pith. sign in

arxiv: 2607.00877 · v1 · pith:X2NJREFTnew · submitted 2026-07-01 · 📊 stat.ML · cs.IT· math.IT

Hierarchical Variational Kalman Filtering

Shilei Li , Dawei Shi , Wei Zheng , Ling Shi This is my paper

Pith reviewed 2026-07-02 06:09 UTC · model grok-4.3

classification 📊 stat.ML cs.ITmath.IT
keywords variational Kalman filteringprocess noise statisticssurrogate variableCAVI reformulationcovariance trackingzero-phase filtersliding-window estimationstate estimation
0
0 comments X

The pith

A surrogate variable for the process-noise-free state enables explicit modeling of process noise in variational Kalman filters, addressing inconsistent covariance estimation and slow convergence.

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

The paper seeks to overcome limitations in traditional variational Kalman filtering when noise statistics are unknown, which causes inconsistent process covariance estimates and slow convergence. By introducing a surrogate variable that stands for the process-noise-free state, the approach allows direct inference of process noise statistics. The authors reformulate standard coordinate ascent variational inference as a marginalized maximum a posteriori problem solved via single-step hyperparameter fitting. This removes the need for multiple inner iterations and separates the design of covariance tracking filters, supporting higher-order filters and sliding-window estimation that turns into a zero-phase filter over complete data history. Validation through numerical simulations shows faster convergence and better accuracy than existing approaches.

Core claim

The method introduces a surrogate variable representing the process-noise-free state to enable explicit modeling and inference of process noise statistics. It reformulates conventional CAVI as a marginalized MAP problem followed by single-step hyperparameter fitting, which eliminates multiple inner iterations, decouples covariance tracking filter design, permits higher-order filters, and enables sliding-window hyperparameter estimation that intrinsically operates as a zero-phase filter when the window covers all historical data.

What carries the argument

The surrogate variable for the process-noise-free state, which enables explicit modeling of process noise, combined with the reformulation of CAVI into marginalized MAP with single-step fitting.

If this is right

  • The architecture permits the deployment of higher-order filters for covariance tracking.
  • Sliding-window hyperparameter estimation is enabled.
  • When the window encompasses all historical data, the covariance tracking estimator operates as a zero-phase filter.
  • Numerical simulations demonstrate enhanced convergence speed and superior estimation accuracy compared with existing methods.
  • The reformulation preserves consistency of process covariance estimation without introducing new approximation errors.

Where Pith is reading between the lines

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

  • This decoupling of filters could allow independent optimization or parallel computation in large-scale tracking systems.
  • The zero-phase property suggests applications in post-processing or smoothing where no phase lag is desired.
  • Similar surrogate variable techniques might apply to other state estimation problems with unknown parameters.
  • Sliding windows offer a way to balance computational cost with adaptation speed in non-stationary environments.

Load-bearing premise

Reformulating CAVI as a marginalized MAP problem with single-step hyperparameter fitting preserves the consistency of process covariance estimation and avoids new approximation errors.

What would settle it

A counterexample where applying the method to a system with unknown but fixed process noise results in inconsistent covariance estimates or no improvement in convergence speed over standard methods.

Figures

Figures reproduced from arXiv: 2607.00877 by Dawei Shi, Ling Shi, Shilei Li, Wei Zheng.

Figure 1
Figure 1. Figure 1: Comparison of CAVI and marginalized MAP+1-step [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Performance analysis of the N = 1 and N = 2 estima￾tors. (a) The inherent speed-precision trade-off in the stan￾dard N = 1 EWMA filter, plotting the time constant (solid blue line) and normalized steady-state variance (dashed red line) against ρ. (b) Pareto front comparison of tracking speed versus precision. By applying the optimal overdamped mo￾mentum gain γ = ρ 2 1+ρ , the N = 2 trajectory (solid red li… view at source ↗
Figure 4
Figure 4. Figure 4: Tracking performance of parameter λ in a single re￾alization. HOHVKF responds faster to abrupt step changes in the ground truth compared to HVKF, but at the cost of slightly larger steady state variance [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figure 3
Figure 3. Figure 3: Overall RMSE comparison over 10 Monte Carlo runs. [PITH_FULL_IMAGE:figures/full_fig_p011_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: Performance comparison of different estimators. The [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 7
Figure 7. Figure 7: Tracking comparison for λ (see (49)) between the HVKF and HOHVKF algorithms. The higher-order ap￾proach (HOHVKF) yields a faster tracking response to sud￾den changes, at the cost of slightly increased fluctuation dur￾ing the constant intervals. 4.3 Example 3: Disturbance Estimation We consider a manipulator tracking example [9]. The augmented state-space model has xk = Φkxk−1 + Gkuk−1 + Bdwd,k + Bθwθ,k yk … view at source ↗
Figure 8
Figure 8. Figure 8: Performance comparison of KF, STKF, VBKF, [PITH_FULL_IMAGE:figures/full_fig_p013_8.png] view at source ↗
read the original abstract

Traditional variational Kalman filtering with unknown noise statistics suffers from inconsistent process covariance estimation and slow convergence speed, limiting its practical utility. To address these issues, we introduce a surrogate variable representing the process-noise-free state, which enables explicit modeling and inference of process noise statistics. In addition, we reformulate the conventional coordinate ascent variation inference (CAVI) as a marginalized maximum a posteriori problem, followed by a single-step hyperparameter fitting. This reformulation obviates the need for multiple inner iterations inherent to CAVI and decouples the design of the covariance tracking filters. Consequently, this architecture permits the deployment of higher-order filters for covariance tracking and enables sliding-window hyperparameter estimation. Notably, when this window encompasses all historical data, the covariance tracking estimator intrinsically operates as a zero-phase filter. Numerical simulations validate the theoretical framework, demonstrating the enhanced convergence speed and superior estimation accuracy compared with existing 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.

Referee Report

1 major / 1 minor

Summary. The paper claims that traditional variational Kalman filtering suffers from inconsistent process covariance estimation and slow convergence. It introduces a surrogate variable for the process-noise-free state to enable explicit modeling of process noise. The method reformulates coordinate ascent variational inference (CAVI) as a marginalized maximum a posteriori problem followed by single-step hyperparameter fitting. This is said to eliminate inner iterations, decouple covariance tracking filter design, permit higher-order filters, support sliding-window hyperparameter estimation, and yield an intrinsic zero-phase filter when the window includes all historical data. Numerical simulations are reported to demonstrate faster convergence and superior estimation accuracy relative to existing methods.

Significance. If the reformulation is shown to preserve the fixed points of the original variational objective and the simulations are rigorous, the work would offer a practical advance in adaptive Kalman filtering under unknown noise statistics by enabling decoupled, higher-order covariance tracking and a zero-phase property. This could improve efficiency in signal processing and state estimation tasks.

major comments (1)
  1. [Abstract / §3 (Proposed Method)] The central claim rests on the reformulation of CAVI as a marginalized MAP problem plus single-step fitting preserving consistency of process covariance estimation. However, no derivation is provided establishing equivalence of stationary points between the new objective and standard CAVI, nor bounding any additional error introduced by the surrogate variable marginalization. This equivalence is load-bearing for the consistency, decoupling, and zero-phase claims.
minor comments (1)
  1. [Abstract] The abstract refers to 'numerical simulations' validating the claims, but provides no information on simulation design, data generation, baselines, error metrics, or statistical significance; these details are needed to assess the reported gains in convergence speed and accuracy.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the careful reading and for identifying the need for explicit justification of the reformulation's properties. The concern regarding stationary-point equivalence is valid and will be addressed by adding the requested derivation and error analysis to the revised manuscript.

read point-by-point responses

  1. Referee: [Abstract / §3 (Proposed Method)] The central claim rests on the reformulation of CAVI as a marginalized MAP problem plus single-step fitting preserving consistency of process covariance estimation. However, no derivation is provided establishing equivalence of stationary points between the new objective and standard CAVI, nor bounding any additional error introduced by the surrogate variable marginalization. This equivalence is load-bearing for the consistency, decoupling, and zero-phase claims.

    Authors: We agree that the manuscript should contain an explicit derivation showing that the stationary points of the proposed marginalized MAP objective coincide with those of standard CAVI, together with a bound on any error introduced by the surrogate-variable marginalization. In the revision we will insert a new subsection (and accompanying theorem with proof) in §3 that (i) establishes exact equivalence of the fixed points for the Gaussian process-noise model and (ii) shows that the marginalization error is identically zero when the surrogate is defined as the process-noise-free state. This addition will directly support the consistency, decoupling, and zero-phase claims. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivation remains self-contained

full rationale

The provided abstract and context describe a surrogate variable for explicit process-noise modeling, a reformulation of CAVI into marginalized MAP plus single-step hyperparameter fitting, and resulting architectural properties (decoupling, higher-order filters, zero-phase when using full history). No equations are shown that reduce a claimed prediction or result to a fitted input by construction, nor any self-citation chains, uniqueness theorems, or ansatzes imported from prior author work. The zero-phase statement is presented as an intrinsic property of the full-history window choice rather than a tautological renaming or fit. The central claims rest on the reformulation's consequences rather than reducing to the inputs themselves.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 1 invented entities

The central claim rests on the introduction of a new surrogate modeling construct and the assumption that the inference reformulation decouples components without compromising consistency; these are not derived from prior literature but postulated to solve the stated problems.

free parameters (1)
  • hyperparameters for covariance tracking
    Single-step hyperparameter fitting is a core part of the proposed architecture.
axioms (1)
  • domain assumption The surrogate variable can represent the process-noise-free state and enable explicit inference of process noise statistics without introducing modeling inconsistencies
    Invoked to address inconsistent process covariance estimation in traditional variational Kalman filtering.
invented entities (1)
  • surrogate variable representing the process-noise-free state no independent evidence
    purpose: To enable explicit modeling and inference of process noise statistics
    New modeling entity introduced to overcome limitations of standard variational Kalman filters

pith-pipeline@v0.9.1-grok · 5678 in / 1439 out tokens · 34549 ms · 2026-07-02T06:09:26.949206+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

45 extracted references · 1 canonical work pages · 1 internal anchor

  1. [1]

    Orb-slam: A versatile and accurate monocular slam system,

    R. Mur-Artal, J. M. M. Montiel, and J. D. Tard´ os, “Orb-slam: A versatile and accurate monocular slam system,” IEEE Transactions on Robotics , vol. 31, pp. 1147–1163, 2015

    2015

  2. [2]

    Consistent and optimal solution to cam- era motion estimation,

    G. Zeng, Q. Zeng, X. Li, B. Mu, J. Chen, L. Shi, and J. Wu, “Consistent and optimal solution to cam- era motion estimation,”IEEE Transactions on Pat- tern Analysis and Machine Intelligence, vol. 47, pp. 12 005–12 020, 2024

    2024

  3. [3]

    Review review of the ensemble kalman filter for atmospheric data as- similation,

    P. Houtekamer and F. Zhang, “Review review of the ensemble kalman filter for atmospheric data as- similation,” 2016

    2016

  4. [4]

    Recursive noise adaptive kalman filtering by variational bayesian approximations,

    S. S¨ arkk¨ a and A. R. Nummenmaa, “Recursive noise adaptive kalman filtering by variational bayesian approximations,” IEEE Transactions on Automatic Control, vol. 54, pp. 596–600, 2009

    2009

  5. [5]

    New results in linear filtering and prediction theory,

    R. E. K´ alm´ an and R. S. Bucy, “New results in linear filtering and prediction theory,” Journal of Basic Engineering, vol. 83, pp. 95–108, 1961

    1961

  6. [6]

    Applications of kalman filtering in aerospace 1960 to the present [historical perspectives],

    M. S. Grewal and A. P. Andrews, “Applications of kalman filtering in aerospace 1960 to the present [historical perspectives],” IEEE Control Systems Magazine, vol. 30, no. 3, pp. 69–78, 2010

    1960

  7. [7]

    The unscented kalman filter for nonlinear estimation,

    E. A. Wan and R. van der Merwe, “The unscented kalman filter for nonlinear estimation,”Proceedings of the IEEE 2000 Adaptive Systems for Signal Pro- cessing, Communications, and Control Symposium (Cat. No.00EX373), pp. 153–158, 2000

    2000

  8. [8]

    Cubature kalman filters,

    I. Arasaratnam and S. Haykin, “Cubature kalman filters,” IEEE Transactions on Automatic Control, vol. 54, no. 6, pp. 1254–1269, 2009

    2009

  9. [9]

    General- ized multikernel maximum correntropy kalman fil- ter for disturbance estimation,

    S. Li, D. Shi, Y. Lou, W. Zou, and L. Shi, “General- ized multikernel maximum correntropy kalman fil- ter for disturbance estimation,”IEEE Transactions on Automatic Control, vol. 69, no. 6, pp. 3732–3747, 2024

    2024

  10. [10]

    Kalman-consensus filter : Opti- mality, stability, and performance,

    R. Olfati-Saber, “Kalman-consensus filter : Opti- mality, stability, and performance,” in Proceedings of the 48h IEEE Conference on Decision and Con- trol (CDC) held jointly with 2009 28th Chinese Con- trol Conference, 2009, pp. 7036–7042

    2009

  11. [11]

    The invariant ex- tended kalman filter as a stable observer,

    A. Barrau and S. Bonnabel, “The invariant ex- tended kalman filter as a stable observer,” IEEE Transactions on Automatic Control , vol. 62, pp. 1797–1812, 2014

    2014

  12. [12]

    Invariant smoother for legged robot state estimation with dy- namic contact event information,

    Z. Yoon, J. ha Kim, and H.-W. Park, “Invariant smoother for legged robot state estimation with dy- namic contact event information,” IEEE Transac- tions on Robotics, vol. 40, pp. 193–212, 2024

    2024

  13. [13]

    Ins/gps integration system using adap- tive filter for estimating measurement noise vari- ance,

    M. Yu, “Ins/gps integration system using adap- tive filter for estimating measurement noise vari- ance,” IEEE Transactions on Aerospace and Elec- tronic Systems, vol. 48, pp. 1786–1792, 2012

    2012

  14. [14]

    Resilient distributed esti- mation against fdi attacks: A correntropy-based ap- proach,

    W. Xia and Y. Zhang, “Resilient distributed esti- mation against fdi attacks: A correntropy-based ap- proach,” Inf. Sci., vol. 635, pp. 236–256, 2023

    2023

  15. [15]

    Generalized kalman smoothing: Modeling and algorithms,

    A. Y. Aravkin, J. V. Burke, L. Ljung, A. C. Lozano, and G. Pillonetto, “Generalized kalman smoothing: Modeling and algorithms,” Autom., vol. 86, pp. 63– 86, 2016

    2016

  16. [16]

    Covariance matching based adaptive unscented kalman filter for direct filtering in ins/gnss inte- 15 gration,

    Y. Meng, S. Gao, Y. Zhong, G. Hu, and A. Subic, “Covariance matching based adaptive unscented kalman filter for direct filtering in ins/gnss inte- 15 gration,” Acta Astronautica, vol. 120, pp. 171–181, 2016

    2016

  17. [17]

    Adaptive kalman filtering for ins/gps,

    A. Mohamed and K. P. Schwarz, “Adaptive kalman filtering for ins/gps,” Journal of Geodesy , vol. 73, pp. 193–203, 1999

    1999

  18. [18]

    T. D. Barfoot, State Estimation for Robotics. USA: Cambridge University Press, 2017

    2017

  19. [19]

    A recursive multi- ple model approach to noise identification,

    X. R. Li and Y. Bar-Shalom, “A recursive multi- ple model approach to noise identification,” IEEE Transactions on Aerospace and Electronic Systems, vol. 30, pp. 671–684, 1994

    1994

  20. [20]

    Variational bayesian inference for a nonlinear forward model,

    M. A. Chappell, A. R. Groves, B. Whitcher, and M. W. Woolrich, “Variational bayesian inference for a nonlinear forward model,” IEEE Transactions on Signal Processing, vol. 57, no. 1, pp. 223–236, 2009

    2009

  21. [21]

    Approaches to adaptive filtering,

    R. Mehra, “Approaches to adaptive filtering,” IEEE Transactions on Automatic Control, vol. 17, no. 5, pp. 693–698, 1972

    1972

  22. [22]

    Kalmannet: Neural network aided kalman filtering for partially known dynamics,

    G. Revach, N. Shlezinger, X. Ni, A. L. Escoriza, R. J. Van Sloun, and Y. C. Eldar, “Kalmannet: Neural network aided kalman filtering for partially known dynamics,” IEEE Transactions on Signal Processing, vol. 70, pp. 1532–1547, 2022

    2022

  23. [23]

    Kalmanformer: using transformer to model the kalman gain in kalman filters,

    S. Shen, J. Chen, G. Yu, Z. Zhai, and P. Han, “Kalmanformer: using transformer to model the kalman gain in kalman filters,” Frontiers in Neuro- robotics, vol. 18, 2025

    2025

  24. [24]

    Adaptive kalmannet: Data-driven kalman filter with fast adaptation,

    X. Ni, G. Revach, and N. Shlezinger, “Adaptive kalmannet: Data-driven kalman filter with fast adaptation,” ICASSP 2024 - 2024 IEEE Interna- tional Conference on Acoustics, Speech and Signal Processing (ICASSP), pp. 5970–5974, 2023

    2024

  25. [25]

    Latent- kalmannet: Learned kalman filtering for tracking from high-dimensional signals,

    I. Buchnik, G. Revach, D. Steger, R. J. Van Sloun, T. Routtenberg, and N. Shlezinger, “Latent- kalmannet: Learned kalman filtering for tracking from high-dimensional signals,”IEEE Transactions on Signal Processing, vol. 72, pp. 352–367, 2023

    2023

  26. [26]

    Bayesian kalmannet: quantifying uncertainty in deep learning augmented kalman filter,

    Y. Dahan, G. Revach, J. Dunik, and N. Shlezinger, “Bayesian kalmannet: quantifying uncertainty in deep learning augmented kalman filter,” IEEE Transactions on Signal Processing, 2025

    2025

  27. [27]

    Diffpf: Differentiable particle filtering with generative sampling via conditional diffusion models,

    Z. Wan and L. Zhao, “Diffpf: Differentiable particle filtering with generative sampling via conditional diffusion models,” IEEE Robotics and Automation Letters, 2026

    2026

  28. [28]

    Normalizing flow-based dif- ferentiable particle filters,

    X. Chen and Y. Li, “Normalizing flow-based dif- ferentiable particle filters,” IEEE Transactions on Signal Processing, vol. 73, pp. 493–507, 2024

    2024

  29. [29]

    Varia- tional bayesian-based maximum correntropy cuba- ture kalman filter with both adaptivity and robust- ness,

    J. He, C. Sun, B. Zhang, and P. Wang, “Varia- tional bayesian-based maximum correntropy cuba- ture kalman filter with both adaptivity and robust- ness,” IEEE Sensors Journal , vol. 21, no. 2, pp. 1982–1992, 2021

    1982

  30. [30]

    Variational nonlinear kalman filtering with unknown process noise covariance,

    H. Lan, J. Hu, Z. Wang, and Q. Cheng, “Variational nonlinear kalman filtering with unknown process noise covariance,”IEEE Transactions on Aerospace and Electronic Systems , vol. 59, pp. 9177–9190, 2023

    2023

  31. [31]

    A novel adaptive kalman filter with inaccu- rate process and measurement noise covariance ma- trices,

    Y. Huang, Y. Zhang, Z. Wu, N. Li, and J. A. Cham- bers, “A novel adaptive kalman filter with inaccu- rate process and measurement noise covariance ma- trices,” IEEE Transactions on Automatic Control, vol. 63, pp. 594–601, 2018

    2018

  32. [32]

    Max- imum correntropy criterion variational bayesian adaptive kalman filter based on strong tracking with unknown noise covariances,

    S. Qiao, Y. Fan, G. Wang, D. Mu, and Z. He, “Max- imum correntropy criterion variational bayesian adaptive kalman filter based on strong tracking with unknown noise covariances,”Journal of the Franklin Institute, vol. 360, no. 9, pp. 6515–6536, 2023

    2023

  33. [33]

    A slide window variational adaptive kalman filter,

    Y. Huang, F. Zhu, G. Jia, and Y. Zhang, “A slide window variational adaptive kalman filter,” IEEE Transactions on Circuits and Systems II: Express Briefs, vol. 67, no. 12, pp. 3552–3556, 2020

    2020

  34. [34]

    A variational bayesian-based unscented kalman filter with both adaptivity and robustness,

    K. Li, L. Chang, and B. Hu, “A variational bayesian-based unscented kalman filter with both adaptivity and robustness,”IEEE Sensors Journal, vol. 16, no. 18, pp. 6966–6976, 2016

    2016

  35. [35]

    Stochastic event-triggered variational bayesian fil- tering,

    X. Lv, P. Duan, Z. Duan, G. Chen, and L. Shi, “Stochastic event-triggered variational bayesian fil- tering,” IEEE Transactions on Automatic Control, vol. 68, no. 7, pp. 4321–4328, 2022

    2022

  36. [36]

    Variational nonlinear kalman filtering with unknown process noise covariance,

    H. Lan, J. Hu, Z. Wang, and Q. Cheng, “Variational nonlinear kalman filtering with unknown process noise covariance,”IEEE Transactions on Aerospace and Electronic Systems , vol. 59, no. 6, pp. 9177– 9190, 2023

    2023

  37. [38]

    Analysis of half- quadratic minimization methods for signal and im- age recovery,

    M. Nikolova and M. K. Ng, “Analysis of half- quadratic minimization methods for signal and im- age recovery,” SIAM Journal on Scientific comput- ing, vol. 27, no. 3, pp. 937–966, 2005

    2005

  38. [39]

    Variational inference: A review for statisticians,

    D. M. Blei, A. Kucukelbir, and J. D. McAuliffe, “Variational inference: A review for statisticians,” Journal of the American Statistical Association, vol. 112, no. 518, pp. 859–877, 2017

    2017

  39. [40]

    K. P. Murphy, Probabilistic Machine Learning: Advanced Topics . MIT Press, 2023. [Online]. Available: http://probml.github.io/book2

    2023

  40. [41]

    Variational Robust Kalman Filters: A Unified Framework

    S. Li, D. Shi, H. Yu, and L. Shi, “Variational ro- bust kalman filters: A unified framework,” arXiv preprint arXiv:2512.15419, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  41. [42]

    A general and adaptive robust loss function,

    J. T. Barron, “A general and adaptive robust loss function,” in Proceedings of the IEEE/CVF con- ference on computer vision and pattern recognition, 2019, pp. 4331–4339

    2019

  42. [43]

    Multi-kernel maximum correntropy kalman filter,

    S. Li, D. Shi, W. Zou, and L. Shi, “Multi-kernel maximum correntropy kalman filter,” IEEE Con- trol Systems Letters, vol. 6, pp. 1490–1495, 2021

    2021

  43. [44]

    Ogata, Discrete-Time Control Systems, 2nd ed

    K. Ogata, Discrete-Time Control Systems, 2nd ed. Englewood Cliffs, NJ, USA: Prentice-Hall, 1995

    1995

  44. [45]

    A Kalman & fading memory co-filter for uncertain systems based on self-perception mechanism,

    X. Luan, W. Xue, S. Zhao, and F. Liu, “A Kalman & fading memory co-filter for uncertain systems based on self-perception mechanism,” IEEE Trans- actions on Automatic Control , 2025. Shilei Li received the B.E. degree in Detection Guidance and Control Tech- 16 nology and M.S. degree in Control Engineering both from Harbin Insti- tute of Technology, Harbin,...

    2025

  45. [46]

    He was a subject editor for International Journal of Robust and Nonlinear Control (2015-2017). He has been serving as an associate editor for IEEE Transac- tions on Control of Network Systems from July 2016, and an associate editor for IEEE Control Systems Let- ters from Feb 2017. He also served as an associate editor for a special issue on Secure Control...

    2015