基于VAE-CWGAN和特征统计重要性融合的网络入侵检测方法

doi:10.11959/j.issn.1000-436x.2024013

Abstract

Abstract:

Considering the problems of traditional intrusion detection methods limited by the class imbalance of datasets and the poor representation of selected features, a detection method based on VAE-CWGAN and fusion of statistical importance of features was proposed.Firstly, data preprocessing was conducted to enhance data quality.Secondly, a VAE-CWGAN model was constructed to generate new samples, addressing the problem of imbalanced datasets, ensuring that the classification model no longer biased towards the majority class.Next, standard deviation, difference of median and mean were used to rank the features and fusion their statistical importance for feature selection, aiming to obtain more representative features, which made the model can better learn data information.Finally, the mixed data set after feature selection was classified through a one-dimensional convolutional neural network.Experimental results show that the proposed method demonstrates good performance advantages on three datasets, namely NSL-KDD, UNSW-NB15, and CIC-IDS-2017.The accuracy rates are 98.95%, 96.24%, and 99.92%, respectively, effectively improving the performance of intrusion detection.

Key words: intrusion detection, network traffic, class imbalance, feature selection, fusion of statistical importance

CLC Number:

TP391

Taotao LIU, Yu FU, Kun WANG, Xueyuan DUAN. Network intrusion detection method based on VAE-CWGAN and fusion of statistical importance of feature[J]. Journal on Communications, 2024, 45(2): 54-67.

Figures/Tables 21

数据集	方法	Acc	Pre	Rec	F1	FAR
NSL-KDD	ROS^[37]	82.17%	90.65%	76.55%	83.07%	9.95%
	BorderlineSMOTE^[37]	84.20%	91.69%	78.96%	84.91%	8.87%
	RENN^[20]	98.02%	98.06%	98.02%	98.03%	—
	ADASYN^[20]	98.35%	98.83%	98.35%	98.57%	—
	ENN^[20]	98.85%	98.87%	98.85%	98.85%	—
	本文方法	$98 . 95 %$	$98 . 95 %$	$98 . 95 %$	$98 . 95 %$	$1 . 01 %$
UNSW-NB15	ROS^[21]	87.31%	$99 . 04 %$	82.14%	89.80%	—
	SMOTE^[21]	87.22%	98.23%	82.70%	89.80%	—
	ADASYN^[21]	90.47%	97.44%	88.31%	92.65%	—
	WGAN^[21]	88.82%	98.90%	84.51%	91.14%	—
	CGAN^[21]	88.92%	98.64%	84.89%	91.25%	—
	本文方法	$96 . 24 %$	96.23%	$96 . 24 %$	$96 . 24 %$	$4 . 74 %$
CIC-IDS-2017	ROS^[22]	$99 . 94 %$	99.74%	99.97%	99.85%	0.07%
	SMOTE^[22]	99.93%	99.69%	$99 . 98 %$	99.84%	0.08%
	ADASYN^[22]	99.92%	99.65%	99.96%	99.80%	0.09%
	RUS+SMOTE^[22]	99.92%	99.65%	99.96%	99.80%	0.09%
	K-Means+SMOTE^[22]	99.93%	99.69%	99.97%	99.83%	0.08%
	本文方法	99.92%	$99 . 92 %$	99.92%	$99 . 92 %$	$0 . 07 %$

数据集	方法	Acc	Pre	Rec	F1	FAR
NSL-KDD	ROS^[25]	78.26%	92.34%	67.41%	77.93%	7.39%
	SMOTE^[25]	81.16%	96.42%	69.48%	80.76%	3.41%
	ADASYN^[25]	80.10%	96.16%	67.74%	79.49%	3.57%
	CVAE^[25]	85.97%	97.39%	77.43%	86.27%	2.74%
	CWGAN^[26]	90.34%	96.74%	85.92%	91.01%	3.83%
	本文方法	$98 . 91 %$	$98 . 91 %$	$98 . 91 %$	$98 . 91 %$	$1 . 07 %$
UNSW-NB15	ROS^[26]	81.70%	77.32%	94.49%	85.05%	33.96%
	SMOTE^[26]	82.44%	78.05%	$94 . 77 %$	85.60%	32.65%
	ADASYN^[26]	82.15%	77.76%	94.65%	85.38%	33.16%
	WGAN^[27]	81.49%	84.71%	82.51%	83.60%	—
	CWGAN^[27]	85.59%	86.11%	85.57%	85.84%	—
	本文方法	$87 . 58 %$	$89 . 22 %$	87.58%	$88 . 39 %$	$8 . 85 %$
CIC-IDS-2017	ROS^[22]	99.76%	99.81%	99.76%	99.77%	—
	SMOTE^[22]	99.76%	$99 . 83 %$	99.76%	99.77%	—
	ADASYN^[22]	99.76%	$99 . 83 %$	99.76%	99.77%	—
	RUS+SMOTE^[38]	99.72%	99.81%	99.72%	99.75%	—
	K-Means+SMOTE^[38]	99.70%	99.81%	99.70%	99.74%	—
	本文方法	$99 . 79 %$	99.79%	$99 . 79 %$	$99 . 79 %$	$0 . 19 %$

数据集	方法	Acc	Pre	Rec	F1	FAR	Size
NSL-KDD	卡方检验法	95.56%	95.59%	95.56%	95.44%	4.46%	13%
	随机森林法	97.84%	97.86%	97.84%	97.81%	1.62%	17%
	互信息法	96.49%	96.54%	96.49%	96.50%	3.96%	15%
	递归消除法	98.07%	98.06%	98.07%	98.06%	2.57%	20%
	遗传算法	98.24%	98.24%	98.24%	98.24%	2.54%	24%
	本文方法	$98 . 91 %$	$98 . 91 %$	$98 . 91 %$	$98 . 91 %$	$1 . 07 %$	22%
UNSW-NB15	卡方检验法	82.43%	85.20%	82.43%	82.52%	11.11%	13%
	随机森林法	84.90%	86.30%	84.90%	85.04%	11.03%	19%
	互信息法	84.18%	86.11%	84.18%	84.30%	11.97%	25%
	递归消除法	85.51%	87.35%	85.51%	85.68%	10.78%	21%
	遗传算法	85.53%	87.27%	85.53%	85.92%	11.57%	32%
	本文方法	$87 . 58 %$	$89 . 22 %$	$87 . 58 %$	$88 . 39 %$	$8 . 85 %$	21%
CIC-IDS-2017	卡方检验法	94.52%	95.25%	94.52%	94.60%	3.97%	13%
	随机森林法	99.30%	99.31%	99.30%	99.31%	0.48%	36%
	互信息法	99.31%	99.31%	99.31%	99.31%	0.56%	35%
	递归消除法	99.37%	99.37%	99.37%	99.37%	0.40%	20%
	遗传算法	99.45%	99.46%	99.45%	99.46%	0.40%	38%
	本文方法	$99 . 79 %$	$99 . 79 %$	$99 . 79 %$	$99 . 79 %$	$0 . 19 %$	41%

模型	数据集	Pre	Rec	F1
MFC	NSL-KDD	94.61%	95.06%	94.83%
	UNSW-NB15	89.03%	90.68%	89.85%
Tad-GAN	NSL-KDD	83.43%	88.71%	85.99%
	UNSW-NB15	86.81%	89.91%	88.33%
MSRC	NSL-KDD	94.43%	94.92%	94.67%
	UNSW-NB15	90.12%	91.79%	90.95%
VAE-CWGAN	NSL-KDD	$98 . 95 %$	$98 . 95 %$	$98 . 95 %$
	UNSW-NB15	$96 . 23 %$	$96 . 24 %$	$96 . 24 %$

方法	Acc	Pre	Rec	F1
KD-TCNN^[29]	98.44%	98.60%	98.47%	98.57%
CSK-CNN^[22]	$99 . 14 %$	94.03%	98.70%	96.31%
ABC-BWO-CONV-LSTM^[40]	98.67%	97.48%	$100 %$	98.73%
Bagging-GBM^[41]	91.57%	98.67%	86.18%	91.50%
本文方法	98.91%	$98 . 91 %$	98.91%	$98 . 91 %$

方法	Acc	Pre	Rec	F1
GMM-WGAN^[42]	$87 . 70 %$	88.46%	87.70%	85.44%
MCNN-DFS^[43]	80.51%	81.00%	81.00%	81.00%
CNN-BiLSTM^[44]	77.16%	82.63%	79.91%	81.25%
FCWGAN-BiLSTM^[27]	85.59%	86.11%	85.57%	85.84%
本文方法	87.58%	$89 . 22 %$	$87 . 58 %$	$88 . 39 %$

方法	Acc	Pre	Rec	F1
TACGAN^[21]	95.86%	96.85%	94.79%	95.81%
SVM-GAC^[45]	$99 . 93 %$	98.73%	99.61%	98.91%
KD-TCNN^[29]	99.44%	99.48%	99.47%	99.46%
ADASYN-RENN^[36]	99.65%	99.63%	99.65%	99.64%
本文方法	99.79%	$99 . 79 %$	$99 . 79 %$	$99 . 79 %$

References 45

[1]	HE J X , WANG X D , SONG Y F ,et al. Network intrusion detection based on conditional Wasserstein variational autoencoder with generative adversarial network and one-dimensional convolutional neural networks[J]. Applied Intelligence, 2023,53(10): 12416-12436.
[2]	王伟 . 基于深度学习的网络流量分类及异常检测方法研究[D]. 合肥:中国科学技术大学, 2018.
	WANG W . Deep learning for network traffic classification and anomaly detection[D]. Hefei:University of Science and Technology of China, 2018.
[3]	PANDA M , PATRA M R . Network intrusion detection using naive bayes[J]. International Journal of Computer Science and Network Security, 2007,7(12): 258-263.
[4]	MEHEDI H M A , NASSER M , PAL B ,et al. Support vector machine and random forest modeling for intrusion detection system (IDS)[J]. Journal of Intelligent Learning Systems and Applications, 2014,6(1): 45-52.
[5]	YAN J Q , JIN D , LEE C W ,et al. A comparative study of off-line deep learning based network intrusion detection[C]// Proceedings of the 2018 Tenth International Conference on Ubiquitous and Future Networks (ICUFN). Piscataway:IEEE Press, 2018: 299-304.
[6]	THAKKAR A , LOHIYA R . Fusion of statistical importance for feature selection in deep neural network-based intrusion detection system[J]. Information Fusion, 2023,90: 353-363.
[7]	DHANABAL L , SHANTHARAJAH S P . A study on NSL-KDD dataset for intrusion detection system based on classification algorithms[J]. International Journal of Advanced Research in Computer and Communication Engineering, 2015,4(6): 446-452.
[8]	MOUSTAFA N , SLAY J . UNSW-NB15:a comprehensive data set for network intrusion detection systems (UNSW-NB15 network data set)[C]// Proceedings of the 2015 Military Communications and Information Systems Conference (MilCIS). Piscataway:IEEE Press, 2015: 1-6.
[9]	SHARAFALDIN I , HABIBI L A , GHORBANI A A . Toward generating a new intrusion detection dataset and intrusion traffic characterization[C]// Proceedings of the 4th International Conference on Information Systems Security and Privacy. Setúbal:SciTePress, 2018: 108-116.
[10]	VIJAYANAND R , DEVARAJ D , KANNAPIRAN B . Support vector machine based intrusion detection system with reduced input features for advanced metering infrastructure of smart grid[C]// Proceedings of the 2017 4th International Conference on Advanced Computing and Communication Systems (ICACCS). Piscataway:IEEE Press, 2017: 1-7.
[11]	TONG D , QU Y R , PRASANNA V K . Accelerating decision tree based traffic classification on FPGA and multicore platforms[J]. IEEE Transactions on Parallel and Distributed Systems, 2017,28(11): 3046-3059.
[12]	FARNAAZ N , JABBAR M A . Random forest modeling for network intrusion detection system[J]. Procedia Computer Science, 2016,89: 213-217.
[13]	KOC L , MAZZUCHI T A , SARKANI S . A network intrusion detection system based on a hidden na?ve Bayes multiclass classifier[J]. Expert Systems with Applications, 2012,39(18): 13492-13500.
[14]	段雪源, 付钰, 王坤 ,等. 基于多尺度特征的网络流量异常检测方法[J]. 通信学报, 2022,43(10): 65-76.
	DUAN X Y , FU Y , WANG K ,et al. Network traffic anomaly detection method based on multi-scale characteristic[J]. Journal on Communications, 2022,43(10): 65-76.
[15]	ALDARWBI M Y , LASHKARI A H , GHORBANI A A . The sound of intrusion:a novel network intrusion detection system[J]. Computers and Electrical Engineering, 2022,104:108455.
[16]	缪祥华, 单小撤 . 基于密集连接卷积神经网络的入侵检测技术研究[J]. 电子与信息学报, 2020,42(11): 2706-2712.
	MIAO X H , SHAN X C . Research on intrusion detection technology based on densely connected convolutional neural networks[J]. Journal of Electronics ＆ Information Technology, 2020,42(11): 2706-2712.
[17]	ALTHOBAITI M M , PRADEEP M K , GUPTA D ,et al. An intelligent cognitive computing based intrusion detection for industrial cyber-physical systems[J]. Measurement, 2021,186:110145.
[18]	THAKUR S , CHAKRABORTY A , DE R ,et al. Intrusion detection in cyber-physical systems using a generic and domain specific deep autoencoder model[J]. Computers ＆ Electrical Engineering, 2021,91:107044.
[19]	DUAN X Y , FU Y , WANG K . Network traffic anomaly detection method based on multi-scale residual classifier[J]. Computer Communications, 2023,198: 206-216.
[20]	CAO B , LI C H , SONG Y F ,et al. Network intrusion detection model based on CNN and GRU[J]. Applied Sciences, 2022,12(9): 4184.
[21]	DING H W , CHEN L Y , DONG L ,et al. Imbalanced data classification:a KNN and generative adversarial networks-based hybrid approach for intrusion detection[J]. Future Generation Computer Systems, 2022,131: 240-254.
[22]	SONG J M , WANG X J , HE M S ,et al. CSK-CNN:network intrusion detection model based on two-layer convolution neural network for handling imbalanced dataset[J]. Information, 2023,14(2): 130.
[23]	KINGMA D P , WELLING M . Auto-encoding variational Bayes[J]. arXiv Preprint,arXiv:1312.6114, 2013.
[24]	GOODFELLOW I J , POUGET-ABADIE J , MIRZA M , et al . Generative adversarial nets[C]// Proceedings of the 27th International Conference on Neural Information Processing Systems. Cambridge:MIT Press, 2014: 2672-2680.
[25]	YANG Y Q , ZHENG K F , WU C H ,et al. Improving the classification effectiveness of intrusion detection by using improved conditional variational autoencoder and deep neural network[J]. Sensors, 2019,19(11): 2528.
[26]	ZHANG G L , WANG X D , LI R ,et al. Network intrusion detection based on conditional Wasserstein generative adversarial network and cost-sensitive stacked autoencoder[J]. IEEE Access, 2020,8: 190431-190447.
[27]	MA Z X , LI J , SONG Y F ,et al. Network intrusion detection method based on FCWGAN and BiLSTM[J]. Computational Intelligence and Neuroscience,2022, 2022:6591140.
[28]	EESA A S , ORMAN Z , BRIFCANI A M A . A novel feature-selection approach based on the cuttlefish optimization algorithm for intrusion detection systems[J]. Expert Systems with Applications, 2015,42(5): 2670-2679.
[29]	WANG Z D , LI Z Y , HE D J ,et al. A lightweight approach for network intrusion detection in industrial cyber-physical systems based on knowledge distillation and deep metric learning[J]. Expert Systems with Applications, 2022,206:117671.
[30]	KHAN N M , MADHAV C N , NEGI A ,et al. Analysis on improving the performance of machine learning models using feature selection technique[C]// Proceedings of 18th International Conference on Intelligent Systems Design and Applications. Berlin:Springer, 2020: 69-77.
[31]	CHANG Y P , LI W , YANG Z M . Network intrusion detection based on random forest and support vector machine[C]// Proceedings of the 2017 IEEE International Conference on Computational Science and Engineering (CSE) and IEEE International Conference on Embedded and Ubiquitous Computing (EUC). Piscataway:IEEE Press, 2017: 635-638.
[32]	KUMAR V , SINHA D , DAS A K ,et al. An integrated rule based intrusion detection system:analysis on UNSW-NB15 data set and the real time online dataset[J]. Cluster Computing, 2020,23(2): 1397-1418.
[33]	AMBUSAIDI M A , HE X J , NANDA P ,et al. Building an intrusion detection system using a filter-based feature selection algorithm[J]. IEEE Transactions on Computers, 2016,65(10): 2986-2998.
[34]	段雪源, 付钰, 王坤 . 基于 VAE-WGAN 的多维时间序列异常检测方法[J]. 通信学报, 2022,43(3): 1-13.
	DUAN X Y , FU Y , WANG K . Multi-dimensional time series anomaly detection method based on VAE-WGAN[J]. Journal on Communications, 2022,43(3): 1-13.
[35]	ARJOVSKY M , CHINTALA S , BOTTOU L . Wasserstein generative adversarial networks[C]// Proceedings of the 34th International Conference on Machine Learning. New York:ACM Press, 2017: 214-223.
[36]	GULRAJANI I , AHMED F , ARJOVSKY M ,et al. Improved training of Wasserstein GANs[J]. arXiv Preprint,arXiv:1704.00028, 2017.
[37]	尹梓诺, 马海龙, 胡涛 . 基于联合注意力机制和一维卷积神经网络双向长短期记忆网络模型的流量异常检测方法[J]. 电子与信息学报, 2023,45(10): 3719-3728.
	YIN Z N , MA H L , HU T . A traffic anomaly detection method based on the joint model of attention mechanism and one-dimensional convolutional neural network-bidirectional long short term memory[J]. Journal of Electronics ＆ Information Technology, 2023,45(10): 3719-3728.
[38]	ZHANG H P , HUANG L L , WU C Q ,et al. An effective convolutional neural network based on SMOTE and Gaussian mixture model for intrusion detection in imbalanced dataset[J]. Computer Networks, 2020,177:107315.
[39]	GEIGER A , LIU D Y , ALNEGHEIMISH S ,et al. TadGAN:time series anomaly detection using generative adversarial networks[C]// Proceedings of the 2020 IEEE International Conference on Big Data (Big Data). Piscataway:IEEE Press, 2020: 33-43.
[40]	KANNA P R , SANTHI P . Hybrid intrusion detection using MapReduce based black widow optimized convolutional long short-term memory neural networks[J]. Expert Systems with Applications, 2022,194:116545.
[41]	LOUK M H L , TAMA B A . Dual-IDS:a bagging-based gradient boosting decision tree model for network anomaly intrusion detection system[J]. Expert Systems with Applications, 2023,213:119030.
[42]	CUI J Y , ZONG L S , XIE J H ,et al. A novel multi-module integrated intrusion detection system for high-dimensional imbalanced data[J]. Applied Intelligence, 2023,53(1): 272-288.
[43]	AL-TURAIKI I , ALTWAIJRY N . A convolutional neural network for improved anomaly-based network intrusion detection[J]. Big Data, 2021,9(3): 233-252.
[44]	JIANG K Y , WANG W Y , WANG A L ,et al. Network intrusion detection combined hybrid sampling with deep hierarchical network[J]. IEEE Access, 2020,8: 32464-32476.
[45]	SHUKLA A K . Detection of anomaly intrusion utilizing self-adaptive grasshopper optimization algorithm[J]. Neural Computing and Applications, 2021,33(13): 7541-7561.

Metrics

Recommended 0

No Suggested Reading articles found!

方法	特征选择技术	数据集	检测性能
文献[28]	CFA	KDD CUP 99	DR=91.50%
文献[29]	BGWOA	NSL-KDD,	DR=98.47%
		CIC-IDS-2017	DR=99.47%
文献[30]	RF	UNSW-NB15	KNN准确率为71.01%
			DT准确率为74.22%
			BME准确率为74.64%
			XGBoost准确率为71.43
			RF准确率为74.87%
文献[31]	RF-SVM	KDD CUP 99	DR=93.00%
文献[32]	IG	UNSW-NB15	准确率为84.83%
文献[33]	MI	KDD CUP 99	DR=99.46%
		NSL-KDD	DR=98.76%

类别	训练样本数/个	测试样本数/个
Normal	67 343	9 711
DoS	45 927	7 460
Probe	11 656	2 421
R2L	995	2 885
U2R	52	67
合计	125 973	22 544

类别	训练样本数/个	测试样本数/个
Normal	56 000	37 000
DoS	12 264	4 089
Fuzzers	18 184	6 062
Analysis	2 000	677
Backdoor	1 746	583
Exploits	33 393	11 132
Generic	40 000	18 871
Reconnaissance	10 491	3 496
Shellcode	1 133	378
Worms	130	44
合计	175 341	82 332

类别	训练样本数/个	测试样本数/个
Benign	351 719	87 964
DoS Hulk	183 998	46 126
DoS GoldenEye	8 307	1 986
DoS Slowloris	4 648	1 148
DoS Slowhttptest	4 444	1 055
合计	553 116	138 279

参数	参数值
参数	VAE-CWGAN	1DCNN
轮	200	10
批大小	256	512
学习率	0.000 1	0.001
优化器	Adam	Adam
激活函数	ReLU	ReLU
卷积核大小	3×3	3×3

Network intrusion detection method based on VAE-CWGAN and fusion of statistical importance of feature

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 21

References 45

Related Articles 15

Metrics

Recommended 0

[1]	Yimin MAO, Zhan ZHOU, Zhigang CHEN. Parallel deep forest algorithm based on Spark and three-way interactive information [J]. Journal on Communications, 2023, 44(8): 228-240.
[2]	Debin WEI, Chengsheng PAN, Li YANG, Zuoren YAN. Adaptive random early detection algorithm based on network traffic level grade prediction [J]. Journal on Communications, 2023, 44(6): 154-166.
[3]	Xin SUN, Guifu ZHANG, Hongyan XING, Wang Zenghui. Research on intrusion detection for maritime meteorological sensor network based on balancing generative adversarial network [J]. Journal on Communications, 2023, 44(4): 124-136.
[4]	Shuangyan YI, Yongsheng LIANG, Jingjing LU, Wei LIU, Tao HU, Zhenyu HE. Robust feature selection method via joint low-rank reconstruction and projection reconstruction [J]. Journal on Communications, 2023, 44(3): 209-219.
[5]	Yifeng WANG, Yuanbo GUO, Qingli CHEN, Chen FANG, Renhao LIN, Yongliang ZHOU, Jiali MA. Method based on contrastive incremental learning for fine-grained malicious traffic classification [J]. Journal on Communications, 2023, 44(3): 1-11.
[6]	Junwei LIANG, Geng YANG, Maode MA, Sadiq Muhammad. Secure federated distillation GAN for CIDS in industrial CPS [J]. Journal on Communications, 2023, 44(12): 230-244.
[7]	Xueyuan DUAN, Yu FU, Kun WANG, Taotao LIU, Bin LI. Network traffic anomaly detection method based on multi-scale characteristic [J]. Journal on Communications, 2022, 43(10): 65-76.
[8]	Yifeng WANG, Yuanbo GUO, Qingli CHEN, Chen FANG, Renhao LIN. Method based on contrastive learning for fine-grained unknown malicious traffic classification [J]. Journal on Communications, 2022, 43(10): 12-25.
[9]	Qixu LIU, Junnan WANG, Jie YIN, Yanhui CHEN, Jiaxi LIU. Application of adversarial machine learning in network intrusion detection [J]. Journal on Communications, 2021, 42(11): 1-12.
[10]	Yongjin HU,Yuanbo GUO,Jun MA,Han ZHANG,Xiuqing MAO. Method to generate cyber deception traffic based on adversarial sample [J]. Journal on Communications, 2020, 41(9): 59-70.
[11]	Youliang TIAN,Yulong WU,Qiuxian LI. Optimum response scheme of intrusion detection based on information theory [J]. Journal on Communications, 2020, 41(7): 121-130.
[12]	Debin WEI,Ting SHEN,Li YANG,Yaowen QI. Network queue scheduling algorithm based on self-similar traffic level grading prediction [J]. Journal on Communications, 2020, 41(4): 182-189.
[13]	Yonghao LI, Liang HU, Ping ZHANG, Wanfu GAO. Multi-label feature selection based on dynamic graph Laplacian [J]. Journal on Communications, 2020, 41(12): 47-59.
[14]	Xinglan ZHANG,Shenglin YIN. Intrusion detection model of random attention capsule network based on variable fusion [J]. Journal on Communications, 2020, 41(11): 160-168.
[15]	Wei SUN,Peng ZHANG,Yongquan HE,Lichao XING. Attack detection method based on spatiotemporal event correlation in intranet environment [J]. Journal on Communications, 2020, 41(1): 33-41.