基于深度学习的区块链蜜罐陷阱合约检测

doi:10.11959/j.issn.1000-436x.2022011

摘要/Abstract

摘要：

针对当前检测方法准确率不高以及模型泛化性较差的问题，提出了基于 KOLSTM 深度学习模型的蜜罐陷阱合约检测方法。首先，通过分析蜜罐陷阱合约的特点，提出了关键操作码的概念，并设计了可用于选取智能合约中关键操作码的关键词提取方法；其次，在传统的LSTM模型中加入关键操作码权重机制，构建了可以同时捕获蜜罐陷阱合约中隐藏的序列特征以及关键操作码特征的 KOLSTM 模型。最后，通过实验表明，该模型具有较高的识别精确率，在二分类和多分类检测场景下的F值较LightGBM模型分别提升2.39%与19.54%。

关键词: 区块链, 以太坊, 智能合约, 蜜罐陷阱合约, 深度学习

Abstract:

Aiming at the problems of low accuracy of current detection methods and poor generalization of model, a honeypot contract detection method based on KOLSTM deep learning model was proposed.Firstly, by analyzing the characteristics of honeypot contract, the concept of key opcode was proposed, and a keyword extraction method which could be used to select the key opcode in smart contract was designed.Secondly, by adding the key opcode weight mechanism to the traditional LSTM model, a KOLSTM model which could simultaneously capture the sequence features and key opcode features hidden in the honeypot contract was constructed.Finally, the experimental results show that the model had a high recognition accuracy.Compared with the existing methods, the F-score is improved by 2.39% and 19.54% respectively in the two classification and multi-classification detection scenes.

Key words: blockchain, Ethereum, smart contract, honeypot contract, deep learning

中图分类号:

TP309.2

张红霞, 王琪, 王登岳, 王奔. 基于深度学习的区块链蜜罐陷阱合约检测[J]. 通信学报, 2022, 43(1): 194-202.

Hongxia ZHANG, Qi WANG, Dengyue WANG, Ben WANG. Honeypot contract detection of blockchain based on deep learning[J]. Journal on Communications, 2022, 43(1): 194-202.

图/表 16

图1

表1

图2

图3

图4

表2

图5

表3

图6

图7

表4

图8

图9

图10

表5

表6

参考文献 24

[1]	ZHENG Z B , XIE S A , DAI H N ,et al. An overview of blockchain technology:architecture,consensus,and future trends[C]// Proceedings of 2017 IEEE International Congress on Big Data (BigData Congress). Piscataway:IEEE Press, 2017: 557-564.
[2]	ZHENG Z B , XIE S A , DAI H N ,et al. An overview on smart contracts:challenges,advances and platforms[J]. Future Generation Computer Systems, 2020,105: 475-491.
[3]	TORRES C F , STEICHEN M . The art of the scam:demystifying honeypots in Ethereum smart contracts[C]// Proceedings of 28th USENIX Security Symposium (USENIX Security 19). Berkeley:USENIX Association, 2019: 1591-1607.
[4]	CAMINO R , TORRES C F , BADEN M ,et al. A data science approach for detecting honeypots in Ethereum[C]// Proceedings of 2020 IEEE International Conference on Blockchain and Cryptocurrency. Piscata-way:IEEE Press, 2020: 1-9.
[5]	CHEN W L , GUO X F , CHEN Z G ,et al. Honeypot contract risk warning on Ethereum smart contracts[C]// Proceedings of 2020 IEEE International Conference on Joint Cloud Computing. Piscataway:IEEE Press, 2020: 1-8.
[6]	曾诗钦, 霍如, 黄韬 ,等. 区块链技术研究综述:原理、进展与应用[J]. 通信学报, 2020,41(1): 134-151.
	ZENG S Q , HUO R , HUANG T ,et al. Survey of blockchain:principle,progress and application[J]. Journal on Communications, 2020,41(1): 134-151.
[7]	CONTI M , SANDEEP KUMAR E , LAL C ,et al. A survey on security and privacy issues of bitcoin[J]. IEEE Communications Surveys ＆Tutorials, 2018,20(4): 3416-3452.
[8]	VASEK M , MOORE T . There’s no free lunch,even using bitcoin:tracking the popularity and profits of virtual currency scams[C]// International Conference on Financial Cryptography and Data Security. Berlin:Springer, 2015: 44-61.
[9]	BARTOLETTI M , CARTA S , CIMOLI T ,et al. Dissecting Ponzi schemes on Ethereum:identification,analysis,and impact[J]. Future Generation Computer Systems, 2020,102: 259-277.
[10]	CHEN W L , ZHENG Z B , NGAI E C H ,et al. Exploiting blockchain data to detect smart ponzi schemes on Ethereum[J]. IEEE Access, 2019,7: 37575-37586.
[11]	CHEN W L , ZHENG Z B , CUI J H ,et al. Detecting ponzi schemes on Ethereum:towards healthier blockchain technology[C]// Proceedings of the 2018 World Wide Web Conference on World Wide Web. New York:ACM Press, 2018: 409-1418.
[12]	张艳梅, 楼胤成 . 基于深度神经网络的庞氏骗局合约检测方法[J]. 计算机科学, 2021,48(1): 273-279.
	ZHANG Y M , LOU Y C . Deep neural network based ponzi scheme contract detection method[J]. Computer Science, 2021,48(1): 273-279.
[13]	CHEN T Q , GUESTRIN C . XGBoost:a scalable tree boosting system[C]// Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining. New York:ACM Press, 2016: 785-794.
[14]	TIKHOMIROV S , VOSKRESENSKAYA E , IVANITSKIY I ,et al. SmartCheck:static analysis of Ethereum smart contracts[C]// Proceedings of 2018 IEEE/ACM 1st International Workshop on Emerging Trends in Software Engineering for Blockchain (WETSEB). Piscataway:IEEE Press, 2018: 9-16.
[15]	LIU C , LIU H , CAO Z ,et al. ReGuard:finding reentrancy bugs in smart contracts[C]// Proceedings of 2018 IEEE/ACM 40th International Conference on Software Engineering:Companion (ICSE-Companion). Piscataway:IEEE Press, 2018: 65-68.
[16]	JIANG B , LIU Y , CHAN W K . ContractFuzzer:fuzzing smart contracts for vulnerability detection[C]// Proceedings of the 33rd ACM/IEEE International Conference on Automated Software Engineering. New York:ACM Press, 2018: 259-269.
[17]	ZHANG Y , JIN R , ZHOU Z H . Understanding bag-of-words model:a statistical framework[J]. International Journal of Machine Learning and Cybernetics, 2010,1(1/2/3/4): 43-52.
[18]	KE G , MENG Q , FINLEY T ,et al. LightGBM:a highly efficient gradient boosting decision tree[J]. Advances in Neural Information Processing Systems, 2017,30: 3146-3154.
[19]	LIU H , YANG Z Q , LIU C ,et al. EClone:detect semantic clones in Ethereum via symbolic transaction sketch[C]// Proceedings of ESEC/FSE 2018:Proceedings of the 2018 26th ACM Joint Meeting on European Software Engineering Conference and Symposium on the Foundations of Software Engineering. New York:ACM Press, 2018: 900-903.
[20]	施聪莺, 徐朝军, 杨晓江 . TFIDF 算法研究综述[J]. 计算机应用, 2009,29(S1): 167-170,180.
	SHI C Y , XU C J , YANG X J . Study of TFIDF algorithm[J]. Journal of Computer Applications, 2009,29(S1): 167-170,180.
[21]	MIKOLOV T , CHEN K , CORRADO G ,et al. Efficient estimation of word representations in vector space[J]. arXiv Preprint,arXiv:1301.3781, 2013.
[22]	WEI J , ZOU K . EDA:easy data augmentation techniques for boosting performance on text classification tasks[C]// Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing (EMNLP-IJCNLP). Boston:Association for Computational Linguistics, 2019: 6382-6388.
[23]	ARLOT S , CELISSE A . A survey of cross-validation procedures for model selection[J]. Statistics Surveys, 2010,4: 40-79.
[24]	GAL Y , GHAHRAMANI Z . A theoretically grounded application of dropout in recurrent neural networks[J]. arXiv Preprint,arXiv:1512.05287, 2015.

操作码	权重值
'SWITCH'	0.774 8
'JUMP'	0.575 1
'CALL'	0.601 9
'DUP'	0.499 9
'CALLDATASIZE'	0.507 6
'DIV'	0.501 0
'SHA'	0.627 4
'ISZERO'	0.497 7
'LT'	0.544 2
'MSTORE'	0.500 4
'STOP'	0.586 0

参数	数值
学习率	0.001
Batch大小	32
训练轮数/轮	20
词向量维度	250
隐藏层维度	150
Dropoutrate	0.5

词向量类型	精确率	召回率	F值
基于合约操作码预训练词向量	96.91%	94.30%	95.59%
维基百科词向量	94.60%	92.37%	93.47%

模型	精确率	召回率	F值
RNN	79.65%	98.62%	88.25%
LSTM	94.60%	92.37%	93.47%
CNN	95.75%	90.69%	93.15%
LightGBM	97.09%	89.62%	93.20%
KOLSTM	96.91%	94.30%	95.59%

类别		KOLSTM			LightGBM
类别	精确率	召回率	F值	精确率	召回率	F值
US	88.17%	91.98%	90.03%	100%	90.55%	95.04%
BD	92.42%	89.17%	91.04%	100%	100%	100%
HSU	88.71%	82.09%	85.27%	100%	92.57%	96.14%
HT	91.91%	100%	95.79%	100%	91.07%	95.33%
ID	73.68%	80.77%	77.06%	100%	82.39%	90.34%
SESL	94.00%	100%	96.91%	0	88.89%	0
SMC	87.00%	98.86%	92.55%	97.73%	92.21%	94.89%
TDO	98.86%	100%	99.43%	0	0	0
平均值	89.34%	92.86%	91.01%	74.71%	79.71%	71.47%