Table 6 Consortium blockchain abnormal behavior risk awareness work comparison
From: Blockchain abnormal behavior awareness methods: a survey
Awareness purpose | Technical methods | Technical details | Applicable scene | Advantage | Disadvantage | Quantitative performance | Data using |
|---|---|---|---|---|---|---|---|
Identity tracing | Group signature and ring signature | The identity of the signer is confidential but verifiable, and the administrator can open the identity of the signer (Zheng et al. 2018; Fujisaki and Suzuki 2007; Liu et al. 2004) | Consortium blockchain acquiring anonymity while can tracing identity in the abnormal situation | Satisfy identity anonymity and linkability | Long preparation time and slow speed before signing | Operations of Multiplication: 20 Operations of Exponentiation: 27 Low computational complexity | – |
Public key and certificate binding | Register the public key with a trusted third party to in- crease the credibility of the address (Ateniese et al. 2014) | Consortium blockchain with single institution of the supervision center | Fast and accurate identity tracing | Abuse of supervisory power | Script efficiency is as good as Bitcoin | – | |
Cooperative registration of public keys based on multi-party secure computing (El Defrawy and Lampkins 2014) | Consortium blockchain with multi-institutions of supervision centers | Can prevent abuse of supervisory power | Low tracking efficiency and slower speed | – | – | ||
With the change of the user’s public key, only a single registration to the supervision center (Li et al. 2021) | Consortium blockchain with single institution of the supervision center | Reduce the burden on users and the supervision center | Abuse of supervisory power | Efficiency is as good as Groth–Sahai proof system (Groth and Sahai 2008) | – | ||
Biometrics blockchain (BBC) | Using biometric information to track malicious accounts on-chain by BBC architecture (Alharthi et al. 2021) | Internet of vehicles with consortium blockchain | Use biometric information to label message senders for privacy and ensure message credibility | Latency time of chain updated | Packet Loss Rate: Less than 5% Computational Cost: 0.1–0.3 ms | Self-made chain with simulation attacks by OMNeT++ (Pongor 1993) | |
Attack on PBFT leaders | Based on reputation schemes | Use reputation model to evaluate leaders’ scores and perceive leader nodes (Lei et al. 2018) | Consortium blockchain using PBFT consensus mechanism | Identify malicious leader leader nodes in time | The effects in non-experimental environments need to be further tested | Feasibility: Delay time with in 22.0 s Reliability: Linear | Self-made chain prototype |
Forensic support based | Use cryptographic primitives such as aggregate signatures and commitments to take BFT forensic support (Sheng et al. 2021) | Consortium blockchain using BFT consensus mechanism | Forensic support can visualize | Large scale test need to be further tested | – | – | |
Auditing under privacy protection | Zero-knowledge proof and commitment | Additive homomorphism commitment to hide sensitive data and complete the audit with zero-knowledge proof guaranteeing audit reliability (Narula et al. 2018) | Consortium blockchain with audit content kept confidential | High privacy and audit reliability | Limited auditing operations | Computational Cost: Linear (Validate for 20 nodes in less than 200 ms) Auditing Time with More Nodes: Linear | Self-made chain prototype |
Collusion attack | Based on reputation schemes | Use Bayesian inference model to evaluate reputation scores and perceive collusion attacks (Yang et al. 2018) | Internet of vehicles with consortium blockchain | High feasibility in consortium ranges | Latency time of chain updated | Feasibility: Less than 1 s Reliability: Exponent | Users in vehicular and blockchain simulation platform |
Use the reputation chain to improve the performance of the transaction chain and perceive collusion attacks (Huang et al. 2020) | E-commerce environment with consortium blockchain | Sharding improves chain’s throughput | The effects in non-experimental environments need to be further tested | Feasibility: 20.4 s delay time at least Reliability: Linear | Self-made chain prototype with simulated users | ||
Use smart contracts to evaluate reputation scores and resist collusion attacks (Zhou et al. 2021) | E-commerce environment with consortium blockchain | High feasibility and reliability | The robust effect on other attacks remains to be verified | Feasibility: Less than 0.8 s Reliability: Constant | Partial Ethereum users participant in | ||
“Govern blockchains by blockchains” | Double-chain architecture | The double-chain consists of a detection chain and a data public chain. The detection chain deploys multi-feature models to detect malicious behaviors on the data public chain (Gu et al. 2018) | Consortium blockchain with multi-institutions of supervision centers | High accuracy and small scale high detection speed | Large scale testing is inefficient | Accuracy: 92.5% Recall: 94.6% F1: 93.5% | Drebin Dataset (Arp et al. 2014) |
The double-chain consists of the transaction chain and the custody chain. The custody chain uses a neural network to identify illegal transactions, and the double-chain anchors the public blockchain (Wu et al. 2020a) | Consortium blockchain with multi-institutions of supervision centers | Fast transaction speed, high scalability, and strong credibility | Practicability needs to be verified | Accuracy: 90.1% Recall: 18.5% F1: 30.8% | Elliptic Dataset |