SEBF: A Single-Chain based Extension Model of Blockchain for Fintech

Yimu Ji, Weiheng Gu, Fei Chen, Xiaoying Xiao, Jing Sun, Shangdong Liu, Jing He, Yunyao Li, Kaixiang Zhang, Fen Mei, Fei WuIJCAI 2020 · DOI

Abstract

Traditional blockchain architectures often struggle with "data type rigidity," limiting their utility for complex Fintech applications that require diverse data structures beyond simple tokens. This paper presents SEBF (Single-Chain based Extension Model of Blockchain for Fintech), a novel four-layer architecture designed to extend a single blockchain's capabilities. By utilizing trusted oracles for non-standard data validation and organizing ledger data into type-specific Merkle sub-trees, SEBF achieves high scalability and faster Simplified Payment Verification (SPV). Experimental results demonstrate that increasing data diversity within the SEBF model actually improves verification efficiency, reducing SPV time by approximately 33% in a 3,000-entry dataset.

Introduction

In the evolving Fintech landscape, there is a critical need to record diverse data types—such as insurance policies, cross-border payment details, and electronic identities—on a secure, immutable ledger. However, established blockchains like Bitcoin and Ethereum are optimized for specific data structures (UTXO or account-based state trees), making the addition of new data types cumbersome. While cross-chain and sidechain solutions exist, they introduce significant architectural complexity and synchronization overhead. SEBF addresses this by enabling a single-chain model to remain extensible and efficient.

System Architecture

The researchers propose a four-layer architecture to decouple data initiation, verification, and storage:

  1. User Layer: Participants initiate service requests.
  2. Financial Terminal Layer: Comprised of light nodes (e.g., mobile devices) that perform initial signing and calculate data digests.
  3. Blockchain Layer: An alliance chain (permissioned) that maintains the ledger and processes transactions.
  4. Oracle Group Regulator Layer: A group of trusted financial authorities and "oracle machines" that use smart contracts to verify and sign extended, non-standard data types.
The four-layer architecture of the SEBF model, illustrating the interaction between users, terminals, the blockchain, and the Oracle Group.

Technical Innovations

Template Data and Oracle Integration

SEBF introduces "template data" (<T-TYPE>) to handle extensibility. While standard tokens are processed natively, new types (e.g., Insurance) are routed to the Oracle Group. This offloads the complexity of understanding diverse data formats from the core blockchain nodes; they only need to verify the Oracle's digital signature, ensuring the data adheres to the required template.

Merkle Sub-trees

A key innovation is the transition from a single global Merkle tree to multiple type-specific Merkle sub-trees. Each data type (e.g., Token, Insurance, Identity) has its own sub-tree, and the roots of these sub-trees are then hashed into the block header. This "divide and conquer" approach reduces the search space for SPV. Mathematically, searching several small trees is more efficient than searching one large tree because the time complexity shifts from $O(\log N)$ to $O(\sum \log N_j)$, where $N_j$ represents the items in a specific sub-tree.

Experimental Evaluation and Results

The model was tested using a Raspberry Pi 3B+ as a terminal and a PC cluster running the FISCO-BCOS platform for blockchain and oracle nodes. The researchers simulated four data types: standard <Token>, Insurance, CrossborderPayment, and ElectronicIdentity.

The primary metric was SPV verification speed. The results showed a clear performance advantage for the SEBF model as data diversity increased.

SPV verification time comparison showing that the SEBF model (B-TICE) outperforms the traditional single-chain model (B-T) as the dataset size grows.

In a simulation with 3,000 data entries, the traditional model (B-T) required approximately 0.42 seconds for verification. In contrast, the SEBF model with four extended types (B-TICE) reduced this to 0.28 seconds. This inverse correlation suggests that the more diverse the data types on the chain, the more efficient the verification process becomes for any individual record, as each sub-tree remains relatively shallow.

Conclusion

SEBF provides a scalable solution for Fintech applications by allowing a single blockchain to handle multiple data types without the overhead of multi-chain architectures. While the model relies on the integrity of the Oracle Group Regulator Layer and is currently optimized for permissioned alliance chains, it offers a significant leap in SPV efficiency. Future work will focus on testing the model in real-world banking scenarios to evaluate its handling of evolving business logic.

Summary updated

BibTeX
@inproceedings{ji2020sebf,
  author = {Yimu Ji and Weiheng Gu and Fei Chen and Xiaoying Xiao and Jing Sun and Shangdong Liu and Jing He and Yunyao Li and Kaixiang Zhang and Fen Mei and Fei Wu},
  title = {{SEBF: A Single-Chain based Extension Model of Blockchain for Fintech}},
  booktitle = {International Joint Conference on Artificial Intelligence},
  year = {2020},
  doi = {10.24963/ijcai.2020/620},
}
Figure