain Super MDS for 3D Localization,“ in Proc. IEEE Workshop Signal Process. and Artif. Intell. Wireless Commun. (SPA WC), pp. 1– 5, 2025

    work page 2025

  25. [25]

    Miron, J

    S. Miron, J. Flamant, N. L. Bihan, P. Chainais and D. Brie, “Quaternions in Signal and Image Processing: A comprehensive and objective overview,“ IEEE Signal Process. Mag. , vol. 40, no. 6, pp. 26–40, 2023

    work page 2023

  26. [26]

    Barth ´elemy, A

    Q. Barth ´elemy, A. Larue and J. I. Mars, “Color Sparse Representa- tions for Image Processing: Review, Models, and Prospects,“ IEEE Trans. on Image Process. , vol. 24, no. 11, pp. 3978–3989, 2015

    work page 2015

  27. [27]

    Shoemake, “Animating rotation with quaternion curves,“ in Proc

    K. Shoemake, “Animating rotation with quaternion curves,“ in Proc. 12th Annu.Conf. Compt. Graph. Interactive Techn. (SIGGRAPH ’85), 1985, pp.245–254

    work page 1985

  28. [28]

    P. D. Fiore, “Efficient linear solution of exterior orientation,“ IEEE Trans. Pattern Anal. Mach. Intell. , vol. 23, no. 2, pp. 140–148, 2001

    work page 2001

  29. [29]

    Zhang, Y

    D. Zhang, Y . Zhang, G. Zheng, B. Deng, C. Feng and J. Tang, “Two- Dimensional Direction of Arrival Estimation for Coprime Planar Arrays via Polynomial Root Finding Technique,“ IEEE Access , vol. 6, pp. 19540–19549, 2018

    work page 2018

  30. [30]

    Dokmanic, R

    I. Dokmanic, R. Parhizkar, J. Ranieri and M. Vetterli, “Euclidean Distance Matrices: Essential theory, algorithms, and applications,“ IEEE Signal Process. Mag. , vol. 32, no. 6, pp. 12–30, 2015

    work page 2015

  31. [31]

    Shabat and A

    G. Shabat and A. Averbuch, J. Ranieri and M. Vetterli, “Interest zone matrix approximation,“ Electronic Journal of Linear Algebra , vol. 23, 2012

    work page 2012

  32. [32]

    Papoulis, S

    A. Papoulis, S. U. Pillai, “Probability, Random Variables, and Stochastic Processes,“ 4th ed., New York, NY , USA: McGraw-Hill, 2002

    work page 2002

  33. [33]

    A. J. Viterbi, “Principles of Coherent Communication“, New York, NY , USA: McGraw-Hill, 1966

    work page 1966

  34. [34]

    Abreu, “On the generation of Tikhonov variates,“ IEEE Trans

    G. Abreu, “On the generation of Tikhonov variates,“ IEEE Trans. Commun., vol. 56, no. 7, pp. 1157–1168, 2008

    work page 2008