基于优化的堆叠集成区块链DDoS攻击检测方法  

Optimized Stacking Ensemble Learning Blockchain DDoS Attack Detection Method

作　　者：王春东 郑泽霖 WANG Chun-dong;ZHENG Ze-lin(School of Computer Science and Engineering,Tianjin University of Technology,Tianjin 300384,China;Tianjin Public Security Police Profession College,Tianjin 300382,China;National Engineering Laboratory for Computer Virus Prevention and Control Technology,Tianjin 300384,China)

机构地区：[1]天津理工大学计算机科学与工程学院,天津300384 [2]天津公安警官职业学院,天津300382 [3]计算机病毒防治技术国家工程实验室,天津300384

出　　处：《计算机技术与发展》2025年第2期54-62,共9页Computer Technology and Development

基　　金：国家自然科学基金联合基金项目(U1536122);国家重点研发计划“区块链”重点专项(2023YFB2703900)。

摘　　要：分布式拒绝服务(Distribute Denial of Service,DDoS)攻击是常见的网络攻击手段之一,对于影响力日益增长的区块链网络构成了较大的威胁。包含堆叠法(Stacking)在内的集成学习模型在DDoS攻击检测方面有很大前景,而Stacking在面对不同类型数据集时需要调整学习器组合。该文使用Stacking方法检测区块链DDoS攻击,利用贝叶斯优化确定各学习器超参数,同时还使用算术优化算法(Arithmetic Optimization Algorithm,AOA)选择基学习器的组合,来解决需要手动调节学习器的问题。在区块链网络攻击流量数据集和比特币交易所交易数据上分别进行了实验,通过准确率、攻击数据漏报率和宏平均精准率三种评价指标进行对比,该方法在这两种不同类型数据集上的性能均优于其他三种常见的集成学习算法。还通过改变实验数据集大小探究出攻击检测性能会随着数据集的增大而上升。通过实验可以证明该方法可以有效检测不同类型数据集上的区块链DDoS攻击。Distribute Denial of Service(DDoS)is one of the common network attack methods,which poses a significant threat to the increasingly influential blockchain network.Ensemble learning models,including the Stacking method,have great potential in DDoS attack detection,and Stacking requires adjusting the combination of learners when facing different types of datasets.We use the Stacking method to detect blockchain DDoS attacks,use Bayesian optimization to determine the hyperparameters of each learner,and also use the Arithmetic Optimization Algorithm(AOA)to select a combination of base learners to solve the problem of manually adjusting learners.The experiments are conducted on blockchain network attack traffic datasets and Bitcoin exchange trading data,and the comparison is carried out by three evaluation indicators:accuracy,attack data omission rate,and macro average accuracy.The proposed method is superior to the other three common ensemble learning algorithms in terms of performance on these two different types of datasets.We also explore how the attack detection performance increases with the increase of the experimental dataset size.Through experiments,it can be proven that the proposed method can effectively detect blockchain DDoS attacks on different types of datasets.

关 键 词：网络空间安全 区块链 DDOS攻击检测 集成学习 堆叠 算术优化算法 

分 类 号：TP309[自动化与计算机技术—计算机系统结构]