Blockchain Frontier Technology (B-Fron.
Vol.
No.
2, 2026, pp.
172O181
E-ISSN: 2808-0009 P-ISSN: 2808-0831.
DOI:10.
ye Blockchain Integration to Enhance Federated Learning Model Integrity Yane Devi Anna1 .
Sherli Triandari2* .
Sigit Anggoro3 .
Ardirra Yolandita4 .
Adele Valerry5 1 Department of Accounting.
Universitas Ekuitas.
Indonesia 2, 4 Faculty of Science and Technology.
University of Raharja.
Indonesia 3 Department of Information System.
Universitas Jenderal Achmad Yani.
Indonesia 5 Department of Science and Technology.
ILearning Incorporation.
Colombia 1 yane.
devi@ekuitas.
id, 2 sherli@raharja.
info, 3 sigit.
anggoro@lecture.
id, 4 ardirra@raharja.
info, 5 vallery.
adele@ilearning.
*Corresponding Author
Article Info
ABSTRACT
Article history:
Federated Learning is a distributed machine learning approach that enables model training without transferring raw data, thereby preserving user privacy.
improve conciseness, overlapping explanations of FLAos privacy benefits across the Abstract.
Introduction, and Literature Review have been consolidated, highlighting its importance in sensitive domains while removing redundancy.
This allows greater emphasis on the studyAos novelty, particularly the Smart Contract design featuring multi-layer verification and reputation checking mechanisms.
Despite its advantages.
FL faces significant challenges related to model integrity, including parameter manipulation, model poisoning attacks, and limited trust among participating nodes.
This study explores the integration of blockchain technology to address these issues.
Leveraging decentralization, immutability, and transparency, blockchain is used to validate model updates, record contributions, and manage node reputation.
The study employs a literature review and technical architecture design for a blockchain-integrated FL system.
The results indicate that blockchain implementation enhances the reliability and security of FL training, especially in low-trust environments, with strong relevance for healthcare, finance, and IoT applications.
Submission, 16-10-2025 Revised, 14-11-2025 Accepted, 29-12-2025 Keywords:
Blockchain Federated Learning Model Integrity Smart Contracts Transparency This is an open access article under the CC BY 4.
0 license.
DOI: https://10.
34306/bfront.
This is an open-access article under the CC-BY license .
ttps://creativecommons.
org/licenses/by/4.
AAuthors retain all copyrights INTRODUCTION The rapid advancement of Artificial Intelligence (AI) and Machine Learning (ML) technologies has led to the emergence of new approaches in data processing, one of which is Federated Learning (FL) .
, .
FL is a distributed model training method that allows data to remain on local devices, thereby preserving user privacy .
This approach is highly relevant to sensitive sectors such as healthcare, finance, and the IoT, where data protection is a top priority.
However, despite its advantages in terms of privacy.
FL still faces serious challenges regarding model integrity and trust among training nodes .
, .
One of the main issues in FL is the potential for attacks such as model poisoning and parameter manipulation by untrusted nodes .
In environments involving multiple parties both individuals and institutions, it is difficult to ensure that every participant provides valid and honest contributions during the training process .
Therefore, a mechanism is needed to enable transparent, tamper proof auditing, verification, and contribution This topic directly aligns with the Ie scope in computing and communication systems, particularly Journal homepage: https://journal.
id/b-front Blockchain Frontier Technology (B-Fron.
ye within secure and distributed machine learning architectures .
The studyAos emphasis on Blockchain integration to ensure transparency, immutability, and trust in FL contributes to ongoing Ie research directions in privacy preserving AI.
IoT, and decentralized network infrastructures .
, .
In this context, blockchain technology emerges as a potential solution to enhance the security and integrity of FL systems .
At the national level, the Government of Indonesia has formally recognized blockchain technology as part of the national digital infrastructure, supporting its adoption for secure, transparent, and accountable digital systems .
With its core characteristics decentralization, transparency, and immutability blockchain has great potential to be integrated into FL architectures to securely record every process and contribution .
The use of blockchain can create a trust layer that ensures each model parameter update is chronologically recorded and immutable while enabling reputation evaluation for all participating nodes .
, .
Hence, this study aims to explore how blockchain integration can strengthen model integrity in FL and propose a conceptual technical architecture as a solution to the existing challenges .
To enhance readability and conceptual focus, redundant descriptions of blockchainAos benefits such as transparency, immutability, and decentralization have been condensed .
The revised narrative now highlights how these characteristics directly support model integrity and trust in FL systems .
Furthermore, the terminology Autechnical architectureAy has been consistently replaced with Auconceptual architectureAy throughout the paper to accurately reflect the theoretical and design oriented nature of this research .
, .
Based on the identified challenges in ensuring model integrity and trust in FL systems, this study seeks to address the following research questions:
A Can smart contracts be utilized within blockchain networks to validate and secure model parameter updates in Federated Learning? A What are the essential architectural components required to effectively integrate blockchain as a trust and verification layer in Federated Learning environments? Ultimately, this research is expected to contribute to the development of a more secure, transparent, and accountable Federated Learning ecosystem by leveraging blockchainAos capabilities in decentralization and trust management .
The proposed conceptual architecture will not only serve as a foundational framework for designing future blockchain-based FL systems but also provide practical insights into mitigating security threats and improving collaboration efficiency among distributed participants .
Moreover, this study aims to bridge the gap between theoretical models and real-world implementation by outlining how blockchain mechanisms, such as consensus algorithms and smart contract automation, can be adapted to support scalable, privacy-preserving, and verifiable learning environments .
, .
Through this approach, it is anticipated that the integration of blockchain and Federated Learning will pave the way for more trustworthy and ethically responsible AI systems in various domains .
, .
In addition, this study aligns with the United Nations Sustainable Development Goals (SDG.
, particularly SDGs 9 (Industry.
Innovation, and Infrastructur.
SDGs 16 (Peace.
Justice, and Strong Institution.
, and SDGs 17 (Partnerships for the Goal.
The integration of blockchain technology into Federated Learning supports the development of resilient digital infrastructure and promotes innovation by enabling secure, transparent, and trustworthy collaborative AI systems.
By ensuring model integrity, accountability, and trust among distributed participants, the proposed architecture contributes to stronger institutional governance in data-driven Furthermore, the decentralized and collaborative nature of blockchain-enabled Federated Learning encourages cross-institutional partnerships while preserving data privacy, making it especially relevant for sustainable digital transformation in sensitive sectors such as healthcare, finance, and smart infrastructure.
LITERATURE REVIEW
Federated Learning: Concepts.
Advantages, and Challenges FL is a machine learning approach that enables model training to be carried out collaboratively by multiple devices or institutions without transferring data to a central server .
Thus.
FL is highly suitable for scenarios that require a high level of privacy, such as in the medical, financial, and edge computing fields .
, .
Although it offers advantages in terms of privacy.
FL still faces several major challenges, particularly related to data and model integrity .
One of the main challenges is the potential for malicious contributions from participating nodes, such as model poisoning attacks and the submission of unauthorized parameters.
Blockchain Frontier Technology (B-Fron.
,Vol 5,No 2,2026: 172-181
E-ISSN: 2808-0009 P-ISSN: 2808-0831
which can degrade the performance and validity of the global model .
, .
Therefore, mechanisms are needed to verify and maintain the integrity of each participantAos contribution in FL.
The explanation is illustrated in Figure 1.
Figure 1.
Integration of Blockchain as a Trust Layer in Federated Learning Figure 1 provides an overview of the key elements of Federated Learning, covering its concept, advantages, and challenges.
The figure highlights that FL enables multiple participants to collaboratively train a machine learning model without sharing raw data, thereby ensuring privacy.
In addition.
FL offers several advantages, particularly in sectors that require strict data protection such as healthcare, finance, and edge computing.
However, the figure also emphasizes that despite its privacy benefits.
FL still faces challenges related to data integrity and security.
Therefore, mechanisms for validating and securing model updates are essential to maintain the reliability of the global model.
Blockchain: Features and Applications in Distributed Systems Blockchain is a distributed technology that enables secure, transparent, and immutable data recording .
Features such as decentralization, distributed consensus, and cryptographic hashing make blockchain suitable for applications that require high security and auditability .
In recent years, blockchain has been applied in various sectors, including logistics, finance, voting systems, and supply chain management .
, .
In the context of Federated Learning, blockchain can be used to record and track all model update activities, maintain contribution history, and prevent data manipulation or falsification by participants .
Furthermore, smart contracts on the blockchain can be utilized to establish automatic verification mechanisms for submitted model parameters .
Current Studies and Research Gaps The integration between Blockchain and FL is a relatively new yet rapidly growing research field .
Several studies have attempted to combine these two technologies to build secure, decentralized, and trustworthy AI systems .
, .
These studies show that blockchain can provide a trust layer within FL systems, assisting in model contribution tracking, parameter validation, and participation incentives through tokenization .
, .
However, several gaps remain to be addressed, such as issues of scalability, system complexity, and communication efficiency.
There are still few studies that specifically examine the design of technical architectures for FL Blockchain systems that can operate efficiently in multi-participant environments .
Therefore, this article seeks to fill that gap by proposing a conceptual and technical approach to integrating blockchain into Federated Learning systems .
, .
From a theoretical perspective, the proposed conceptual architecture is grounded in distributed trust and consensus principles inspired by Byzantine Fault Tolerance (BFT) models, which assume the presence of potentially unreliable or malicious nodes in the network.
The blockchain layer functions as a consensus mechanism to mitigate Byzantine behaviors through immutable logging and decentralized verification .
Additionally, the incentive aspect aligns with game-theoretic concepts, where rational nodes are motivated to behave honestly to maximize their long-term reputation and reward outcomes.
This formal grounding reinforces the systemAos theoretical robustness and aligns the conceptual design with foundational principles in distributed system reliability and trust management .
Blockchain Frontier Technology (B-Fron.
Vol.
No.
2, 2026: 172Ae181 Blockchain Frontier Technology (B-Fron.
METHODOLOGY
This study employs a qualitative descriptive approach using a literature review and conceptual system design method.
The primary objective of this method is to explore, analyze, and synthesize various scholarly sources discussing the integration between FL and Blockchain, as well as to design a technical architecture framework that can enhance model integrity within Federated Learning scenarios .
, .
Type and Research Approach This research is exploratory and descriptive in nature, employing a technical conceptual approach.
The study does not focus on direct implementation through coding or simulation but rather on developing a system framework that can serve as a reference for future implementations .
, .
This study is positioned as a System Framework Paper, emphasizing the conceptual integration of blockchain and federated learning within a structured theoretical model.
Rather than presenting empirical data or simulation outcomes, it synthesizes existing literature to develop a framework that can guide future technical implementations .
The objective is to provide a conceptual foundation that identifies architectural components, trust mechanisms, and integration logic for blockchain-enabled federated learning systems .
, .
Data Sources and Collection Techniques The data for this study were obtained through a comprehensive literature review from various relevant These sources included international scientific journal articles from publishers such as Ie.
Springer, and ACM, as well as technology whitepapers discussing Blockchain and Federated Learning .
Additionally, technical documentation from open source platforms such as TensorFlow Federated.
Hyperledger, and IPFS was utilized, along with case studies and reports on similar system developments .
, .
The data collection process involved compiling and selecting relevant literature, which was then analyzed based on key themes, including the security of Federated Learning, the role of blockchain in ensuring data integrity, and the integrative architecture of FL Blockchain systems.
Data Analysis Technique The data were analyzed using a thematic analysis approach consisting of several structured steps.
First, data reduction was conducted by filtering and refining information from relevant sources to ensure accuracy, quality, and relevance.
Next, thematic categorization was carried out by grouping the extracted information according to major themes such as model security, contribution validation, and system architecture.
Finally, concept synthesis was performed to develop conceptual solutions and design an integration framework that connects FL and blockchain based on the synthesized findings from the reviewed literature.
Stage Data Reduction Thematic Catego- Concept Synthesis Table 1.
Thematic Analysis Process Description Purpose/Outcome Filtering and refining information To eliminate redundant or from relevant sources to ensure the non-relevant data and retain inclusion of only high-quality and the most credible sources for contextually relevant studies.
Grouping the extracted information To organize findings into according to major themes such as key analytical categories model security, contribution valida- that align with the research tion, and system architecture.
Developing conceptual solutions To produce a comprehensive and designing the integration conceptual framework that framework between Federated connects literature insights Learning and Blockchain based on into a unified system design.
the synthesized findings.
The Table 1 above explains the thematic analysis process used in the study, which consists of three main stages data reduction, thematic categorization, and concept synthesis.
In the data reduction stage, information from various sources is filtered to include only high quality and relevant studies, ensuring that the Blockchain Frontier Technology (B-Fron.
,Vol 5,No 2,2026: 172-181
E-ISSN: 2808-0009 P-ISSN: 2808-0831
analysis is based on credible data.
The thematic categorization stage involves organizing the extracted information into key themes such as model security, contribution validation, and system architecture to align the findings with the main research focus.
Finally, in the concept synthesis stage, the researcher integrates these themes to develop a comprehensive conceptual framework that illustrates how FL and Blockchain can be effectively combined into a unified system design.
RESULT AND DISCUSSION
Integrative Architecture Design of Blockchain Federated Learning The study proposes a conceptual architecture that integrates Blockchain technology into FL to enhance data integrity, transparency, and trust during the model training process.
The architecture focuses on the logical interaction between FL nodes and the blockchain network rather than on full technical implementation.
Each model update is hashed using cryptographic functions and verified by smart contracts, which check contributor metadata such as ID, timestamp, and accuracy threshold before aggregation.
A key innovation of this design lies in its multi-layer verification logic within smart contracts.
These contracts not only record hashes but also validate model parameters, assess node reliability through a reputation ledger, and enforce differential privacy constraints to prevent malicious updates.
This structured trust and privacy aware validation system differentiates the model from previous Blockchain FL integrations.
To overcome scalability and latency challenges, the architecture adopts an off chain data management approach, storing large model parameters in distributed systems like IPFS, while keeping only hashes and validation metadata on the blockchain.
This approach maintains transparency and verifiability while reducing network overhead.
Overall, the proposed architecture serves as a theoretical framework that ensures data security, privacy preservation, and auditability within decentralized AI environments, aligning with IeAos principles of efficient and trustworthy system design.
Figure 2.
Integrative Architecture of Blockchain-Federated Learning The diagram in Figure 2 illustrates the integration of Blockchain into FL, where each node trains a local model and submits the model parameters for verification via a smart contract, recording them as a hash on the blockchain.
This process ensures data integrity, transparency, and security by preventing unauthorized changes, malicious contributions, and model poisoning, while also maintaining the privacy of local data.
leveraging blockchain.
Federated Learning becomes more secure, reliable, and suitable for applications in sensitive sectors like healthcare, finance, and edge computing, offering a robust solution for decentralized machine learning.
Functional Advantages of the Resulting System The integration of Blockchain and FL in the proposed architecture provides several significant advantages.
First, the integrity of model parameters is preserved, as every contribution is recorded on the blockchain Blockchain Frontier Technology (B-Fron.
Vol.
No.
2, 2026: 172Ae181 Blockchain Frontier Technology (B-Fron.
ye and cannot be tampered with.
Second, the system becomes more transparent, since the entire history of model updates can be traced by all participants through the blockchain ledger.
Third, the implementation of smart contracts enables parameter validation processes to be conducted automatically and consistently, without manual Fourth, this approach opens the opportunity to implement reputation based incentive mechanisms, where nodes that consistently provide valid contributions can be rewarded, while suspicious nodes may have their contributions restricted.
Table 2.
Advantages of Blockchain in Model Management Description Each model update is securely recorded on the blockchain, preventing tampering or data manipulation.
Transparency All participants can trace the full history of model updates through the blockchain ledger.
Automated Validation Smart contracts automatically validate parameters, ensuring consistency without manual intervention.
Reputation-Based Incentives Nodes providing valid contributions can earn rewards, while unreliable nodes face contribution limits.
Advantage Model Integrity Table 2 highlights the main advantages of integrating Blockchain with Federated Learning.
Blockchain ensures model integrity by preventing data tampering, promotes transparency through traceable updates, enables automated validation via smart contracts, and supports reputation-based incentives that reward reliable participants while limiting untrustworthy ones.
Overall, these features enhance the systemAos security, trust, and Analysis of Challenges and Limitations Although the proposed architecture offers solutions to the issue of model integrity in FL, several challenges must be considered.
One challenge is the increased system complexity due to the addition of the blockchain layer, affecting both inter node communication and transaction management.
Furthermore, public blockchains such as Ethereum have limitations in terms of throughput and transaction costs, making private or permissioned blockchains like Hyperledger potentially more suitable for this system.
Another challenge is latency during the verification and aggregation processes, particularly at large scales with a high number of Therefore, real world implementation of this system requires adjustments and optimizations based on the specific needs of the sector or use case.
An important trade-off arises when adopting a private or permissioned blockchain to mitigate latency and throughput limitations.
While such frameworks enhance performance and scalability, they may partially reduce decentralization by introducing controlled access and centralized governance over node participation.
This shift transitions the trust model from a fully trustless consensus to a semi-trusted environment relying on preauthorized nodes for validation.
Nevertheless, transparency and auditability can still be preserved through appropriate cryptographic proof mechanisms and distributed access policies.
Recognizing this balance between efficiency and decentralization is essential for applying the proposed architecture in real-world multiinstitutional Federated Learning scenarios.
It should be noted that these observations are derived from a synthesis of existing scholarly works rather than empirical testing.
Consequently, the identified challenges including latency, transaction costs, and communication complexity should be interpreted as theoretical insights guiding future experimental research.
Overall, the studyAos findings indicate that integrating blockchain into FL provides a potential solution to address FLAos key weaknesses, namely the lack of contribution verification mechanisms and vulnerability to model manipulation.
The proposed system ensures integrity, transparency, and trust crucial aspects in the development of secure and responsible collaborative AI systems.
However, the full effectiveness of this system can only be demonstrated through real world implementation and testing across various contexts.
MANAGERIAL IMPLICATIONS
The integration of blockchain into Federated Learning provides strategic value for organizations that manage sensitive, distributed, or cross-institutional data.
Managers in sectors such as healthcare, finance.
Blockchain Frontier Technology (B-Fron.
,Vol 5,No 2,2026: 172-181
E-ISSN: 2808-0009 P-ISSN: 2808-0831
education, and IoT can utilize the proposed architecture to establish a verifiable audit trail that strengthens compliance, transparency, and accountability.
The use of smart contracts to automate parameter validation reduces operational risks associated with human error, minimizes opportunities for data manipulation, and ensures that only high-quality model updates are accepted in the learning process.
This contributes to more reliable decision-making within organizations by ensuring that AI models are trained using trustworthy and validated contributions.
From an operational and governance perspective, the systemAos reputation-based mechanism enables managers to incentivize honest participation while restricting unreliable nodes, creating a more stable and performance-oriented collaborative ecosystem.
The use of off-chain storage and modular architectural components also allows organizations to optimize resource allocation and adapt the system to their technical capacity.
However, implementing this integration requires managers to carefully balance the trade-off between decentralization and efficiency, especially when choosing between public and permissioned blockchain platforms.
These considerations underscore the importance of managerial decision-making in ensuring the sustainable adoption, governance, and scalability of blockchain-enabled Federated Learning systems.
CONCLUSION
This research successfully designed an integrative architecture that combines blockchain technology with FL to enhance the integrity of the training model.
By utilizing blockchain as a transparent and immutable trust layer, the system can securely record and verify each participant nodeAos contribution.
The proposed architecture conceptually demonstrates the potential to enhance integrity, transparency, and trust within Federated Learning systems through blockchain integration.
While the design provides a strong theoretical foundation, these benefits remain at a conceptual stage and require empirical validation through real world implementation or simulation studies.
The integration, therefore, creates a reliable and transparent framework for auditability rather than a fully verified technical solution.
The use of smart contracts strengthens the automatic validation of model parameters, preventing manipulation and preserving the authenticity of the data used in the distributed learning process.
The main advantage of this integration lies in creating a learning system that not only protects user data privacy but also provides a clear and reliable audit mechanism.
The transparency enabled by blockchain allows for complete traceability of the model training history while also opening opportunities for implementing a reputation based incentive system to improve the quality of contributions.
This represents an important step in addressing the main challenges of FL, particularly regarding the security and reliability of the resulting model.
However, the real world implementation of this system faces several challenges, such as increased system complexity, verification latency, and performance limitations in public blockchains.
Therefore, future research is recommended to focus on developing a prototype with technical optimizations, including selecting the appropriate blockchain platform and testing at a large scale.
In doing so, the potential of the proposed blockchain-integrated Federated Learning architecture for scalable, efficient, and trustworthy real-world deployment can be empirically evaluated.
DECLARATIONS
About Authors Yane Devi Anna (YD) https://orcid.
org/0009-0001-1909-5105 Sherli Triandari (ST) https://orcid.
org/0009-0007-9323-4373 Sigit Anggoro (SA) https://orcid.
org/0009-0008-3161-86352 Ardirra Yolandita (AY) Adele Valerry (AV) https://orcid.
org/0009-0001-9358-3588 https://orcid.
org/0009-0009-1433-1058 Author Contributions Conceptualization: YD.
ST, and SA.
Methodology: AY.
Software: AV.
Validation: AV and YD.
Formal Analysis: SA and ST.
Investigation: YD.
Resources: AY.
Data Curation: AY.
Writing Original Draft Blockchain Frontier Technology (B-Fron.
Vol.
No.
2, 2026: 172Ae181 Blockchain Frontier Technology (B-Fron.
ye Preparation: ST and YD.
Writing Review and Editing: SA.
Visualization: AV.
All authors.
YD.
ST.
SA.
AY and AV, have read and agreed to the published version of the manuscript.
Data Availability Statement The data presented in this study are available on request from the corresponding author.
Funding The authors received no financial support for the research, authorship, and/or publication of this article.
Declaration of Conflicting Interest The authors declare that they have no conflicts of interest, known competing financial interests, or personal relationships that could have influenced the work reported in this paper.
REFERENCES