Sisi Zhou

As smart devices become more prevalent, consumer electronics have become mainstream, leading to widespread concerns about data privacy and security. In response to these challenges, decentralized federated learning (DFL) has emerged, allowing work nodes to collaboratively train models in a serverless, peer-to-peer manner, thereby eliminating single points of failure and enhancing the privacy protection of private data. However, the decentralized nature of DFL makes it vulnerable to poisoning attacks, where malicious work nodes can manipulate the system by launching attacks on any arbitrary work node. This security challenge is particularly pronounced in resource-constrained environments of consumer electronics, necessitating innovative security measures to ensure both efficiency and safety. To address this issue, we propose MultiSecDFL, a multi-metric filtering method for secure aggregation aimed at enhancing security within distributed and collaborative learning frameworks. In MultiSecDFL, each work node utilizes its local model as a benchmark to evaluate the models exchanged with neighboring nodes based on three dimensions: the distance, statistical characteristics, and directional sensitivity. This approach effectively excludes Byzantine work nodes from local model aggregation, ensuring both the security and efficiency of the learning process. The experimental results indicate that MultiSecDFL effectively defends against poisoning attacks across various interaction topologies, outperforming baseline methods. Overall, compared to the baselines, our method achieves an average improvement of 7.27% in model accuracy and reduces training time by an average of 24.09%. Notably, the performance is even more significant in scenarios with attacks, demonstrating a solid balance between security and efficiency in our approach.

The hierarchical federated learning (FL) framework significantly reduces communication burdens on central servers; however, the integration of edge servers introduces potential risks such as single point of failure (SPOF) and ongoing challenges like imbalanced data distribution. To tackle these challenges, we propose TrustHFL, an innovative aggregation method designed for secure hierarchical FL. TrustHFL enhances training efficiency by employing group training that clusters clients with similar data distribution characteristics. Inside each cluster, synchronous aggregation is implemented, while asynchronous aggregation is utilized between clusters to alleviate delays from bottleneck clients. We also introduce a robust access control mechanism for secure interactions between clients and edge servers, ensuring data privacy and system integrity. Moreover, our design favors off-chain computation and training, limiting on-chain storage to essential information and thereby minimizing both storage and computational demands on the blockchain, ultimately enhancing training efficiency. Extensive experimental results demonstrate that the proposed method accelerates convergence speed and enhances model accuracy. Compared to existing classical FL methods, the model accuracy is improved by an average of 1.98% under various data distribution scenarios, while the time required to achieve the same accuracy is reduced by an average of 65.54%.

Energy-aware path planning for industrial autonomous vehicles requires more than a geometric shortest path: terrain, payload, drivetrain efficiency, turning, and a hard task-level energy budget jointly affect feasibility. We present Neuro-PathNet-EC, a physics-aware constrained planner with learned multimodal guidance. The planner operates on heading-aware grid states and evaluates directed edge energy from friction and elevation. Exact Pareto labels enforce the hard energy budget, while a goal-conditioned ResNet-18 network fuses four raster channels with vehicle semantics and predicts an eight-channel dense cost-to-go map. The learned heuristic is used only for guidance; it is not assumed admissible or consistent. Exact Energy-A* obtains its guarantee from an analytic admissible anchor, and a FOCAL extension provides the bound C ≤ wC*. On a layout-disjoint synthetic industrial dataset, five training seeds achieve a test relative heuristic MAE of 0.0289 ± 0.0014 and Spearman correlation of 0.9963 ± 0.0004. Across 512 paired binding-budget queries per seed (2,560 seed-query runs), full multimodal neural-constrained search returns the exact constrained objective in every run while reducing expanded labels by 10.73% relative to Energy-A*. The semantic branch also reduces labels relative to an otherwise matched image-only model. FOCAL satisfies its declared bound in all evaluated cases, although its current correctness-first OPEN management does not improve runtime. The evaluation is based on synthetic layouts and does not constitute public-benchmark or real-vehicle validation.

The development of the Internet of Things (IoT), Big Data, and deep learning technologies has brought convenience to people’s lives. As personal privacy data protection laws and regulations tighten, the cost of acquiring high-quality annotated data from vast IoT datasets has significantly increased, resulting in prevalent issues such as data acquisition challenges and label noise in training data. In this work, we focus on the demand for privacy protection and trustworthy sharing of IoT data, and propose a method for addressing data bias in IoT through federated learning and blockchain by utilizing the theory of local intrinsic dimension (LID), incorporating committee consensus to achieve noise label identification and correction at the data level, reducing information loss in the training data. Additionally, it performs screening of low-quality local model updates at the model level, leveraging blockchain technology that addresses the single point of failure issues in traditional federated learning, ensuring the performance and security of the federated learning models. Analysis, proof of convergence, and experimentations on the proposed framework demonstrate good security and robustness in noisy environments, effectively addressing data bias in intelligent IoT settings. In scenarios with a noise level of 0.3, 0.6, and 0.9, the average model accuracy improved respectively by 7.75%, 7.30%, and 14.04% compared to FedAvg. Similarly, when compared to FedCorr, the average improvement in model accuracy is 5.19%, 3.63%, and 8.74% respectively. Moreover, the training time remains within an acceptable range for all cases.

Since the birth of Bitcoin, blockchain has shifted from a critical cryptocurrency technology to an enabling technology. Due to its immutability and trustworthiness, blockchain has revolutionized many fields requiring credibility and high-quality data for decision making. Particularly in business intelligence and business process management, users can use blockchain to build their blockchain-enabled collaboration and data-sharing ecosystem with their partners. In this paper, we present the development process of blockchain and consensus mechanisms, where important blockchain consensus mechanisms are introduced. The consensus mechanism is the kernel among various blockchain components to ensure security and performance. Again, we present a comparison of these consensus mechanisms from different perspectives. We take the blockchain-enabling business as an example and analyze the relationship between blockchain and business process characteristics and the ideas and principles for selecting consensus mechanisms. Finally, we describe the differences and connections among many consensus mechanisms while laying a foundation for selecting appropriate consensus mechanisms for different scenarios and fields of application.