CONNECTOR: Enhancing the Traceability of Decentralized Bridge Applications via Automatic Cross-chain Transaction Association

Dan Lin, Jiajing Wu, Yuxin Su, Ziye Zheng, Yuhong Nan, Qinnan Zhang, Bowen Song, Zibin Zheng

TL;DR

CONNECTOR is a new automated tool that links transactions across different blockchains by analyzing public smart contract data, eliminating the need for centralized APIs or private ledgers. It achieves near-perfect accuracy in identifying cross-chain deposits and successfully matches over 95% of corresponding withdrawals, providing a powerful new lens for DeFi transparency and anti-money laundering efforts.

What Problem Does This Paper Address?

In the multi-chain world of Decentralized Finance (DeFi), cross-chain bridges are the essential infrastructure that allows users to move assets from one blockchain (e.g., Ethereum) to another (e.g., BNB Chain). However, these bridges create a "traceability gap." Because blockchains are isolated islands, a "deposit" transaction on the source chain and a "withdrawal" transaction on the destination chain appear as two unrelated events to standard block explorers.

Figure 1: A user making a cross-chain transaction from Ethereum to BNB chain includes three main steps: ①User locks assets to bridge contract on Ethereum (deposit trans- action txdep); ②The bridge verifies the transaction; ③The bridge contract unlocks the corresponding value of the asset to the user on the BNB chain (withdrawal transaction txwth).

Existing methods to link these transactions often rely on centralized APIs provided by the bridge developers themselves. This is problematic because it violates the decentralized ethos of blockchain—if a bridge operator is malicious or their API is down, the transactions become untraceable. The authors tackle the challenge of cross-chain transaction association using only publicly available on-chain data, such as contract traces and execution logs.

Methods

The authors propose a two-step framework called CONNECTOR:

  1. Deposit Transaction Identification (S1): The system first needs to find which transactions are actually bridge deposits. It uses two types of features:
  2. Functional Features: It applies Natural Language Processing (NLP) techniques (specifically Word2Vec) to the semantic data in transaction inputs (function names and parameters) to understand the "intent" of the transaction.
  3. Structural Features: It models the internal contract calls as a graph and uses "network motifs"—recurring sub-graph patterns—to identify the unique structural signature of a bridge deposit.
Figure 3: Overview of CONNECTOR.
  1. Withdrawal Transaction Matching (S2): Once a deposit is identified, CONNECTOR parses the contract logs to extract metadata like the intended recipient address and destination chain. It then searches the destination chain using a heuristic matcher based on three rules:
  2. Asset Type: Do the tokens match?
  3. Time Window: Did the withdrawal happen within a reasonable interval (initially 30 minutes) after the deposit?
  4. Amount: Is the withdrawal amount equal to the deposit minus a small bridge fee (typically 0-3%)?

Datasets

The researchers built the first comprehensive dataset for this problem, consisting of 24,392 cross-chain transaction pairs. They focused on three major bridges representing different technical mechanisms:

  • Celer cBridge (Hash Time Lock Contract - HTLC)
  • Multichain (Notary Mechanism)
  • Poly Network (Relay/Relay chain)

The evaluation covered transactions moving from Ethereum to the BNB Chain and Polygon, collected between September 2020 and April 2023.

Key Insights and Ideas

  • The Privacy Paradox: The study found that the vast majority of users use the exact same address on both the source and destination chains. While convenient, this makes them trivial to track and suggests a widespread lack of privacy awareness among DeFi users.
  • Official Explorers are Fallible: The authors discovered cases where a bridge's official explorer returned "No result found" for valid transactions (possibly due to contract upgrades), whereas CONNECTOR successfully associated them.
  • Structural vs. Semantic: While function names (semantics) are helpful, they change when contracts are upgraded. The structural "motifs" of how tokens move between internal contracts are more stable and generalizable across different bridge designs.
Examples of directed network motifs used to identify the structural patterns of bridge transactions.

Key Results

  • High Accuracy: CONNECTOR identified 100% of deposit transactions and matched 95.81% of withdrawal transactions across the test set.
  • Efficiency: Manual association by experts takes 5–10 minutes per transaction; CONNECTOR does it in just 0.84 seconds.
  • Superiority: It significantly outperformed existing heuristics designed for centralized bridges (CeFi), which often achieved less than 10% matching rates when applied to decentralized bridge data without API access.
  • Real-world Impact: The tool was used to successfully trace funds from major security incidents, including the Akropolis and Kucoin hacks, demonstrating its utility for anti-money laundering (AML).

Limitations and Future Work

  • Open-Source Dependency: CONNECTOR currently relies on the availability of smart contract ABIs (Application Binary Interfaces) to parse logs. While ~80% of DeFi bridges are open-source, those that aren't remain a challenge.
  • Scope Constraints: The current version does not support NFT bridges, privacy-preserving bridges (which use zero-knowledge proofs to intentionally break traceability), or non-EVM compatible chains (like Solana or Bitcoin).
  • Complex Relationships: The authors noted that "many-to-many" transaction relationships (e.g., one deposit resulting in multiple withdrawals) are currently out of scope and represent a promising direction for future research.

Summary updated

Abstract

Decentralized bridge applications are important software that connects various blockchains and facilitates cross-chain asset transfer in the decentralized finance (DeFi) ecosystem which currently operates in a multi-chain environment. Cross-chain transaction association identifies and matches unique transactions executed by bridge DApps, which is important research to enhance the traceability of cross-chain bridge DApps. However, existing methods rely entirely on unobservable internal ledgers or APIs, violating the open and decentralized properties of blockchain. In this paper, we analyze the challenges of this issue and then present CONNECTOR, an automated cross-chain transaction association analysis method based on bridge smart contracts. Specifically, CONNECTOR first identifies deposit transactions by extracting distinctive and generic features from the transaction traces of bridge contracts. With the accurate deposit transactions, CONNECTOR mines the execution logs of bridge contracts to achieve withdrawal transaction matching. We conduct real-world experiments on different types of bridges to demonstrate the effectiveness of CONNECTOR. The experiment demonstrates that CONNECTOR successfully identifies 100% deposit transactions, associates 95.81% withdrawal transactions, and surpasses methods for CeFi bridges. Based on the association results, we obtain interesting findings about cross-chain transaction behaviors in DeFi bridges and analyze the tracing abilities of CONNECTOR to assist the DeFi bridge apps.
BibTeX
@article{lin2024connector,
  author = {Dan Lin and Jiajing Wu and Yuxin Su and Ziye Zheng and Yuhong Nan and Qinnan Zhang and Bowen Song and Zibin Zheng},
  title = {{CONNECTOR: Enhancing the Traceability of Decentralized Bridge Applications via Automatic Cross-chain Transaction Association}},
  journal = {IEEE Trans. Inf. Forensics Secur.},
  year = {2024},
}
Figure