Cooperative Optimization of Grid-Edge Cyber and Physical Resources for Resilient Power System Operation

Ana E. Goulart; Katherine R. Davis; Khandaker Akramul Haque; Leen Al Homoud; Shining Sun; Xiang Huo

REVIEW 4 major objections 5 minor 16 references

Cooperative Optimization of Grid-Edge Cyber and Physical Resources for Resilient Power System Operation

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

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

Pith's one-line read A coordinated two-layer optimization—rerouting the communication network and dispatching backup storage—keeps power-system voltages within limits during a combined cyber-physical attack.

desk verdict A promising cyber-physical co-optimization framework whose written model forgets to force critical nodes active; fixable, but the current claims outrun the equations. read the letter →

arxiv 2502.05768 v1 pith:JWK5JQGS submitted 2025-02-09 eess.SY cs.SY

Xiang Huo , Shining Sun , Khandaker Akramul Haque , Leen Al Homoud , Ana E. Goulart , Katherine R. Davis This is my paper · ORCID

classification eess.SYcs.SY

keywords cyber-physicalpowersystemgridresiliencecommunicationnetworkreconfigurationspanningtreebi-leveloptimizationenergystorageACoptimalflowcyberattackmitigation

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

Power grids are cyber-physical systems: when an attacker simultaneously takes down a communication node and the generator it controls, the grid needs an answer that covers both layers. This paper proposes a cooperative optimization framework in which the cyber layer reconfigures its communication topology—switching control to a neighboring node—while the physical layer redispaches generation and activates a backup energy storage system. The two layers are coupled through a bi-level problem: the cyber reconfiguration is solved first, then the physical response is optimized subject to the new topology. The paper's central claim is that this coordinated response keeps bus voltages inside their normal bounds after a combined cyber-physical attack, and it demonstrates the claim on a modified IEEE 14-bus system where a generator and its control node are knocked out together.

What carries the argument

The load-bearing mechanism is a flow-based reformulation of the communication network inspired by the Spanning Tree Protocol. Each active cyber node consumes one unit of flow that originates at the root node, and the net flow balance at every node forces the active nodes to form a tree, while link-activation variables tie flows to communication links. This reformulation turns cyber topology design into a mixed-integer linear program whose objective is the cost of activated links and deployed cyber resources. Coupled with this is a bi-level decomposition: the upper level solves the cyber MILP, and the lower level solves an AC optimal power flow augmented with energy-storage constraints, so the two layers exchange only the identity of the replacement cyber node and the physical resources it controls.

What would settle it

Rerun the paper's IEEE 14-bus attack scenario with the backup ESS at the selected replacement node forced to zero energy capacity; if the voltage at Bus 6 drops below $0.94$ p.u. at any post-attack time step, the claim that backup ESS dispatch keeps voltages within limits is falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that grid resilience during a cyber-physical event can be restored by coordinating a small set of actions across two layers rather than by hardening either layer alone. On the cyber side, a flow-based spanning-tree optimization isolates the corrupted node and reroutes control to the least-expensive neighboring node that keeps all critical nodes connected. On the physical side, an AC optimal power flow problem reschedules the remaining generators and starts a backup ESS at the affected bus to cover the lost generation. The two problems are arranged as a bi-level program—cyber decisions in the upper level, physical decisions in the lower level—and solved sequentially by the proposed algorithm. On the modified IEEE 14-bus test case, the method keeps the voltage at Bus 6 within $[0.94, 1.06]$ p.u. after the attack, which the authors present as evidence that the cooperative framework can guide coupled cyber-physical resource control.

Load-bearing premise

The load-bearing premise is that the flow-based cyber constraints really do force every critical node to remain active and connected after rerouting, and that the selected neighboring node actually has a backup energy storage resource available to serve the affected bus.

Editorial extensions

If this is right

Operators can precompute a ranked list of neighboring cyber nodes for each critical asset, so that when an attack is detected the response reduces to solving two small optimization problems instead of one large coupled problem.
The same formulation can be used for day-ahead planning to decide where backup energy storage should be placed, since the lower-level problem reveals which buses need ESS support to hold voltages during contingencies.
Because link activation costs can encode communication latency, the cyber reconfiguration naturally trades monetary cost against control speed, which gives a quantitative rationale for choosing one replacement node over another.
The three objectives—cyber cost, generation cost, and resilience cost—are combined with adjustable weights, so a utility can tune the same model for cheap normal operation or for aggressive emergency response.

Reading between the lines

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

The paper's coupling is one-directional: the cyber layer chooses the replacement node, and only then does the physical layer respond. A natural extension would feed voltage violations back into the cyber choice, which would likely pick a different replacement when the first candidate's ESS is too small.
The method assumes a backup ESS exists at or near the replacement node. If that ESS has limited energy, the voltage guarantee holds only for a finite number of time steps, so sizing the ESS against the attack duration is a testable extension the paper does not address.
The flow-based spanning-tree model allows the critical-node set to be any specified list, so the same algorithm could be applied to protect loads or control centers rather than only generator buses.
Scaling to large grids would likely require decomposition or heuristics, since the upper-level MILP and the lower-level AC OPF are each computationally difficult in general; the paper explicitly leaves large-scale deployment as future work.

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.

Desk Editor's Note

read the letter

Quick take: this paper describes a plausible cyber-physical co-optimization framework, but the written model doesn't actually enforce that critical cyber nodes stay active, so the headline connectivity guarantee doesn't follow from the equations. The fix is straightforward, but the simulation needs strengthening too.

What's new: the combination of a spanning-tree flow formulation for cyber topology reconfiguration with AC OPF and backup ESS dispatch in a bi-level structure, plus an adaptive algorithm that re-routes around a compromised node. Each ingredient is known, but the integration is not something I've seen in the cited references. The paper is honest in Remark 2 that the bi-level solution is not globally optimal—good.

The main problem is exactly what the stress-test flags: constraints (7)-(10) only ensure the active nodes form a rooted tree; they never require x_i=1 for the critical set K. Section III-B says K 'must be active and connected,' but that condition isn't in problem (11) or the upper level of (14). With positive deployment costs, the optimizer can set every x_i=0 except the root, satisfy all constraints, and the 'critical' nodes are ignored. Algorithm 1 inherits the gap when it updates K. So the central claim—that the reconfiguration keeps critical nodes connected—is not supported by the written formulation. This is fixable: add x_i=1 for all i in K, and then re-derive the flow constraints. But as written, the model doesn't do what the paper says.

Two lesser issues. First, (11) is called a MILP, but the solver is IPOPT, an NLP solver, and the paper doesn't mention how the binary variables are handled. That needs clarification. Second, the simulation is a single scenario on the IEEE 14-bus case with no quantitative comparison—no cost numbers, no comparison to a non-cooperative approach, just a voltage plot that shows the ESS helps. That's a demonstration, not a validation.

Overall, the framework is a legitimate direction, and the missing-constraint issue is a fixable oversight rather than a dead end. With the K-activation constraint added and a more rigorous simulation, the paper could be solid. As it stands, it deserves a serious referee, but the referee should insist on those revisions.

Referee Report

4 major / 5 minor

Summary. The paper proposes a cooperative cyber-physical optimization framework for resilient power system operation. The cyber layer is modeled as a flow-based spanning-tree problem that minimizes the cost of deploying cyber resources and activating communication links, while the physical layer is modeled as an AC optimal power flow problem. A bi-level formulation is introduced, and an adaptive algorithm reconfigures the cyber topology and dispatches backup energy storage after a combined cyber-physical attack. The approach is demonstrated on a modified IEEE 14-bus system with a simulated attack on a cyber node and its associated generator.

Significance. If the formal gap in the cyber-layer model is repaired, the paper could provide a useful constructive method for coordinating communication-network reconfiguration with physical resource dispatch during cyber-physical contingencies. The approach is not circular: it is a new optimization-based construction on standard test-case data, solved with an external solver, and it does not rely on fitted parameters from the target scenario. However, the central cyber-topology guarantee is not enforced by the written constraints, and the numerical evaluation lacks a baseline, so the significance of the current version is limited.

major comments (4)

[III-B, Eq. (11) and (14)] The model does not impose x_i=1 for i in K, despite the statement in Section III-B that critical cyber nodes must be active and connected. Constraints (7)-(10) only require that the set of active nodes forms a rooted tree; they do not force any non-root node to be active. With positive deployment costs, an optimizer can satisfy (11) with only the root active, so the claimed connectivity guarantee for K is not a consequence of the written formulation. The same gap appears in the upper level of (14) and in Algorithm 1, where the updated critical set is never forced to be active. Please add explicit membership constraints for the critical set and re-derive the stated guarantee.
[IV, Figs. 3-4] The simulation section contains no baseline or quantitative performance metric. Figures 3 and 4 show post-attack trajectories, but there is no comparison with a no-mitigation case, with an alternative neighboring cyber node, or with another control strategy; the text only states that voltage is regulated to stay within limits. Consequently, the abstract's claim that the method 'improves grid resilience' is not quantitatively established. Please add a baseline and report numerical improvements (e.g., voltage deviation, load served, or total cost).
[III-C, Remark 2 and Contribution 3] The paper lists 'bi-level optimal cyber and physical response actions' as a contribution, but Remark 2 correctly observes that the solution does not ensure global optimality across the two levels, and Algorithm 1 is a sequential procedure rather than a simultaneous bi-level solve. This terminology should be revised to 'cooperative local optimization' or 'sequential decomposition'; otherwise the contribution claim is contradicted by the manuscript's own remark.
[III-C, Eq. (14)] The coupling between cyber and physical layers is not formalized. The upper level (14a) contains only cyber variables and constraints, while the lower level (14c) contains physical and resilience variables; the resilience cost f_res in (14c) includes \tilde x6, but \tilde x6 is not linked to the cyber topology decisions in (14a)-(14b). Please specify how the bi-level decisions interact and how the 'cyber-physical couplings' described in Section III-B enter the optimization.

minor comments (5)

[Abstract] The word 'phyiscal' should be 'physical'.
[I] 'Sahu el al.' should be 'Sahu et al.', and 'doesnot' should be 'does not'.
[II-B] The set of non-root nodes is denoted with an unclear symbol 'V¯ˆır'; define the notation explicitly before Eq. (8).
[IV, Fig. 3] The caption reads 'shows the shows the generators'; remove the duplication.
[IV] Please report the cyber topology cost coefficients and ESS parameters used in the simulation, or state that they are available from the cited standards, to improve reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the bi-level cyber-physical optimization is a constructive model benchmarked on an external IEEE 14-bus case; the notable issue is a missing critical-node membership constraint, which is a modeling gap rather than a circular reduction.

full rationale

Walking the paper's derivation chain, no claimed prediction reduces by construction to fitted inputs or to a self-citation chain. The cyber-topology objective (6) and constraints (7)-(10) are a standard flow-based arborescence formulation inspired by Spanning Tree Protocol [11]; the physical layer is a standard ACOPF with objective (2) and constraints (1), (3), (4); the bi-level problem (14) combines these with stated weighting coefficients alpha. Neither the cyber nor the physical optimization is calibrated to the outcome it later reports (voltage recovery or ESS dispatch), and the numerical section uses the external IEEE 14-bus case, MATPOWER-based cost data, and the external solver IPOPT. The authors' prior work is cited only for background context (e.g., [1], [3], [5], [15]); none of these citations supplies a load-bearing uniqueness or existence result that forces the paper's choice of model, and no fitted parameter is renamed as a prediction. The skeptical reading of the manuscript identifies a genuine modeling gap: Section III-B asserts that the predefined critical set K 'must be active and connected when constructing the cyber topology,' but problem (11) and the upper level of problem (14) constrain only (7)-(10), so nothing in the written formulation forces x_i = 1 for i in K; with positive deployment costs the optimizer can satisfy all written constraints with only the root node active. This is an omitted constraint or correctness risk, not a circular step, because the asserted guarantee is not equivalent to the model's inputs by construction and the reported resilience improvement is not an artifact of a fitted parameter or a self-citation. Accordingly, the circularity score is 0.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The framework introduces no new physical entities. It relies on standard cyber-physical infrastructure models, but the omission of the critical-node activation constraint is an unstated assumption that the connectivity guarantee depends on.

free parameters (1)

Balancing coefficients α1, α2, α3 = Not reported
The bi-level objective in (14c) scales cyber, physical, and resilience costs by these coefficients; the simulation uses unspecified values, and the reported results may depend on their choice.

assumptions (3)

domain assumption The cyber communication network can be represented as a rooted directed arborescence with unit flows, as in the STP flow model in (7)-(10).
Section II-B assumes information flows from a root node to every active node and that this model captures real cyber network operation (STP). This is a modeling assumption that may not hold for arbitrary communication networks with loops or meshed topologies.
ad hoc to paper Critical cyber nodes must be active, x_i=1 for i in K, although this constraint is not written in problem (11).
The stated constraints (7)-(10) do not force critical nodes to be active; the paper's connectivity guarantee relies on this unstated condition.
domain assumption A neighboring cyber node m in M_ic can fully assume the compromised node's control functions, including controlling the ESS at the attacked bus.
Section III-A and Algorithm 1 replace the compromised node with a neighbor and then let that neighbor control physical resources; this assumes functional equivalence of cyber nodes.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Cooperative Optimization of Grid-Edge Cyber and Physical Resources for Resilient Power System Operation." pith.science (2026). https://pith.science/paper/JWK5JQGS

@misc{pith2026250205768,
  author       = {Pith},
  title        = {Pith review of: Cooperative Optimization of Grid-Edge Cyber and Physical Resources for Resilient Power System Operation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JWK5JQGS}},
  note         = {Machine review of arXiv:2502.05768}
}

read the original abstract

The cooperative operation of grid-edge power and energy resources is crucial to improving the resilience of power systems during contingencies. However, given the complex cyber-physical nature of power grids, it is hard to respond timely with limited costs for deploying additional cyber and/or phyiscal resources, such as during a high-impact low-frequency cyber-physical event. Therefore, the paper examines the design of cooperative cyber-physical resource optimization solutions to control grid-tied cyber and physical resources. First, the operation of a cyber-physical power system is formulated into a constrained optimization problem, including the cyber and physical objectives and constraints. Then, a bi-level solution is provided to obtain optimal cyber and physical actions, including the reconfiguration of cyber topology (e.g., activation of communication links) in the cyber layer and the control of physical resources (e.g., energy storage systems) in the physical layer. The developed method improves grid resilience during cyberattacks and can provide guidance on the control of coupled physical side resources. Numerical simulation on a modified IEEE 14-bus system demonstrates the effectiveness of the proposed approach.

Figures

Figures reproduced from arXiv: 2502.05768 by the authors.

**Figure 2.** Construction of cyber communication networks (black [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗

**Figure 3.** Active power generation of all generators after the [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Active charging/discharging power of the ESS, and [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

16 extracted references · 16 canonical work pages

[1]

A cyber-physical modeling and assessment framework for power grid infrastructures,

K. R. Davis, C. M. Davis, S. A. Zonouz, R. B. Bobba, R. Berthier, L. Garcia, and P. W. Sauer, “A cyber-physical modeling and assessment framework for power grid infrastructures,” IEEE Trans. Smart Grid , vol. 6, no. 5, pp. 2464–2475, 2015

work page 2015
[2]

Learning- based predictive control via real-time aggregate flexibility,

T. Li, B. Sun, Y . Chen, Z. Ye, S. H. Low, and A. Wierman, “Learning- based predictive control via real-time aggregate flexibility,” IEEE Trans. Smart Grid , vol. 12, no. 6, pp. 4897–4913, 2021

work page 2021
[3]

A review of scalable and privacy-preserving multi-agent frameworks for distributed energy resources,

X. Huo, H. Huang, K. R. Davis, H. V . Poor, and M. Liu, “A review of scalable and privacy-preserving multi-agent frameworks for distributed energy resources,” Adv. Appl. Energy , p. 100205, 2024

work page 2024
[4]

A review of the measures to enhance power systems resilience,

M. Mahzarnia, M. P. Moghaddam, P. T. Baboli, and P. Siano, “A review of the measures to enhance power systems resilience,” IEEE Syst. J , vol. 14, no. 3, pp. 4059–4070, 2020

work page 2020
[5]

Exploration of for-purpose decentral- ized algorithmic cyber attacks in EV charging control,

M. F. Fard, X. Huo, and M. Liu, “Exploration of for-purpose decentral- ized algorithmic cyber attacks in EV charging control,” in Proc. 32nd Int. Symp. Ind. Electron. , Helsinki, Finland, Jun. 19-21 2023, pp. 1–6

work page 2023
[6]

Generation of firewall configurations for a large scale synthetic power system,

A. Sahu, P. Wlazlo, N. Gaudet, A. Goulart, E. Rogers, and K. Davis, “Generation of firewall configurations for a large scale synthetic power system,” in Proc. IEEE Tex. Power Energy Conf. , College Station, TX, USA, Feb. 28-Mar. 01 2022, pp. 1–6

work page 2022
[7]

Reliability as- sessment of cyber-physical power systems considering the impact of predicted cyber vulnerabilities,

A. Rostami, M. Mohammadi, and H. Karimipour, “Reliability as- sessment of cyber-physical power systems considering the impact of predicted cyber vulnerabilities,” Int. J. Electr . Power Energy Syst. , vol. 147, p. 108892, 2023

work page 2023
[8]

A tri-level optimization model to mitigate coordinated attacks on electric power systems in a cyber-physical environment,

K. Lai, M. Illindala, and K. Subramaniam, “A tri-level optimization model to mitigate coordinated attacks on electric power systems in a cyber-physical environment,”Appl. Energy, vol. 235, pp. 204–218, 2019

work page 2019

Show all 16 references

[9]

Decision support framework for resilience-oriented cost-effective distributed generation expansion in power systems,

T. R. B. Kushal and M. S. Illindala, “Decision support framework for resilience-oriented cost-effective distributed generation expansion in power systems,” IEEE Trans. Ind. Appl. , vol. 57, no. 2, pp. 1246–1254, 2020

2020
[10]

History of optimal power flow and formulations,

M. B. Cain, R. P. O’neill, A. Castillo et al. , “History of optimal power flow and formulations,” FERC, vol. 1, pp. 1–36, 2012

2012
[11]

Spanning Tree Protocols

“Spanning Tree Protocols.” [Online]. Avail- able: https://www.ieee802.org/1/files/public/docs2009/ aq-seaman-merged-spanning-tree-protocols-0509.pdf
[12]

Analyzing the cyber-physical impact of cyber events on the power grid,

R. Liu, C. Vellaithurai, S. S. Biswas, T. T. Gamage, and A. K. Srivastava, “Analyzing the cyber-physical impact of cyber events on the power grid,” IEEE Trans. Smart Grid , vol. 6, no. 5, pp. 2444–2453, 2015

2015
[13]

IEEE 14-bus system

Illinois Information Trust Institute, “IEEE 14-bus system.” [Online]. Available: https://icseg.iti.illinois.edu/ieee-14-bus-system/
[14]

CASE14 power flow data for IEEE 14 bus test case

Matpower, “CASE14 power flow data for IEEE 14 bus test case.” [Online]. Available: https://matpower.org/docs/ref/matpower5.0/case14. html
[15]

Generalized contingency analysis based on graph theory and line outage distribution factor,

M. R. Narimani, H. Huang, A. Umunnakwe, Z. Mao, A. Sahu, S. Zonouz, and K. Davis, “Generalized contingency analysis based on graph theory and line outage distribution factor,” IEEE Syst. J. , vol. 16, no. 1, pp. 626–636, 2021

2021
[16]

Large-scale nonlinear programming using IPOPT: An integrating framework for enterprise-wide dynamic optimization,

L. T. Biegler and V . M. Zavala, “Large-scale nonlinear programming using IPOPT: An integrating framework for enterprise-wide dynamic optimization,” Comput. Chem. Eng. , vol. 33, no. 3, pp. 575–582, 2009

2009

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.

[1] [1]

A cyber-physical modeling and assessment framework for power grid infrastructures,

K. R. Davis, C. M. Davis, S. A. Zonouz, R. B. Bobba, R. Berthier, L. Garcia, and P. W. Sauer, “A cyber-physical modeling and assessment framework for power grid infrastructures,” IEEE Trans. Smart Grid , vol. 6, no. 5, pp. 2464–2475, 2015

work page 2015

[2] [2]

Learning- based predictive control via real-time aggregate flexibility,

T. Li, B. Sun, Y . Chen, Z. Ye, S. H. Low, and A. Wierman, “Learning- based predictive control via real-time aggregate flexibility,” IEEE Trans. Smart Grid , vol. 12, no. 6, pp. 4897–4913, 2021

work page 2021

[3] [3]

A review of scalable and privacy-preserving multi-agent frameworks for distributed energy resources,

X. Huo, H. Huang, K. R. Davis, H. V . Poor, and M. Liu, “A review of scalable and privacy-preserving multi-agent frameworks for distributed energy resources,” Adv. Appl. Energy , p. 100205, 2024

work page 2024

[4] [4]

A review of the measures to enhance power systems resilience,

M. Mahzarnia, M. P. Moghaddam, P. T. Baboli, and P. Siano, “A review of the measures to enhance power systems resilience,” IEEE Syst. J , vol. 14, no. 3, pp. 4059–4070, 2020

work page 2020

[5] [5]

Exploration of for-purpose decentral- ized algorithmic cyber attacks in EV charging control,

M. F. Fard, X. Huo, and M. Liu, “Exploration of for-purpose decentral- ized algorithmic cyber attacks in EV charging control,” in Proc. 32nd Int. Symp. Ind. Electron. , Helsinki, Finland, Jun. 19-21 2023, pp. 1–6

work page 2023

[6] [6]

Generation of firewall configurations for a large scale synthetic power system,

A. Sahu, P. Wlazlo, N. Gaudet, A. Goulart, E. Rogers, and K. Davis, “Generation of firewall configurations for a large scale synthetic power system,” in Proc. IEEE Tex. Power Energy Conf. , College Station, TX, USA, Feb. 28-Mar. 01 2022, pp. 1–6

work page 2022

[7] [7]

Reliability as- sessment of cyber-physical power systems considering the impact of predicted cyber vulnerabilities,

A. Rostami, M. Mohammadi, and H. Karimipour, “Reliability as- sessment of cyber-physical power systems considering the impact of predicted cyber vulnerabilities,” Int. J. Electr . Power Energy Syst. , vol. 147, p. 108892, 2023

work page 2023

[8] [8]

A tri-level optimization model to mitigate coordinated attacks on electric power systems in a cyber-physical environment,

K. Lai, M. Illindala, and K. Subramaniam, “A tri-level optimization model to mitigate coordinated attacks on electric power systems in a cyber-physical environment,”Appl. Energy, vol. 235, pp. 204–218, 2019

work page 2019