Simulative Analysis of Multi-Agent Systems in Energy Systems: Impact

of Communication Networks

Malin Radtke

and Emilie Frost

Distributed Artiﬁcial Intelligence, OFFIS - Institute for Information Technology, Oldenburg, Germany

{malin.radtke, emilie.frost}@ofﬁs.de

Keywords:

Multi-Agent Systems, Simulative Analysis, Cyber-Physical Energy Systems, Communication Systems.

Abstract:

This paper addresses the growing use of Multi-Agent Systems (MASs) in power systems, particularly within

the context of Cyber-Physical Energy Systems (CPES). The reliance on Information and Communication

Technologies (ICT) is critical for coordinating control and ensuring reliable information exchange. Disruptions

in the ICT system can degrade overall system performance. Given the complexity of these systems, systematic

testing and accurate simulation of MAS behavior under the inﬂuence of communication networks are essential

to ensure stability and security of supply. The paper provides a structured perspective on how to analyze

MASs performance under different communication conditions in CPES, offering recommendations based on

literature. It serves as a guide to understanding the challenges posed by the integration of ICT into power

systems, with guidelines that can be used and extended to evaluate and improve system performance.

1 INTRODUCTION

Multi-Agent Systems (MASs) have a wide range

of applications in various domains, including smart

grids, smart manufacturing, sensor networks, and in-

telligent transportation systems (Zhang et al., 2021).

Extensive literature highlights the use of MASs in

these ﬁelds, particularly in smart grids, where re-

searchers are exploring their broad potential (Mahela

et al., 2022).

As these applications evolve, the integration of In-

formation and Communication Technology (ICT) into

power systems is becoming increasingly critical. Tra-

ditional power systems, once reliant solely on phys-

ical infrastructure, are now increasingly integrated

with ICT, encompassing advanced control, com-

puting, and communication functions (Yohanandhan

et al., 2020). This integration has led to the devel-

opment of Cyber-Physical Energy Systems (CPES),

which merge physical components of the power grid

with cyber systems that monitor, control, and opti-

mize the performance of the entire network (Hasanuz-

zaman Shawon et al., 2019).

The reliance on ICT for coordinating control and

information exchange among agents in CPES means

that any disruption in communication networks can

lead to signiﬁcant system degradation (Zhou et al.,

https://orcid.org/0009-0009-9902-1744

https://orcid.org/0000-0003-4791-2333

2021). For instance, unexpected physical faults or

cyber attacks can compromise the system’s ability

to maintain stability, efﬁciency, and security (Zhang

et al., 2021). The complexity of these interactions

highlights the need for systematic testing and mod-

eling methods that can accurately simulate the be-

havior of MASs under various network conditions

(Yohanandhan et al., 2020).

Given these challenges, it is crucial to conduct a

thorough and systematic investigation of the complex

interactions between MASs and the communication

networks within CPES. This investigation should fo-

cus on understanding how communication networks

inﬂuence the overall performance and reliability of

MASs. Moreover, there is a growing need to develop

robust strategies and architectures that can mitigate

these risks and ensure the secure operation of CPES

(Wang and Govindarasu, 2020a).

In this context, this position paper introduces a

structured methodology for simulating and analyzing

MASs in CPES, with an emphasis on how communi-

cation networks impact their performance and relia-

bility. Rather than delivering deﬁnitive solutions, this

paper offers a structured perspective on how such an

analysis might be conducted. Speciﬁcally, the mor-

phological box presented herein is not intended to be

prescriptive, but rather serves as a theoretical guide

based on current literature, reﬂecting one possible ap-

proach to tackling these complex issues.

Radtke, M. and Frost, E.

Simulative Analysis of Multi-Agent Systems in Energy Systems: Impact of Communication Networks.

DOI: 10.5220/0013241400003890

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 17th International Conference on Agents and Artiﬁcial Intelligence (ICAART 2025) - Volume 1, pages 523-530

ISBN: 978-989-758-737-5; ISSN: 2184-433X

523

By examining various parameters and scenarios

that impact MAS performance, this paper seeks to

provide a conceptual framework to understand and

address the challenges posed by the integration of ICT

in modern power systems.

In particular, the paper includes research on re-

lated work for analysis of MASs, as discussed in sec-

tion 2. In section 3, selected literature is used to illus-

trate the basis for the proposed morphological box,

which demonstrates key objectives and design deci-

sions critical to such a simulative analysis. Finally,

the implications of these ﬁndings are discussed and

concluded in section 4.

2 RELATED WORK

In this section, we categorize and identify contri-

butions relevant to the simulative analysis of MASs

in CPES under the inﬂuence of communication net-

works. Additionally, we highlight limitations in ex-

isting related work, that provide a foundation for the

more detailed analysis guidance.

Agent-Based Applications in CPES.

Hasanuzzaman Shawon et al. (2019) and Mahela

et al. (2022) provide comprehensive reviews of

agent-based applications in smart grids, focusing

on MAS concepts, design, challenges, and tech-

nological frameworks. Agents are categorized

based on their roles in energy management (e.g.,

pricing, scheduling) and on ensuring reliability

and security (e.g., fault handling). However, they

do not delve into how communication network

issues impact system performance, particularly

under fault conditions or cyber attacks. This

paper builds on their insights by speciﬁcally

addressing these gaps through the proposed

simulative analysis guideline.

Safety and Security Analysis in MASs. Zhang

et al. (2021) provide an in-depth survey on

the safety and security of MASs, with a focus

on fault estimation, detection, diagnosis, and

fault-tolerant control. While this research offers a

strong theoretical framework, its abstract nature

limits practical applicability, particularly in

energy systems. The focus is primarily on fault

detection rather than simulative investigations

that explore MAS behavior under real-world

fault and attack scenarios. Yohanandhan et al.

(2020) review various modeling and simulation

methods related to cybersecurity in CPES. This

includes a detailed overview of cyber attack types,

such as Denial of Service (DoS), Malware, and

Man-in-the-Middle attacks, and their potential

impacts on physical systems, including stability

and economic factors. However, the review

remains general, lacking speciﬁc guidelines or

best practices for analyzing MASs under the

inﬂuence of communication networks.

Research Gap. The reviewed literature provides

valuable foundations for understanding MASs in

CPES, safety and security concerns, and the role

of ICT. However, there are notable gaps, particu-

larly in the simulative analysis of MAS behavior

under the inﬂuence of communication networks.

Existing research tends to focus either on ab-

stract, theoretical concepts or general cybersecu-

rity in CPES without providing speciﬁc guidelines

or practical approaches for investigating MASs in

the context of energy systems. This position paper

aims to ﬁll these gaps by proposing a comprehen-

sive guide for simulative analysis, addressing the

key objectives and design decisions necessary for

robust and reliable MAS operation in CPES.

3 SIMULATIVE ANALYSIS OF

MAS

The analysis of MASs in CPES necessitates a com-

prehensive understanding of the role communication

networks play in determining the performance and be-

havior of these agent systems. Communication net-

works, which serve as the critical infrastructure for

data exchange between agents, can signiﬁcantly im-

pact system stability, reliability, and efﬁciency. To in-

vestigate these impacts, a simulative analysis offers a

robust approach to assess how variations in commu-

nication conditions, such as latency, packet loss, or

network failures, inﬂuence MAS performance.

This section presents a structured methodology

for conducting a simulative analysis of MASs under

the inﬂuence of communication networks in CPES.

Drawing on insights from existing literature, the guid-

ance is divided into several key areas:

• the deﬁnition of analysis objectives,

• the speciﬁcation of system requirements and com-

munication threats,

• and the design of the simulation study itself.

Using a morphological box (see Figure 1), we cate-

gorize critical parameters that affect MASs in CPES,

providing a comprehensive guideline to systemati-

cally assess the interplay between MASs and com-

munication networks. The aforementioned key areas

and parameters are presented and detailed below, as

shown in Figure 1.

ICAART 2025 - 17th International Conference on Agents and Artiﬁcial Intelligence

524

Analysis

Objective

(Analysis)

Metrics

Multi-agent system in Cyber-physical Energy Systems

Analysis of MAS-behavior under

communication network influence

may be violated through

Requirements

lead to

Threats

Objective

(MAS)

Fault location Fault isolation

Fault reaction

Monitoring

Control

Security Safety

ReliabilityPerformanceFunctionality Resilience

Robustness

Autonomy

Cyber attacksCommunication system influences

Faults

Deception

attacks

Communication

Modeling Method

Detailed Simulation

Abstract application

representation

Communication system

parameter

Power system parameter

(Control) Algorithm

parameter

Illustrative example

Comprehensive

parameter analysis

Normal

operation

Graph-based model

Evaluate proposed architecture for

fault location/isolation/reaction

Study

Effects

Design

Baseline and

boundaries

Power system

topology & configuration

Communication system

topology & configuration

Denial-of-Service

attack

...

Hardware

component faults

Failures

Increased

message volume

Traffic Delays

Packet losses

Agent system

configurations

...

Availability Integrity

Figure 1: Overview of important parameters in the simulative analysis of MASs in CPES.

3.1 Analysis: Objective

The analysis objective addresses the purpose of con-

ducting the simulative analysis itself. It deﬁnes what

the analysis aims to achieve, such as understanding

how communication network conditions affect MAS

behavior or assessing the efﬁcacy of fault-handling ar-

chitectures.

One key objective when conducting a simulative anal-

ysis may be to understand how MAS behavior and

performance are inﬂuenced by different communica-

tion network conditions. This process assists in the

identiﬁcation of potential vulnerabilities and the opti-

mization of the system. It includes investigating the

impact of communication network performance met-

rics such as latency and packet drops, as well as fail-

ures or attacks on power system applications.

For example, Xiahou et al. (2021) studied the reli-

ability of power systems under random delay attacks,

while Radtke et al. (2023) investigated the inﬂuence

of a realistic communication network on the power

system applications redispatch and voltage control

under varying communication network conditions.

Secondly, the evaluation of the proposed architec-

tures for fault detection, isolation, and reaction pro-

vides valuable insights into the effectiveness of MASs

Simulative Analysis of Multi-Agent Systems in Energy Systems: Impact of Communication Networks

525

in handling communication faults and cyber attacks.

For instance, Wang and Govindarasu (2020a) eval-

uated their proposed anomaly detection mechanism,

while Albarakati et al. (2023) demonstrated the efﬁ-

cacy of adaptive protection mechanisms. Other stud-

ies, such as the work of Ilyes et al. (2019), focused

on frameworks to detect and reconﬁgure systems af-

ter cyber attacks and physical faults.

3.2 System: Objective

In contrast to the objective of analysis, the objective of

the system focuses on the intended goals of the MAS

within the context of the energy system. This subsec-

tion highlights the primary functions that the MAS

is designed to perform, such as energy management

or fault detection and reaction (Mahela et al., 2022),

and emphasizes how these speciﬁc system objectives

shape the focus of the analysis.

If the primary objective of the MAS is related to

monitoring and control functions in the power grid

(such as in Xiahou et al. (2021); Yang et al. (2020)),

the analysis would usually focus primarily on under-

standing the MAS behavior under the inﬂuence of

communication networks.

Conversely, when the objective of the agent sys-

tem lies more in fault location, isolation, and reaction

(such as in Albarakati et al. (2023); Ilyes et al. (2019);

Abdelhamid et al. (2022)), the analysis shifts towards

evaluating proposed architectures for handling com-

munication faults and responding to cyber attacks. In

this case, the focus is on testing the effectiveness of

the agent’s ability to detect, isolate, and recover from

faults in communication networks in the analysis.

3.3 System: Requirements

In the context of energy systems, different require-

ments must be met by MASs. Depending on the ob-

jective in the development of the MAS and the appli-

cation, these requirements differ. Many approaches

consider security as a requirement for MASs. Ac-

cording to Hines et al. (2014), security in the context

of power systems means that no component will cause

the system to violate the operating limits. In the con-

text of smart grids, security often refers to cyber secu-

rity, as designing communication systems resistant to

attempted cyber attacks. This is also done in the liter-

ature, as many approaches consider security of MASs

against cyber attacks (Ilyes et al., 2019; Zhou et al.,

2021; Wang and Govindarasu, 2020a; Choi et al.,

2020) or faults (Abdelhamid et al., 2022). Accord-

ing to Hines et al. (2014), security is often associated

with robustness, meaning that a secure system is ro-

bust against attacks or failures. Therefore, similar to

security, robustness is also required for MASs. Al-

barakati et al. (2023) deﬁne robustness as the system’s

ability to cope with disturbances while maintaining

functionality. Thus, approaches in the literature in-

vestigate the robustness of their MAS regarding faults

and attacks, for example, to ensure that the system is

robust under packet drops or time delays (Oest et al.,

2021; Yang et al., 2020).

Additionally, reliability is considered a require-

ment for MASs, which can be deﬁned as continuous

delivery of a correct service (Sanislav et al., 2017).

The actual service depends on the application of the

MAS, as service availability, latency, duration of out-

ages (Hines et al., 2014) or the delivery of electric-

ity (Arghandeh et al., 2016). In the literature, relia-

bility of MASs is considered regarding cyber attacks

(Fang et al., 2017; Sanislav et al., 2017; Choi et al.,

2020; Oest et al., 2021) or evaluated for distributed

approaches (Kou et al., 2021). Similar to the previ-

ous metrics, functionality is a requirement for MASs

(Zhu et al., 2019). Robust systems are maintaining

functionality during disturbances (Arghandeh et al.,

2016). However, the deﬁnition of functionality is also

depending on the application and the system itself.

The performance is often analyzed as a requirement

for MASs, however, the interpretation depends on the

use case. For different applications, different met-

rics are deﬁned, such as solution quality (Oest et al.,

2021). Furthermore, resilience, meaning the ability of

a system to recover from a failure (Hines et al., 2014)

is required in MASs (Wang and Govindarasu, 2020a).

This includes for example service restoration (Ab-

delhamid et al., 2022). Other requirements are also

possible, depending on the application and objective.

These include for example autonomy (Sanislav et al.,

2017; Oest et al., 2021), safety (Sanislav et al., 2017;

Ilyes et al., 2019), availability (Fang et al., 2017; Kou

et al., 2021) or integrity (Fang et al., 2017).

3.4 System: Threats

The aforementioned requirements that apply to MASs

in energy systems may be violated through a variety

of threats concerning the impact of the communica-

tion network on the system. These threats can be clas-

siﬁed into two main categories: those that affect the

system under normal operating conditions or in the

event of technical faults, and those that are of a cyber

attack nature.

In examining the impact of communication net-

works under normal operational conditions, it is pos-

sible to consider the integration of packet drops and

packet delays in message dispatch (Yang et al., 2020),

ICAART 2025 - 17th International Conference on Agents and Artiﬁcial Intelligence

526

as well as the evaluation of system behavior under

the inﬂuence of different communication technolo-

gies (Oest et al., 2021). Furthermore, the inﬂuence of

technical faults, such as communication link or base

station failure, may also be analyzed (Abdelhamid

et al., 2022; Oest et al., 2021).

Cyber attacks have recently been perceived as par-

ticularly threatening and therefore have also been an-

alyzed regarding their impact on MASs. Fang et al.

(2017) regard cyber attacks in general, while other

authors focus on speciﬁc attacks, as deception at-

tacks (Wang and Govindarasu, 2020a). Often inves-

tigated are DoS attacks, as the effects of such an at-

tack can take many forms, such as system failures or

overloads (Wang and Govindarasu, 2020a; Albarakati

et al., 2023). Other attacks deal with false data be-

ing transferred via the system. As this also impacts

a MAS, these types of attacks are extensively inves-

tigated in the scientiﬁc literature. Most of the time,

false data injection attacks are considered in this case

(Ilyes et al., 2019; Zhou et al., 2021; Choi et al., 2020)

or man-in-the-middle attacks (Sanislav et al., 2017).

Other approaches investigate false data or data manip-

ulations in speciﬁc areas, such as alert manipulation

attacks (Zhou et al., 2021) or conﬁguration change at-

tacks (Sanislav et al., 2017).

3.5 System: Effects

While some approaches investigate the threat itself,

others focus on the effect of such threats. Differ-

ent threats have different effects on the system. The

investigated effects are hardware component faults

due to attacks (Sanislav et al., 2017), failures (Wang

and Govindarasu, 2020b), increased message volume

(Frost. et al., 2024), trafﬁc (Radtke et al., 2023), de-

lays or packet losses (Oest et al., 2021; Frost et al.,

2020; Yang et al., 2020; Xu et al., 2021). Several of

these effects are caused by multiple threats, such as

failures or attacks, as failures of assets can be caused

by environment or denial of service attacks.

3.6 Study: Baseline and Boundaries

Establishing the baseline and boundaries for the sim-

ulative analysis is a critical ﬁrst step to accurately as-

sess the behavior of MASs within CPES in a simula-

tion study. This process involves deﬁning the initial

conﬁguration of the system, agent behavior, charac-

teristics of the communication network, and the topol-

ogy of the power system. By setting a clear baseline,

the analysis can systematically vary different param-

eters to assess the system performance under a range

of conditions.

The baseline should include a well-deﬁned model

of the MAS, specifying the number and type of

agents, and the speciﬁc objectives they are designed

to achieve (e.g., fault detection, optimization, or con-

trol). For example, studies such as Sanislav et al.

(2017) and Oest et al. (2021) include detailed conﬁg-

urations of agent systems and communication tech-

nologies, which serve as starting points for their anal-

yses.

The topology and conﬁguration of the communi-

cation system are also critical components of the base-

line. This includes deﬁning communication technolo-

gies and network architectures. Variations in commu-

nication conditions, such as the introduction of delays

or failures, can then be explored based on this base-

line conﬁguration. For instance, Frost et al. (2024)

assessed the inﬂuence of a wired communication net-

work with ring and star topologies in three distinct

scenarios: an ideal communication environment, an

unimpaired scenario, and an attack situation.

Finally, the power system topology and conﬁgu-

ration must be speciﬁed, including details about the

physical grid model, such as the number of buses,

feeders, and generators. The baseline might be based

on real-world systems, such as the distribution feeder

used in Albarakati et al. (2023) or standard models

such as the IEEE 9-bus system (Ilyes et al., 2019).

By carefully deﬁning the baseline and boundaries,

the analysis can ensure that any variations introduced

in the scenarios (whether related to the agent system,

communication conditions, or power system conﬁgu-

rations) are meaningfully compared against a consis-

tent reference point. This approach facilitates a com-

prehensive understanding of how different parameters

inﬂuence the overall performance and reliability of

the MAS within CPES.

3.7 Study: Communication Modeling

Method

An important aspect of simulative analysis is de-

termining the method for communication modeling.

Different approaches exist, each varying in complex-

ity, accuracy, and suitability depending on the purpose

of the analysis. The choice of method affects the level

of detail that can be captured regarding the inﬂuence

of communication networks on MASs in CPES.

One common approach is the integration of Key

Performance Indicators (KPIs). This method mod-

els basic communication parameters such as delays

or packet losses using stochastic distributions, simu-

lating network inﬂuence without detailed network be-

havior. While this approach is simpler and less com-

putationally expensive, it may lack the realism re-

Simulative Analysis of Multi-Agent Systems in Energy Systems: Impact of Communication Networks

527

quired to model complex interactions. For instance,

Xiahou et al. (2021) use this approach to integrate

random delay attacks into their analysis, providing a

general understanding of how such disruptions affect

system performance.

A more comprehensive approach is a detailed sim-

ulation conducted using specialized communication

network simulation frameworks. This method allows

for modeling speciﬁc communication technologies,

protocols, and phenomena like interference, provid-

ing a more accurate and realistic representation of

network behavior. However, it comes with draw-

backs, such as increased complexity, higher compu-

tational overhead, and potentially costly licenses for

simulation tools. Studies like Wang and Govindarasu

(2020a) and Oest et al. (2021) demonstrate this ap-

proach, leveraging detailed simulations to evaluate

performance under varying network conditions.

Alternatively, the abstract application representa-

tion method focuses on the effects of the communi-

cation network on the messages exchanged by agents,

without explicitly modeling the network itself. This

approach simpliﬁes the analysis by assuming certain

conditions, such as attack probabilities or transmis-

sion errors, and analyzing their impact on system per-

formance. This method is used in studies like Fang

et al. (2017), where attack probabilities are assumed

to study system behavior, and in Albarakati et al.

(2023), which models cyber attacks in data transmis-

sion without detailed network modeling.

Some studies also use graph-based models to rep-

resent communication networks. In this method,

nodes in the graph represent communication nodes

(e.g., agents), and edges represent communication

links, which may be weighted based on network prop-

erties such as bandwidth or latency. This approach

provides a ﬂexible way to analyze network topologies

and their impact on communication. For example, Es-

lami et al. (2022) and Yang et al. (2020) use graph-

based models to represent and simulate network per-

formance, particularly in scenarios involving packet

drops or communication delays.

3.8 Study: Metrics

The selection of appropriate metrics is a critical com-

ponent of study design, as it deﬁnes how the perfor-

mance of the MAS under the inﬂuence of the com-

munication network will be evaluated. These metrics

should be aligned with the overall objectives of the

analysis and tailored to the speciﬁc scenario under in-

vestigation. Typically, the metrics can be categorized

based on their origin: communication system param-

eters, power system parameters, and MAS or control

algorithm performance.

The parameters of the communication system are

crucial to assess how the network inﬂuences MAS

performance. These metrics typically focus on net-

work delays, latency, and message losses. For ex-

ample, Xiahou et al. (2021) measure delay times to

evaluate the effects of random delay attacks, while

Oest et al. (2021) examine end-to-end latency in dif-

ferent communication technologies to assess overall

communication performance.

The parameters of the power system provide in-

sight into how the grid itself is affected by commu-

nication issues or cyber attacks. Metrics such as cur-

rent levels, breaker status during disruptions, or power

output of generators are often used to evaluate grid

stability and resilience. Studies like Albarakati et al.

(2023) measure breaker status during cyber attacks,

while Yang et al. (2020) track the optimized power

output of generators to assess system performance un-

der various network conditions.

Agent system and control algorithm parameters

are essential for evaluating how well the agents or

control algorithms perform within the system (under

the inﬂuence of the communication network). Met-

rics in this category might include fault location ac-

curacy, failure rates, and optimization run-times. For

instance, Wang and Govindarasu (2020a) evaluate

anomaly detection performance, and Albarakati et al.

(2023) measure fault location accuracy. Oest et al.

(2021) focus on metrics like negotiation times and

error rates for optimization tasks, while Ilyes et al.

(2019) track system reliability and failure rates.

3.9 Study: Design

The design of the study is important for shaping the

insights that can be derived from the simulative anal-

ysis. Two common approaches are illustrative case

studies and comprehensive parameter analysis.

An illustrative example or case study is often used

to showcase the behavior of the system under a spe-

ciﬁc set of conditions. These case studies are typically

designed around predeﬁned scenarios, such as cyber

attacks, communication faults, or system failures, and

are meant to provide qualitative insights into the sys-

tem’s performance in speciﬁc contexts. For example,

Fang et al. (2017) explore ﬁve different attack sce-

narios, while Wang and Govindarasu (2020a) focus

on a single attack scenario to evaluate the system’s

response. Similarly, Albarakati et al. (2023) present

three different scenarios, and Zhou et al. (2021) eval-

uate different fault and attack types. These case stud-

ies offer targeted insights, but may not capture the full

range of system behaviors.

ICAART 2025 - 17th International Conference on Agents and Artiﬁcial Intelligence

528

In contrast, a comprehensive parameter analysis

systematically explores how changes in key parame-

ters affect system performance across a wider range of

conditions. This approach involves running multiple

simulations while varying parameters such as agent

system size, network conditions, or disturbance inten-

sity to capture a more holistic understanding of sys-

tem dynamics. For instance, Kou et al. (2021) con-

duct an extensive analysis of reliability metrics, and

Oest et al. (2021) investigate how different agent sizes

and communication technologies impact system per-

formance in various scenarios.

Both approaches have their advantages. Case

studies are ideal for demonstrating the system re-

sponse to speciﬁc conditions, while comprehensive

parameter analysis allows for a more generalized un-

derstanding of system behavior under a range of dis-

turbances. The choice between these approaches

should be aligned with the objective of the analysis

and the level of detail required for the analysis.

4 DISCUSSION AND

CONCLUSION

A principal conclusion to be drawn from this guide-

line for simulative analysis is the necessity for

MASs to be communication-aware. This implies that

decision-making algorithms in energy systems should

be constructed in a manner that enables them to func-

tion with incomplete or delayed information. The

concrete deﬁnition of requirements for the system al-

lows the identiﬁcation of potential impairments re-

sulting from the inﬂuence of the simulated communi-

cation network. This enables analyzing the system’s

behavior under adverse conditions (e.g., due to cyber

attacks) and the implementation of suitable security

mechanisms. The use of appropriate detection mech-

anisms allows for the mitigation of faults and attacks

before system stability is compromised. Investiga-

tions into the scalability of such systems also beneﬁt

from the simulation analysis proposed, as these stud-

ies highlight the importance of considering potential

(communication) bottlenecks in large systems. Fur-

thermore, simulative analysis of MASs under inﬂu-

ences of the communication network assists in select-

ing appropriate communication technologies and pro-

tocols by evaluating and comparing diverse options.

Although the guideline provides a solid founda-

tion, it may not cover all possible scenarios or sys-

tem conﬁgurations and must be applied to real-world

use cases. Users are encouraged to expand and adapt

the guideline based on their own needs, incorporating

additional parameters, or exploring new technologies,

such as emerging cyber threats and communication

technologies.

In summary, this paper serves as a guide for under-

standing and addressing the challenges posed by the

interconnections of ICT and power system, and pro-

poses a structured approach to evaluating the perfor-

mance of MASs under communication conditions in

CPES in a simulative analysis. Recommendations are

given on what aspects should be considered in such an

analysis, based on the current literature. Future work

may include examples of the practical application of

these guidelines to new research projects.

ACKNOWLEDGEMENTS

This manuscript is based on a project funded by the

Federal Ministry for Economic Affairs and Climate

Action as part of the “Edge Data Economy” tech-

nology program. We gratefully acknowledge our

”DEER” project partners’ support in this research.

The authors would like to thank the German Fed-

eral Government, the German State Governments,

and the Joint Science Conference (GWK) for their

funding and support as part of the NFDI4Energy con-

sortium. The work was partially funded by the Ger-

man Research Foundation (DFG) – 501865131 within

the German National Research Data Infrastructure

(NFDI, www.nfdi.de).

REFERENCES

Abdelhamid, A. M., Zakzouk, N. E., and El Safty, S. (2022).

A Multi-Agent Approach for Self-Healing and RES-

Penetration in Smart Distribution Networks. Mathe-

matics, 10(13).

Albarakati, A. J., Azeroual, M., Boujoudar, Y., EL Iysaouy,

L., Aljarbouh, A., Tassaddiq, A., and EL Markhi,

H. (2023). Multi-Agent-Based Fault Location and

Cyber-Attack Detection in Distribution System. En-

ergies, 16(1).

Arghandeh, R., Von Meier, A., Mehrmanesh, L., and Mili,

L. (2016). On the deﬁnition of cyber-physical re-

silience in power systems. Renewable and Sustainable

Energy Reviews, 58:1060–1069.

Choi, I.-S., Hong, J., and Kim, T.-W. (2020). Multi-

Agent Based Cyber Attack Detection and Mitigation

for Distribution Automation System. IEEE Access,

8:183495–183504. Conference Name: IEEE Access.

Eslami, A., Abdollahi, F., and Khorasani, K. (2022).

Stochastic fault and cyber-attack detection and

consensus control in multi-agent systems. In-

ternational Journal of Control, 95(9):2379–

2397. Publisher: Taylor & Francis eprint:

https://doi.org/10.1080/00207179.2021.1912394.

Simulative Analysis of Multi-Agent Systems in Energy Systems: Impact of Communication Networks

529

Fang, Z. H., Mo, H. D., and Wang, Y. (2017). Reliability

analysis of cyber-physical systems considering cyber-

attacks. In 2017 IEEE International Conference on

Industrial Engineering and Engineering Management

(IEEM), pages 364–368, Singapore. IEEE.

Frost., E., Heiken., J., Tr

oschel., M., and Nieße., A.

(2024). Detecting and analyzing agent communica-

tion anomalies in distributed energy system control.

pages 625–632.

Frost, E., Radtke, M., Nebel-Wenner, M., Oest, F., and

Stark, S. (2024). cosima-mango: Investigating Multi-

Agent System Robustness Through Integrated Com-

munication Simulation. SoftwareX.

Frost, E., Veith, E. M., and Fischer, L. (2020). Robust

and deterministic scheduling of power grid actors. In

2020 7th International Conference on Control, De-

cision and Information Technologies (CoDIT), vol-

ume 1, pages 100–105. IEEE.

Hasanuzzaman Shawon, M., Muyeen, S. M., Ghosh, A., Is-

lam, S. M., and Baptista, M. S. (2019). Multi-Agent

Systems in ICT Enabled Smart Grid: A Status Update

on Technology Framework and Applications. IEEE

Access, 7:97959–97973. Conference Name: IEEE

Access.

Hines, P., Veneman, J., and Tivnan, B. (2014). Smart grid:

Reliability, security, and resiliency. University of Ver-

mont.

Ilyes, N., Ben Smida, M., Khalgui, M., and Bachir, A.

(2019). Multi Agent System-based Approach for En-

hancing Cyber-Physical Security in Smart Grids.

Kou, Y., Bie, Z., Li, G., Liu, F., and Jiang, J. (2021). Relia-

bility evaluation of multi-agent integrated energy sys-

tems with fully distributed communication. Energy,

224:120123.

Mahela, O. P., Khosravy, M., Gupta, N., Khan, B., Alh-

elou, H. H., Mahla, R., Patel, N., and Siano, P. (2022).

Comprehensive overview of multi-agent systems for

controlling smart grids. CSEE Journal of Power and

Energy Systems, 8(1):115–131.

Oest, F., Radtke, M., Blank-Babazadeh, M., Holly, S., and

Lehnhoff, S. (2021). Evaluation of Communication

Infrastructures for Distributed Optimization of Virtual

Power Plant Schedules. Energies 2021, 14:1226.

Radtke, M., Stucke, C., Trauernicht, M., Montag, C., Oest,

F., Frost, E., Bremer, J., and Lehnhoff, S. (2023). In-

tegrating Agent-Based Control for Normal Operation

in Interconnected Power and Communication Systems

Simulation. Pages: 233.

Sanislav, T., Zeadally, S., Mois, G., and Fouchal, H. (2017).

Multi-agent architecture for reliable Cyber-Physical

Systems (CPS). In 2017 IEEE Symposium on Com-

puters and Communications (ISCC), pages 170–175.

Wang, P. and Govindarasu, M. (2020a). Multi-Agent

Based Attack-Resilient System Integrity Protection

for Smart Grid. IEEE Transactions on Smart Grid,

11(4):3447–3456.

Wang, P. and Govindarasu, M. (2020b). Multi-agent based

attack-resilient system integrity protection for smart

grid. IEEE Transactions on Smart Grid, 11(4):3447–

3456.

Xiahou, K. S., Liu, Y., and Wu, Q. H. (2021). Robust

Load Frequency Control of Power Systems Against

Random Time-Delay Attacks. IEEE Transactions on

Smart Grid, 12, 1:909–911.

Xu, L., Guo, Q., Wang, Z., and Sun, H. (2021). Modeling

of time-delayed distributed cyber-physical power sys-

tems for small-signal stability analysis. IEEE Trans-

actions on Smart Grid, 12(4):3425–3437.

Yang, Q., Chen, G., and Wang, T. (2020). ADMM-based

distributed algorithm for economic dispatch in power

systems with both packet drops and communication

delays. IEEE/CAA Journal of Automatica Sinica, 7,

3:842–852.

Yohanandhan, R. V., Elavarasan, R. M., Manoharan, P., and

Mihet-Popa, L. (2020). Cyber-Physical Power Sys-

tem (CPPS): A Review on Modeling, Simulation, and

Analysis With Cyber Security Applications. IEEE Ac-

cess, 8:151019–151064.

Zhang, D., Feng, G., Shi, Y., and Srinivasan, D. (2021).

Physical Safety and Cyber Security Analysis of Multi-

Agent Systems: A Survey of Recent Advances.

IEEE/CAA Journal of Automatica Sinica, 8(2):319–

333.

Zhou, T. L., Xiahou, K. S., Zhang, L. L., and Wu, Q. H.

(2021). Multi-agent-based hierarchical detection and

mitigation of cyber attacks in power systems. Interna-

tional Journal of Electrical Power & Energy Systems,

125:106516.

Zhu, Y., Li, H., Duan, H., and Guan, X. (2019). Optimal

Energy Consumption for Consensus of Multi-Agent

Systems With Communication Faults. IEEE Access,

7:138941–138949. Conference Name: IEEE Access.

ICAART 2025 - 17th International Conference on Agents and Artiﬁcial Intelligence

530