2026· IEEE Transactions on Network and Service Management· Vol 23, pp. 5815-5831· 0 citations· 46 references
Computer Science
Abstract
Effective scaling of blockchain-enabled Industrial Internet of Things (IIoT) requires sharding that simultaneously ensures transaction locality, strict committee-size feasibility, and robustness against malicious node concentration. Existing methods often fail to balance this trilemma, risking either infeasible deployments or increased shard-takeover vulnerabilities. To address this, we propose GNN-OSS, a deployable sharding framework that decouples topology-aware preference learning from hard constraint enforcement. It first employs a trust-repulsion graph neural network to learn locality-aware preferences while discouraging low-trust nodes from collapsing into the same representation region. A Post-Hoc Capacity-Constrained Projection (PH-CCP) then maps these soft preferences into strictly feasible shard assignments. Finally, an entropy-driven Overlapping Sparse Scheme (OSS) selectively replicates boundary nodes to reduce residual cross-shard overhead without altering primary consensus membership. Evaluations demonstrate that, under the evaluated settings, GNN-OSS achieves a favorable performance–security trade-off. Against 20% malicious nodes, it substantially mitigates shard-takeover risks. Furthermore, it improves throughput by up to 33% over strictly feasible baselines and lowers the cross-shard ratio from 6.4% to 4.4% with minimal per-epoch overhead. Overall, GNN-OSS provides a practical sharding framework for open or hybrid blockchain-enabled IIoT environments.
Blockchain technology enables decentralized trust, yet traditional blockchain networks face critical scalability limitations under large-scale deployments. Sharding improves throughput through parallel processing, but existing sharded BFT architectures still suffer from severe hierarchical coupling between shards and the verification committee. Moreover, the complex asynchronous competition and backoff/retransmission dynamics in sharded blockchain services remain largely unmodeled, leaving the network-level steady-state behavior of sharded blockchains poorly understood. To address these challenges, we propose D2S-BFT, a novel Decoupled Double-Star Byzantine Fault-Tolerant architecture, which physically decouples local intra-shard consensus from global verification. For rigorous performance evaluation, we establish a randomized-service double-star service system and cast the cross-shard competition mechanism as a finite-source Markov chain. We derive the state transition probability matrix under general load conditions, compute the extended sojourn time, and construct an end-to-end transaction on-chain latency equation that explicitly incorporates encryption overhead, network delay, and queuing delay. The resulting D2S queuing model, expressed in the non-classical Kendall notation L/G/n=2/inf/L-RSS, provides strict theoretical boundary constraints on system performance. It demonstrates that D2S-BFT can effectively alleviate transaction congestion and ensure robust operation, while also laying a rigorous analytical foundation for model-driven configuration optimization in large-scale dynamic blockchain environments.
Ji-Qiang Liu, Lijun Sun, Xiao Chen et al.· 2026 International Conferenc...· 0 citations
Sharding technology divides the blockchain network into multiple parallel-processing subnetworks, achieving high throughput and scalability. However, it also faces challenges, including the risk of 51% attack caused by malicious node clustering and systemic load imbalances. While reputation mechanisms are widely employed to mitigate these risks, existing approaches remain constrained by unidimensional evaluations. Specifically, most studies focus on node behavior and assess node security metrics, neglecting node performance metrics and node heterogeneity. To address these challenges, this paper presents a multi-granularity reputation model to quantify the efficiency and reliability of nodes. This model accounts for the performance and security differences arising from node heterogeneity and behavioral dynamics. Based on this model, we further propose a neighborhood-constrained simulated annealing-based node partition algorithm, NCSA-NP, that achieves balanced security and performance across shards. Experimental results demonstrate that the proposed approach achieves significant improvements in throughput and latency compared to other baselines.
Jie Li, Zhihao Guo, Yibing Li et al.· Annual International Compute...· 0 citations
Simulation results show that compared with standard PBFT, Q-PBFT, and APBFT, H-PBFT exhibits significant advantages in consensus latency, throughput, and view switching recovery time, and maintains high system robustness even in complex network environments with malicious nodes.
Zhenhua Wang, Jiangang Hu, Xinmeng Wang et al.· Future Internet· 0 citations
Elastic Proof-of-Location Byzantine Fault Tolerance is proposed, a privacy-preserving and location-aware blockchain consensus framework for IoT systems that reduces communication overhead and improves consensus efficiency compared with conventional PBFT-based approaches while strengthening resilience against location-based and identity-based attacks.
Yunus Kareem, D. Djenouri, Essam Ghadafi· Future Internet· 0 citations
Federated Learning (FL) enables collaborative model training while preserving data privacy by keeping data decentralized. However, traditional FL architectures suffer from critical trust issues, including vulnerability to model poisoning attacks, a lack of incentives for honest participation, and the inability to verify the quality of submitted model updates. Blockchain-based solutions have been proposed to address these challenges, but they often require participants to bootstrap independent validator networks and lock additional capital, limiting scalability and practical adoption. This paper proposes EigenFL, a novel FL framework secured through blockchain restaking, leveraging EigenLayer to provide cryptoeconomic security and decentralized validation by reusing Ethereum's existing validator infrastructure. Participants are economically incentivized to submit honest updates, while malicious behavior is discouraged through slashing mechanisms. Model updates are validated off-chain by independent operators, and their outcomes are immutably recorded on-chain, ensuring accountability and robustness. EigenFL demonstrates how blockchain restaking can transform FL from a trust-based paradigm into a cryptoeconomically secured framework, enabling verifiable and decentralized training.
Giovanni Maria Cristiano, Salvatore D’Antonio, Giovanni Mazzeo· Annual International Compute...· 0 citations
The rapid expansion of Internet of Things (IoT) ecosystems has intensified the need for distributed intelligence mechanisms that reduce direct raw-data exposure while remaining resilient to adversarial manipulation. Federated learning (FL) addresses part of this challenge by enabling collaborative model training without centralizing raw data, but it remains vulnerable to malicious client behavior, particularly model poisoning attacks that can substantially degrade global model quality. This paper investigates a blockchain-assisted, trust-aware FL framework for privacy-aware distributed intelligence in resource-constrained IoT systems, where the blockchain layer is used to support trust coordination, auditability, traceability, and tamper-resistant metadata recording rather than to directly improve predictive performance. The empirical study compares six aggregation strategies: FedAvg, coordinate-wise Median, Trimmed Mean, FLTrust, trust-aware weighted aggregation, and a hybrid trust-trimmed mean method. The primary evaluation is conducted on the UCI Human Activity Recognition (UCI HAR) dataset under Dirichlet-based non-IID client partitioning α = 0.1 and α = 1.0, partial client participation, and sign-flip model poisoning. Each configuration is evaluated over five independent runs. Under the severe 40% malicious-client stress test, FLTrust achieves the strongest mean robustness among the evaluated methods, reaching 0.4195 ± 0.1295 accuracy and 0.3108 ± 0.1294 macro-F1 for α = 0.1, and 0.6472 ± 0.0725 accuracy and 0.6014 ± 0.0949 macro-F1 for α = 1.0. In lower-intensity attack controls with 10% and 20% malicious clients, the trust-aware and hybrid trust-trimmed strategies are the most competitive, achieving the highest or near-highest mean performance without requiring a clean server-side reference set. Mechanism-level analysis shows that FLTrust is particularly effective in the severe setting because its reference-based scoring assigns near-zero weights to malicious clients, whereas trust-aware and hybrid methods rely on relative update consistency and are more affected by the interaction between poisoning and statistical heterogeneity. Globally, the results indicate that no single aggregation rule dominates all adversarial regimes. Instead, reference-based trust provides strong protection under severe poisoning, while trust-aware and hybrid aggregation offer competitive root-free alternatives under lower attack intensities. The findings also clarify that the privacy-preserving scope of the framework derives from FL-based data locality and does not constitute a formal cryptographic privacy guarantee.
M. Reis, Carlos Serôdio, Frederico Branco· Electronics· 0 citations