Intelligent Health & Mission Management Architecture for Autonomous

and Resilient Distributed Space Systems

Mohammad Reza Jabbarpour

1 a

, Ghaith El-Dalahmeh

, Bao Quoc Vo

1 b

and Ryszard Kowalczyk

2,3 c

School of Science, Computing and Engineering Technologies, Swinburne University of Technology, Melbourne, Australia

University of South Australia, Adelaide, Australia

Polish Academy of Sciences, Warsaw, Poland

{rjabbarpoursattari, geldalahmeh, bvo}@swin.edu.au, Ryszard.Kowalczyk@unisa.edu.au

Keywords:

Distributed Space System, AI-Enabled Satellite Constellation, Mission Planning, Resiliency.

Abstract:

Distributed Space Systems (DSS) play a vital role in the success of multi-spacecraft missions, which are gar-

nering considerable attention because of their affordability through lower costs of multiple smaller spacecraft,

adaptability through reconﬁguration, and resilience to failure through redundancy. These systems enable col-

laborative endeavours among spacecraft, thus amplifying exploration capabilities within such missions. Never-

theless, the presence of multiple satellites ampliﬁes the system’s complexity and raises the probability of fault

occurrences. Consequently, an efﬁcient health and mission management (HMM) system capable of accurately

detecting and identifying faults within such a complex system is imperative to enhance mission success. In

this study, we introduce an innovative Intelligent Agent-based HMM (IHMM) architecture for multi-spacecraft

systems, leveraging Intelligent Agents (IAs) to seamlessly integrate mission success with satellite health and

resilience. A thorough exploration and classiﬁcation of diverse data sources suitable for integration into IAs

is conducted, categorised according to their deployment type and intended roles. To evaluate and validate our

proposed architecture, we conducted a preliminary analysis using one-time and continuous friction faults on a

reaction wheel. The experiments show our approach outperforms traditional methods by proactively adapting

control strategies in real-time and preventing saturation of other reaction wheels.

1 INTRODUCTION

Over the past decade, there has been a noticeable shift

towards utilising distributed space systems, such as

satellite constellations, as opposed to monolithic sys-

tems. This trend, bolstered by increased private in-

vestment and the launch of large constellations, is

part of the ”Space 4.0” or ”New Space” movement

(Iacomino, 2019). Educational institutions have also

become key players in space exploration, particularly

through the use of CubeSats, small satellites initially

introduced by California Polytechnic State University

in 1999. The advancement of commercial off-the-

shelf (COTS) technologies has transformed CubeSats

from educational tools to commercially viable plat-

forms. Leading examples of this shift include satellite

constellations like Starlink, OneWeb, and Lemur-2,

https://orcid.org/0000-0003-2631-1168

https://orcid.org/0000-0002-7404-110X

https://orcid.org/0000-0003-0937-4028

some of which are expected to exceed 1000 satellites,

highlighting a signiﬁcant departure from traditional

space approaches. While ”mass production” of satel-

lites is now feasible, the challenge remains in man-

aging their ”mass operations.” This shift, driven by

economic and strategic advantages of distributed sys-

tems over monolithic ones such as feasibility, redun-

dancy, resiliency, and cost-effectiveness, requires new

approaches. Traditional manual control by skilled op-

erators, even with batched telecommands, becomes

increasingly complex with larger numbers of space-

craft. This method is not scalable for large satellite

constellations, necessitating new operational methods

and greater automation (Ben-Larbi et al., 2021). Tran-

sitioning to large-scale constellations introduces sig-

niﬁcant challenges, including higher complexity of (i)

basic communication tasks and ground resources al-

location, (ii) coordination and higher probability of

anomalies, (iii) mission objectives, and (iv) space

situational awareness (SSA) functionalities (Krupke

116

Jabbarpour, M. R., El-Dalahmeh, G., Vo, B. Q. and Kowalczyk, R.

Intelligent Health & Mission Management Architecture for Autonomous and Resilient Distributed Space Systems.

DOI: 10.5220/0013094600003890

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 3, pages 116-123

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

et al., 2019). In the dynamic realm of space explo-

ration and satellite technology, ensuring the success

and safety of space missions has become increasingly

dependent on robust SSA capabilities and subsys-

tems health. Intelligent Health and Mission Manage-

ment (IHMM) represents a promising technological

approach that integrates mission success with health

of satellite’s subsystems. To motivate research ef-

forts in this area and pave the way for the automated

management of large distributed satellite systems, this

paper presents a resilience-oriented architecture for

IHMM by considering the ﬂow of pre-processed data

as an input for IAs. In summary, this paper provides

the following contributions:

• Proposes two distinct classiﬁcations of data

sources based on sensor deployment type and

their respective roles for IHMM.

• Proposes a novel resilience-oriented architecture

for IA-enabled IHMM that integrates various AI-

based technologies to increase mission success.

• Presents preliminary analysis to evaluate pro-

posed architecture effectiveness considering reac-

tion wheel friction faults as a use case scenario.

The paper is structured as follows: Section 2

reviews related concepts and approaches, focusing

on resiliency in DSSs. Section 3 classiﬁes data

sources and sensors for IHMM. Section 4 details the

resilience-oriented IHMM architecture and compo-

nents. Section 5 presents preliminary analysis of the

proposed architecture. Conclusions and future work

are in Section 6.

2 LITERATURE REVIEW

Spacecraft face numerous challenges during mis-

sions, such as harsh environments and the inability to

repair or replace malfunctioning equipment. Extreme

conditions in space, such as temperature cycling,

vacuum, and radiation, degrade critical instruments

and systems, especially in Low-Earth Orbit (LEO)

and Geosynchronous satellite operations. Compo-

nents like Telemetry, Tracking and Command (TTC)

transponders, Attitude Determination and Control

Systems (ADCS), Solar Arrays and Electrical Power

Systems (EPS) are particularly susceptible. Inte-

grated System Health Management (ISHM) repre-

sents a promising technological approach that inte-

grates sensor data from different sources with histor-

ical records of component and subsystem health sta-

tus (Figueroa et al., 2006). This fusion facilitates the

generation of actionable insights, empowering intel-

ligent decision-making concerning satellite constella-

tion operation, maintenance, and resilience. ISHM

primarily depends on evaluations and forecasts of sys-

tem health, which encompass early fault detection and

the estimation of Remaining Useful Life (Jennions,

2011). Various AI techniques have been applied to

space ISHM systems, with a notable link between

the advancement of AI technology and its integra-

tion into space health management. Many AI tech-

nologies initially developed within the aerospace in-

dustry have matured and found application in space

health management systems. In the 1970s and 1980s,

rule-based expert systems were prevalent, coinciding

with NASA’s Integrated Vehicle Health Management

(IVHM). Subsequently, the focus shifted towards ad-

vancing technologies like decision support systems

and emerging ﬁelds like Machine Learning (ML).

NASA developed the Remote Agent, the ﬁrst fully

autonomous AI system for spacecraft, featuring the

Livingstone model-based reasoning system and de-

cision support. It was used in NASA’s X-37 pro-

gram, where Vehicle Management Software (VMS)

diagnosed and predicted faults. Although the X-

37 IVHM experiment concluded, it advanced real-

time ISHM for space missions. In 1992, NASA pri-

oritized IVHM for future space systems. By the

2000s, AI development expanded to deep learning,

robotics, and autonomous vehicles, while aerospace

ISHM emphasized model-based reasoning. Systems

like MAPGEN enabled constraint-based planning for

Mars missions (Ai-Chang et al., 2004). Recently, fo-

cus has shifted to onboard autonomy in diagnostics,

prognostics, and navigation using AI

The next advancement in the ISHM paradigm,

known as Intelligent Health and Mission Manage-

ment (IHMM) (Ranasinghe et al., 2022), has emerged

to seamlessly integrate mission success with space-

craft health. IHMM systems bring forth the ability

to forecast declines in the operational performance of

subsystems, allowing for timely identiﬁcation of suit-

able corrective or reconﬁguration measures. This en-

sures that the system maintains an acceptable level of

operational capability well in advance of any poten-

tial failure event (Ranasinghe et al., 2022). Our pro-

posed resilience-oriented IHMM architecture which

integrates different AI-based technologies for HMM

is discussed in the section 4.

3 DATA CLASSIFICATION

The effectiveness of IHMM relies on diverse data

from space-based and ground-based assets, providing

vital insights into satellite health, resilience, and mis-

sion facilitation. Advanced sensors and technologies

Intelligent Health & Mission Management Architecture for Autonomous and Resilient Distributed Space Systems

117

enhance safety, efﬁciency, and security in space oper-

ations while supporting exploration and resource uti-

lization. Given the critical role of satellite health in

mission success, sensors are categorized by deploy-

ment (space or ground) and function in health man-

agement, mission success, or both (see Table 1), clar-

ifying their role in satellite operations.

3.1 Deployment-Based Classiﬁcation

Ground-based sensors are vital for satellite operations

but have limitations such as line-of-sight obstructions

and atmospheric interference. With growing mis-

sion complexity, integrating space-based sensors is

essential. These sensors provide continuous, unob-

structed observations, enabling precise monitoring,

early anomaly detection, and real-time space weather

monitoring. They enhance SSA, improve communi-

cation with GSs, and mitigate risks, making them crit-

ical for optimising mission success in space explo-

ration.

3.2 Role-Based Classiﬁcation

Satellite health is critical for space missions, requiring

effective monitoring and maintenance (Ranasinghe

et al., 2022). Beyond health, mission success involves

meeting objectives and handling space complexities,

aided by various sensors. Some focus on health, oth-

ers on mission goals or space navigation, while cer-

tain sensors serve dual roles. This section classiﬁes

sensors by their roles—health management, mission

success, or dual-purpose—helping operators optimize

decisions for mission success and satellite longevity.

4 PROPOSED ARCHITECTURE

The objective of IHMM is to develop an opera-

tional management solution using AI to monitor sys-

tem conditions and provide real-time health assess-

ments for safety-critical and mission-essential sys-

tems. It anticipates faults or performance degrada-

tion, enabling timely recovery actions and adjust-

ments to maximise mission performance. IHMM en-

hances safety and reliability through precise real-time

insights into system health and situational capabili-

ties. It considers mission objectives, operational con-

ditions, and environmental indicators for informed

decision-making. The system forecasts performance

declines, identiﬁes components needing reconﬁgura-

tion, and adjusts mission proﬁles to maintain opera-

tional capability, reducing the risk of failure or mis-

sion aborts. Main drivers of IHMM system are dis-

cussed this section. Moreover, a generic architecture

of IHMM system by considering pre-processed data

ﬂow as an input for IA is proposed at the end of the

section.

4.1 Health and Usage Monitoring

System (HUMS)

A HUMS collects, processes, and analyzes critical

system component data using onboard sensors to

monitor key systems and environmental indicators.

These sensors enable early fault detection and proac-

tive measures to prevent disruptions, as discussed in

previous sections. Sensor data is sent to a central unit

for logging and storage, processed onboard or at a

GS, and critical information is transmitted to GS op-

erators as needed. Pre-processing involves ﬁltering,

fusion, and analysis to remove noise, ensuring ac-

curate normal behavior representation (Gregory and

Liu, 2021). Data fusion integrates diverse sources,

extracting features for ML models. Fault condi-

tion indicators are monitored by analyzing anoma-

lies against a trained model, updated continuously to

adapt to operational changes, ensuring accurate diag-

nostics and prognostics. Reasoning methods include

knowledge-based, model-based, and data-driven ap-

proaches. Data-driven methods, such as ML (Neu-

ral Networks, SVMs, Clustering, Fuzzy Logic) and

statistical techniques (Gaussian Process Regression,

least squares regression, Hidden Markov Models), are

preferred for their independence from physical mod-

els, lower costs, and faster processing times.

4.2 Fault Detection, Isolation, and

Recovery (FDIR)

FDIR is vital in aerospace and spacecraft opera-

tions, monitoring sensory data or computed vari-

ables against nominal values derived from normal

operations. Monitoring parameters are identiﬁed by

analysing failure mechanisms. Signiﬁcant deviations

are tested to exclude false positives and input into an

inference module with a fault model covering poten-

tial faults. This enables real-time adjustments or re-

pairs to maintain performance until full maintenance

is feasible (Holland, 2010).

Failure Mode, Effects, and Criticality Analysis

(FMECA) is widely used to understand fault mecha-

nisms and identify critical failure paths. By analysing

system parameters (e.g., temperature, mass, voltage),

FMECA aids in pinpointing the fault’s nature and lo-

cation.

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

118

Table 1: List of Sensors and Data Sources by considering 2 Classiﬁcations.

Sensors

Deployment-based Role-based

Space Ground Health Mission

Infrared Cameras ✓ ✓ ✓

Star Trackers ✓ ✓

Solar Array Sensors ✓ ✓

Battery Sensors ✓ ✓

Thermal Infrared Sensors ✓ ✓ ✓

Gyroscopes ✓ ✓

Accelerometers ✓ ✓

Magnetometers ✓ ✓ ✓

Telemetry Data ✓ ✓

Inertial Measurement Units (IMUs) ✓ ✓

GPS/ GNSS Receivers ✓ ✓ ✓ ✓

Antenna Pointing Sensors ✓ ✓

Transponder Sensors ✓ ✓

Central Processing Unit (CPU) Load Sensors ✓ ✓

Memory Utilisation Sensors ✓ ✓

Thruster Status Sensors ✓ ✓

Fuel Level Sensors ✓ ✓

Error Logs ✓ ✓

Onboard Clock Sensors ✓ ✓

High-Resolution telescopes & Optical Cameras ✓ ✓ ✓ ✓

multi-spectral (MSP) and hyper-spectral (HSP) optical sensors ✓ ✓

Radar Sensors (e.g., Synthetic Aperture Radar (SAR)) ✓ ✓ ✓ ✓

Automatic Identiﬁcation System (AIS) Receivers/ Transponders ✓ ✓ ✓

RF Sensors ✓ ✓ ✓

Satellite Communication Systems (Laser or RF) ✓ ✓ ✓

EO/IR (Electro-Optical/Infrared) Cameras ✓ ✓ ✓

LIDAR (Light Detection and Ranging) ✓ ✓ ✓

Radiation Detectors (Dosimeters, Particle Detectors, Spectrometers) ✓ ✓ ✓ ✓

Solar Observatories ✓ ✓ ✓

4.3 Predictive Fault Analytics

Fault prediction is vital for ensuring the reliability

of complex engineering systems, including aerospace

and manufacturing. By leveraging advanced data

analytics, ML algorithms, and predictive modeling,

fault prediction anticipates failures or anomalies be-

fore they occur. This proactive strategy enables timely

maintenance, reduces downtime and costs, and en-

hances system performance and resilience. System

integrity reﬂects trust in performance based on stan-

dards, with integrity ﬂags signaling unmet criteria.

Traditional health monitoring is reactive, lacking im-

mediate prevention of critical issues or smooth degra-

dation of safety-critical systems. Therefore, fault pre-

diction is crucial for predictive reliability.

The need for methods combining reactive and pre-

dictive approaches is especially important for com-

plex autonomous missions (Thangavel et al., 2024).

Predictive integrity, derived from Avionics-Based In-

tegrity Augmentation (ABIA), was developed for

safety-critical GNSS applications.

4.4 Resilience-Oriented IHMM

Architecture

Satellite avionics oversee critical operations, includ-

ing housekeeping, task execution, resource allocation,

payload handling, and information security. With the

growing sophistication of small satellites, their soft-

ware complexity has increased, driven by a shift of

functionalities from ground-based systems to onboard

computers, enhancing autonomy and mission capabil-

ities. To address this complexity, agent-based archi-

tecture is proposed for its unique features, such as

decentralisation, modularity, robustness, scalability,

ﬂexibility, adaptability, and resiliency, among others.

The resilience-oriented IHMM architecture, in-

corporating AI-based technologies for data prepro-

cessing and input to intelligent agents (IA), is de-

picted in Fig. 1. Fault detection, the initial and

critical phase of FDIR systems, enables spacecraft to

autonomously handle issues without ground operator

support. Built-in testing (BIT), a widely used fault

detection and isolation method, identiﬁes and isolates

faults without external equipment. FDIR systems as-

sess fault severity using fault models, monitored pa-

rameters, and BIT outputs, providing inputs to IA for

further action.

The TT&C subsystem serves as the primary com-

munication link between a satellite and GS, facili-

tating control and monitoring operations. It accom-

plishes this by gathering telemetry data from all satel-

lite subsystems and transmitting it to the GS, while

also receiving commands from the ground. TT&C is

responsible for three key tasks (Anyaegbunam, 2014):

Intelligent Health & Mission Management Architecture for Autonomous and Resilient Distributed Space Systems

119

Figure 1: Resilience-oriented IHMM Architecture.

1. Health monitoring, achieved by analysing re-

ceived telemetry data via AI algorithms at the GS

to assess the satellite’s condition.

2. Satellite tracking to determine its location.

3. Satellite control by executing commands received

from the GS.

These functions are vital for satellite operations, en-

abling system health monitoring and abnormal be-

havior detection (Abdelghafar et al., 2020). Ground-

based sensors such as infrared cameras, GPS, AIS,

and LiDAR can be integrated into ground stations

(GS) to offer valuable inputs for intelligent agents

(IA). Thus, anomaly detection and command signals

from GSs can serve as IA inputs.

Telemetry data supports HUMS (space-based or

ground-based) in assessing subsystem health and

identifying failure modes (Chen et al., 2022). Envi-

ronmental factors like space weather, radiation lev-

els, and communications from neighboring satellites

can be processed directly by IA or at a GS, contribut-

ing relevant information. Payload data from mission-

related sensors is another critical input for IA, link-

ing spacecraft health to mission success (Sibilla,

2022). Consequently, pre-processed data from AI-

based units can replace raw data as IA inputs, enhanc-

ing mission capabilities and objectives.

5 PRELIMINARY ANALYSIS

In order to evaluate and validate our proposed archi-

tecture, a preliminary analysis is conducted in this

section. According to (Perumal et al., 2021) the at-

titude and orbit control system (AOCS) contributes

to 32% of small satellite failures. Reaction wheels

(RWs) are critical components for AOCS which are

used for satellite attitude control. By spinning the

reaction wheel at different speeds, the satellite can

change its orientation without using fuel. This is

based on the principle of conservation of angular mo-

mentum. RWs are prone to various failure and degra-

dation, among them friction fault is the one which is

considered in this paper. A friction fault in the reac-

tion wheels of a satellite refers to an increase in the

internal friction within the reaction wheel assembly,

which can impede the wheel’s ability to spin freely.

This friction can lead to reduced performance or com-

plete failure of the reaction wheel, compromising the

satellite’s ability to control its orientation (attitude) in

space.

Two different friction faults are considered for

evaluation purpose:

1. Deﬁnes an event that injects a large, one-time

static friction fault.

2. Deﬁnes an event that is always active, and adds a

smaller static friction fault with small probability

Basilisk (Kenneally et al., 2020), an Astrody-

namics Simulation Framework, is utilised for simu-

lation environment. The simulated scenario involves

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

120

Figure 2: One-time static friction fault.

Figure 3: Continuous smaller static friction fault.

a 6-degree-of-freedom (6-DOF) small satellite’s at-

titude control over a 60-minute period while orbit-

ing Earth in LEO, focusing on maintaining orienta-

tion during an eclipse and handling a reaction wheel

fault. The satellite uses ”Honeywell HR16” reaction

wheels controlled by a PD controller to align with the

desired attitude. The eclipse module simulates the

Earth’s shadow, temporarily affecting the sun’s vis-

ibility and the satellite’s sensors. A friction fault is

injected in one of the reaction wheels, testing the sys-

tem’s resilience. The simulation logs various param-

eters, including attitude error, reaction wheel torque,

and speed, and generates plots to illustrate the satel-

lite’s performance in maintaining orientation despite

the injected fault.

Figures 2 and 3 show one-time and continuous

static friction faults in RW

, respectively. In Figure 2,

experiences a friction fault at the 3rd minute of

the simulation, with static friction sharply increasing

and stabilizing at a higher level—indicating potential

mechanical issues like bearing wear. RW

and RW

maintain low, stable friction, showing normal opera-

tion. The early fault in RW

could impair its attitude

control efﬁciency.

Figure 3 depicts a continuous fault in RW

, with

static friction rising stepwise over 60 minutes. Fre-

quent faults, such as RW

saturation, suggest ongoing

degradation, while RW

and RW

remain unaffected.

This fault could also reduce the spacecraft’s attitude

control performance.

Each of above mentioned faults can be happened

due to the following events, respectively. For each

event, the cause & effectare discussed as follows.

1. Bearing Wear

• Cause & effect: Reaction wheels are mounted

Figure 4: Impact of one-time static friction fault on attitude.

Figure 5: Impact of continuous smaller static friction fault

on attitude.

on bearings that enable smooth spinning. How-

ever, prolonged use can cause the bearings to

wear, leading to increased friction and resis-

tance, which hinders the wheel’s ability to spin

at the desired speed.

2. RW saturation:

• Cause & effect: Reaction wheel saturation hap-

pens when wheels reach their maximum spin

rate, preventing additional torque. Persistent

saturation can strain bearings and mechanical

parts, leading to increased friction.

The effects of on-time and continuous friction

faults on satellite attitude are shown in Figures 4 and

5, respectively. These ﬁgures depict the attitude error

norm, indicating the deviation from the desired ori-

entation over simulation time. Figure 4 shows initial

ﬂuctuations due to a friction fault, while Figure 5 ex-

hibits more frequent corrections and ﬂuctuations with

continuous faults.

In the traditional architecture (without IA), at both

cases, the PD controller adjusts control torque to

minimize attitude error, while other reaction wheels

(RWs) compensate by increasing torque to stabilize

the spacecraft. This method allows for quick sta-

bilization but overlooks broader factors like overall

satellite status, mission goals, and long-term conse-

quences. Consequently, this leads to increased stress

on the remaining RWs, risking saturation and reduc-

ing the satellite’s ability to maintain precise pointing,

which can compromise the mission.

Figure 6 illustrates the RWs’ speeds over time.

The blue line, representing the affected wheel’s speed,

remains nearly constant, showing it is no longer effec-

tively contributing to attitude control due to increased

Intelligent Health & Mission Management Architecture for Autonomous and Resilient Distributed Space Systems

121

Figure 6: Reaction Wheels speed with PD controller.

friction. The other two lines display signiﬁcant speed

variations, indicating they are compensating for the

degraded wheel’s performance loss. These ﬂuctua-

tions highlight the increased effort needed to maintain

stability as the control system adjusts. Consequently,

current FDIR methods and PD controllers are mainly

reactive, offering only short-term responses.

In contrast, The IA assesses the friction fault con-

sidering the satellite’s overall status, including sub-

system health, mission phase, and environmental fac-

tors like space weather. This enables the IA to eval-

uate fault severity beyond immediate attitude errors.

By combining real-time HUMS data with historical

trends, the IA can predict if the fault will result in

a total reaction wheel failure. If so, it can take pre-

emptive measures to reduce impact. Additionally, the

IA considers the mission phase and objectives: prior-

itizing precise attitude control during critical phases

(e.g., imaging or communication alignment) or opt-

ing to conserve resources by reducing control effort

during less critical periods.

In our scenario, the IA dynamically adjusts its

control strategy in real-time to account for the reduced

effectiveness of the faulty reaction wheel (RW). To

avoid saturating the remaining RWs, it switches to

an alternative control algorithm better suited to the

situation. The IA proactively employs magnetic tor-

quers alongside the remaining RWs to distribute the

control load, reduce the control burden, and sustain

their effectiveness during the mission-critical phase.

This approach is validated in the literature, including

(Avanzini et al., 2019) and (He et al., 2023). The

dipole moment for the torque rods is shown in Fig-

ure 7. Initial spikes indicate the activation of torque

rods for attitude control and desaturation of the other

RWs. The observed reduction and stabilization of the

dipole moment over time demonstrate that the initial

intervention effectively mitigated the effects of RW

degradation, allowing other RWs to regain control.

Figure 8 shows the effectiveness of magnetic tor-

quers in controlling the speed of reaction wheels. The

rotational speeds ﬂuctuate initially, but these ﬂuctu-

ations diminish over time, particularly after 20 min-

utes. Compared to using only the PD controller (see

Figure 6), the reaction wheels’ speeds are signiﬁ-

Figure 7: Dipole moment for the torque rods.

Figure 8: Reaction Wheels speed using magnetic torquers.

cantly reduced, indicating lower stress and reduced

risk of saturation.

In summary, the IA’s proactive and predictive ca-

pabilities enhance response to friction faults in a re-

action wheel by leveraging satellite status, historical

data, and mission objectives. By integrating predic-

tive analysis and adaptive control with the traditional

PD controller, it offers a more resilient, mission-

aware fault management strategy. This approach sta-

bilizes the satellite more effectively, extends opera-

tional life, and ensures mission success. The proposed

architecture resolved these faults promptly, primarily

due to centralized decision-making and recovery by a

single IA that processes all relevant data. This cen-

tralization minimizes conﬂicts, ensures system-wide

consistency, and reduces errors common in decentral-

ized systems. Additionally, it enables faster response

times by streamlining decision-making and improv-

ing resource management through optimized alloca-

tion of processing power, memory, and communica-

tion bandwidth. Maintaining a holistic view of the

satellite’s status improves fault detection and supports

proactive interventions. Consequently, the architec-

ture enhances resiliency and signiﬁcantly boosts the

mission success rate, offering a robust solution for

complex space missions.

6 CONCLUSION AND FUTURE

WORK

In this article, we explored the factors driving the

rise of DSS in multi-spacecraft missions and empha-

sised the necessity of a resilience-oriented architec-

ture that integrates satellite health and mission suc-

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

122

cess. By classifying diverse data sources and inves-

tigating mission-based DSS literature, we proposed

a novel resilience-oriented IHMM architecture using

AI-based approaches for enhanced fault detection and

mission success. IAs in small satellites are primarily

responsible for health management and mission suc-

cess, relying on lightweight, onboard-executable al-

gorithms. Key measures to enhance DSS resilience

include onboard health management, cluster-based

mission planning, and self-recovery mechanisms. A

preliminary analysis with one-time and continuous

friction faults on the reaction wheel validated our ar-

chitecture. The system efﬁciently addressed these

faults using proactive adjustments and centralised

decision-making, preventing reaction wheel satura-

tion and maintaining stability. Future work will in-

volve implementing and validating the architecture

in a simulated constellation space environment to

demonstrate its effectiveness.

ACKNOWLEDGEMENTS

This work has been supported by the SmartSat CRC,

whose activities are funded by the Australian Govern-

ment’s CRC Program. We also acknowledge the col-

laborative contributions of the research team (James

Barr, Travis Bessell) from Saab Australia.

REFERENCES

Abdelghafar, S., Darwish, A., and Hassanien, A. E. (2020).

Intelligent health monitoring systems for space mis-

sions based on data mining techniques. Machine

learning and data mining in aerospace technology,

pages 65–78.

Ai-Chang, M., Bresina, J., Charest, L., Chase, A., Hsu, J.-J.,

Jonsson, A., Kanefsky, B., Morris, P., Rajan, K., Ygle-

sias, J., et al. (2004). Mapgen: mixed-initiative plan-

ning and scheduling for the mars exploration rover

mission. IEEE Intelligent Systems, 19(1):8–12.

Anyaegbunam, B. (2014). Design elements of satellite

telemetry, tracking and control subsystems for the pro-

posed nigerian made satellite. Int. J. Eng. Sci. Inven-

tion, 3(1):5–13.

Avanzini, G., de Angelis, E., Giulietti, F., and Serrano, N.

(2019). Attitude control of low earth orbit satellites

by reaction wheels and magnetic torquers. Acta astro-

nautica, 160:625–634.

Ben-Larbi, M. K., Pozo, K. F., Haylok, T., Choi, M., Grze-

sik, B., Haas, A., Krupke, D., Konstanski, H., Schaus,

V., Fekete, S. P., et al. (2021). Towards the automated

operations of large distributed satellite systems. part

1: Review and paradigm shifts. Advances in Space

Research, 67(11):3598–3619.

Chen, C., Yin, Y., Liu, Y., Wang, D., Chen, R., and Jia, Y.

(2022). Design and discussion of on-orbit autonomous

health management system for mission-constrained

spacecraft.

Figueroa, F., Holland, R., Schmalzwel, J., and Duncav-

age, D. (2006). Integrated system health management

(ishm): systematic capability implementation. In SAS

2006: 2006 IEEE Sensors Applications Symposium,

number SSTI-2200-0066.

Gregory, J. W. and Liu, T. (2021). Introduction to ﬂight

testing. John Wiley & Sons.

He, Y., Yang, S., Lin, S., He, W., and Yang, Z. (2023). Satel-

lite attitude control for emergency mode using reac-

tion wheels and magnetic torquers. In 2023 6th In-

ternational Conference on Robotics, Control and Au-

tomation Engineering (RCAE), pages 207–211. IEEE.

Holland, S. (2010). Integrated vehicle health management

in the automotive industry. Citeseer.

Iacomino, C. (2019). Commercial space exploration: po-

tential contributions of private actors to space explo-

ration programmes. Springer.

Jennions, I. K. (2011). Integrated vehicle health manage-

ment: perspectives on an emerging ﬁeld. SAE Inter-

national.

Kenneally, P. W., Piggott, S., and Schaub, H. (2020).

Basilisk: A ﬂexible, scalable and modular astrody-

namics simulation framework. Journal of aerospace

information systems, 17(9):496–507.

Krupke, D., Schaus, V., Haas, A., Perk, M., Dippel, J.,

Grzesik, B., Larbi, M. K. B., Stoll, E., Haylock, T.,

Konstanski, H., et al. (2019). Automated data retrieval

from large-scale distributed satellite systems. In 2019

IEEE 15th international conference on automation

science and engineering (CASE), pages 1789–1795.

IEEE.

Perumal, R. P., Voos, H., Vedova, F. D., and Moser, H.

(2021). Small satellite reliability: A decade in review.

In 35th Annual Small Satellite Conference.

Ranasinghe, K., Sabatini, R., Gardi, A., Bijjahalli, S.,

Kapoor, R., Fahey, T., and Thangavel, K. (2022). Ad-

vances in integrated system health management for

mission-essential and safety-critical aerospace appli-

cations. Progress in Aerospace Sciences, 128:100758.

Sibilla, M. (2022). Autonomous Onboard Health and Usage

Management System for Smart Satellites. PhD thesis,

Politecnico di Torino.

Thangavel, K., Sabatini, R., Gardi, A., Ranasinghe, K.,

Hilton, S., Servidia, P., and Spiller, D. (2024). Artiﬁ-

cial intelligence for trusted autonomous satellite oper-

ations. Progress in Aerospace Sciences, 144:100960.

Intelligent Health & Mission Management Architecture for Autonomous and Resilient Distributed Space Systems

123