Briefing
This research introduces the Proof of Data Sharing (PoDaS) algorithm, a foundational breakthrough addressing data silos and information asymmetry in traditional supply chain management. PoDaS innovatively combines the strengths of Proof of Work (PoW) and Proof of Stake (PoS) consensus mechanisms with federated learning (FL) and convolutional neural networks (CNNs). This integration utilizes the computational consumption from FL as a verifiable proof of workload, enhancing both the fairness and efficiency of the consensus process. The core implication is a future blockchain architecture capable of real-time, secure, and transparent information sharing, significantly improving operational efficiency and trust across complex, multi-party systems.
Context
Before this research, traditional supply chain information sharing struggled with inherent limitations such as data silos and significant information asymmetry. These challenges led to inefficiencies, a lack of transparency, and vulnerabilities in data integrity, hindering real-time verification and collaborative trust among diverse participants. Established blockchain consensus mechanisms, while providing immutability, often presented trade-offs between computational resource consumption (PoW) and potential centralization risks (PoS), leaving a gap for a more optimized, context-specific solution for complex data environments like supply chains.
Analysis
The paper’s core mechanism, the PoDaS algorithm, proposes a novel approach to blockchain consensus by leveraging computational work derived from federated learning (FL) processes as its proof. In this model, the effort expended in local model training and subsequent model update verification within a federated learning framework serves as the “proof of data sharing.” This inherently links the consensus process to meaningful data analysis, where convolutional neural networks (CNNs) are employed to analyze supply chain data securely. The algorithm’s design combines the robustness of PoW in ensuring fairness with the resource efficiency of PoS, effectively using FL to improve model training and CNNs for enhanced data processing. This fundamentally differs from previous approaches by integrating AI-driven data utility directly into the consensus mechanism, ensuring that computational effort contributes to both network security and practical data insights, leading to improved block generation times and superior model accuracy compared to standalone PoW or PoS systems.
Parameters
- Core Concept → Proof of Data Sharing (PoDaS) algorithm
- Key Technology Integration → Federated Learning (FL), Convolutional Neural Networks (CNNs)
- Problem Addressed → Supply Chain Data Silos, Information Asymmetry
- Performance Metrics → Block generation time, Model accuracy (96.00%)
- Comparison → Superior to PoW in block time, slightly better than PoS in block time, significantly superior to PoW/PoS in model accuracy
- Authors → Lu Cai, Aijun Liu, Yongcai Yan
- Source Publication → Sci Rep. 2025 Sep 30;15(1):33916
Outlook
This research paves the way for a new generation of blockchain applications that seamlessly integrate artificial intelligence with decentralized consensus. In the next 3-5 years, this theoretical framework could unlock real-world applications beyond supply chains, extending to other data-intensive, multi-party systems such as healthcare data sharing, collaborative research platforms, and secure industrial IoT networks. It opens new avenues for academic inquiry into hybrid consensus models that derive proof from meaningful computational work, potentially leading to more sustainable and performant decentralized architectures where network security is intrinsically linked to the utility of shared data.
