Proof of Finalization: A Self-Fulfilling Function of Blockchain
Aixian Deng,Qian Ren,Yingjun Wu,Hong Lei,Bangdao Chen
DOI: https://doi.org/10.1109/tifs.2024.3451355
IF: 7.231
2024-09-14
IEEE Transactions on Information Forensics and Security
Abstract:Blockchain has been widely used in various industries for providing trustworthy data. On-chain data can be regarded as trusted after it is finalized by blockchain consensus, namely after the data is believed to be immutable. Unfortunately, nodes with poor/isolated network conditions are still susceptible to data spoofing attacks of blockchain view, spawning kinds of severe attacks. For example, a light node newly joining a blockchain network may request the blockchain view from a malicious full node and accept a spoof view, leading to a double spending attack. Besides, a Trusted Execution Environment (TEE), the network stack of which is fully controlled by its host, may be fed spoofed blockchain data as input, undermining the trustworthiness of TEE-based computation by cheating inputs. To resist data spoofing, existing methods rely on a trusted authority to identify trusted data, or timely provide sufficient confirmation blocks for a block b to prove the finalization of b (since the adversary holding less hash power than the honest blockchain node cannot generate the confirmation blocks timely). These methods either suffer the risks caused by centralized trust base or are only PoW-oriented and high-latency. As promising blockchains including Ethereum migrate to energy-saving consensus, e.g., PoS, designing consensus-agnostic approaches against data spoofing becomes an urgent need of the industries. In this paper, we introduce a Proof of Finalization (PoF) problem for proving the finalization of blockchain to prevent data spoofing attacks of blockchain. We also contrive a novel PoF scheme, which leverages the chain quality property of blockchain to establish a trustworthy committee for proof generation. The scheme is chain-agnostic, non-interactive, non-authority-involved, and with negligible latency. Once blockchain data is finalized, the latency of proof generation in our scheme is only 106 milliseconds. Therefore, our scheme paves the way for any system, e.g., light nodes, cross-chain bridges, and layer-2 systems, to read blockchains with various consensus securely.
computer science, theory & methods,engineering, electrical & electronic