基于深度学习的无线传感器网络流量异常检测

doi:10.13229/j.cnki.jdxbgxb.20250045

摘要/Abstract

摘要：

针对现有无线传感器网络（WSNs）流量异常检测方法泛化能力较弱、异常检测准确率不高等缺陷，提出一种融合改进卷积神经网络（CNN）和密度峰值聚类（Density peak clustering，DPC）的WSNs流量异常检测模型。该模型由数据处理层、特征提取层和异常判定输出层组成。数据处理层，利用格拉姆角场（GAF）变换技术，将一维WSNs流量时间序列转换生成为二维图像并传输至特征提取层。特征提取层，运用CNN提取WSNs数据高维空间特征；设计改进的黏菌优化算法和引入最小绝对收缩和选择算子（Lasso）回归对CNN进行改进优化，以增强CNN网络泛化能力。异常判定输出层，采用DPC对提取的高维空间特征进行聚类，通过定义异常判定规则最终实现对WSNs流量的异常检测。实验结果表明：相比于其他WSNs流量异常检测方案，所提模型检测准确率提高了约15.33%。

关键词: 计算机科学与技术, 无线传感器网络, 异常检测, 卷积神经网络, 黏菌优化算法, 密度峰值聚类

Abstract:

To address weak generalization and low accuracy in existing detection methods， a WSNs traffic anomaly detection model based on an improved convolutional neural network （CNN） and density peak clustering （DPC） is proposed. This model consists of a data processing layer， a feature extraction layer， and an anomaly detection output layer. In the data processing Layer， the gramian angular fields （GAF） is used to transform one-dimensional WSNs traffic time series into two-dimensional images. In the feature extraction layer， CNN is used to extract high-dimensional spatial features of WSNs data. An improved slime mold optimization algorithm is designed and the least absolute shrinkage and selection operator （Lasso） is introduced to enhance the generalization ability of CNN network. In the traffic anomaly detection output layer， DPC is used to cluster the extracted high-dimensional spatial features， and anomaly detection of WSNs data is ultimately achieved by defining anomaly detection rules. The experimental results show that compared to other WSNs traffic anomaly detection schemes， the detection accuracy of the proposed detection model has increased by 15.33%.

Key words: computer science and technology, wireless sensor networks, anomaly detection, convolutional neural network, slime mold optimization algorithm, density peak clustering

中图分类号:

TP393

刘洲洲,金聪,蒋光毅,贾楠,刘超,张杨梅. 基于深度学习的无线传感器网络流量异常检测[J]. 吉林大学学报(工学版), 2026, 56(8): 2201-2209.

Zhou-zhou LIU,Cong JIN,Guang-yi JIANG,Nan JIA,Chao LIU,Yang-mei ZHANG. Traffic anomaly detection in wireless sensor networks based on deep learning[J]. Journal of Jilin University(Engineering and Technology Edition), 2026, 56(8): 2201-2209.

图/表 8

图1

图2

表1

表2

ISMA优化CNN超参数配置结果"

数据集	超参数
数据集	$m 1$	$m 2$	$m 3$	$m 4$	$Δ$
WSN-DS	14	51	124	30	39
KDD CUP'99	22	67	144	43	51

表2

图3

图4

表3

不同模型评价指标对比结果 (%)"

模型	评级指标	WSN-DS					KDD CUP'99
模型	评级指标	N1	B1	G1	F1	S1	N2	D2	R2	U2	P2
ICNN-DPC	$A C$	99.4	98.7	97.4	89.3	98.8	98.9	97.4	96.2	97.3	97.9
	$F A$	0.03	0.05	0.04	0.16	0.04	0.05	0.04	0.06	0.03	0.04
	$R$	/	100	98.9	87.2	100	/	98.2	100	99.6	100
	$Λ 1$	97.1	97.6	96.4	86.6	98.3	98.1	96.6	96.0	96.9	97.3
ICNN-Softmax	$A C$	90.2	89.3	89.1	87.2	90.6	89.5	90.4	88.7	90.1	91.4
	$F A$	0.12	0.14	0.11	0.19	0.09	0.11	0.08	0.15	0.12	0.10
	$R$	/	90.2	91.1	86.2	91.1	/	93.2	92.7	91.0	93.3
	$Λ 1$	90.4	90.7	91.6	80.3	90.7	91，4	92.2	91.9	90.5	92.4
CNN-Softmax	$A C$	87.1	86.2	88.1	77.1	88.3	87.2	84.4	82.7	85.5	84.1
	$F A$	0.22	0.21	0.16	0.27	0.15	0.19	0.16	0.22	0.16	0.15
	$R$	/	85.4	88.1	79.3	89.8	/	85.2	84.1	86.4	92.3
	$Λ 1$	80.2	81.1	83.2	76.4	85.3	84.1	82.2	83.3	84.6	90.5
GCN-GRU	$A C$	98.6	97.2	96.9	93.2	97.7	98.3	96.1	95.8	96.4	97.0
	$F A$	0.04	0.06	0.05	0.07	0.06	0.06	0.05	0.08	0.07	0.06
	$R$	/	98.9	97.2	94.1	100	/	97.9	99.4	99.1	99.6
	$Λ 1$	96.3	95.2	95.5	93.1	97.4	98.2	95.4	95.3	96.2	96.8
BIRCH	$A C$	88.6	87.0	89.4	79.2	89.1	88.4	85.2	83.1	86.0	85.4
	$F A$	0.19	0.18	0.14	0.22	0.13	0.16	0.15	0.19	0.14	0.13
	$R$	/	88.3	90.4	80.7	91.5	/	86.6	85.4	87.6	86.2
	$Λ 1$	88.3	87.4	89.5	78.2	90.4	88.7	86.4	82.3	86.2	84.1

表3

表4

WSNs异常检测对别结果"

模型	指标/%
模型	$A C$	$F A$	$R$	$Λ 1$
ICNN-DPC	98.21	0.07	99.27	98.38
ICNN-Softmax	90.34	0.11	92.36	95.29
CNN-Softmax	85.15	0.22	89.42	88.41
GCN-GRU	97.13	0.09	98.96	97.23
BIRCH	89.24	0.18	93.45	91.62

表4

参考文献 20

[1]	刘景美, 闫义博. 人工鱼群特征选择的网络入侵检测系统[J]. 西安电子科技大学学报, 2023, 50(4): 132-138.
	Liu Jing-mei, Yan Yi-bo. Artificial fish feature selection network intrusion detection system [J]. Journal of Xidian University, 2023, 50(4): 132-138.
[2]	金明, 丁蓉. 一种联合时域和空域残差的网络异常检测与节点定位方法[J]. 电子学报, 2023, 51(5): 1172-1178.
	Jin Ming, Ding Rong. Detection and localization of outlier nodes in wireless sensor networks via jointing temporal and spatial residuals [J]. Acta Electronica Sinica, 2023, 51(5): 1172-1178.
[3]	朱梦圆, 陈卓, 刘鹏飞, 等. 基于雾计算的无线传感器网络联合入侵检测算法[J]. 北京航空航天大学学报,2022, 48(10): 1943-1950.
	Zhu Meng-yuan, Chen Zhuo, Liu Peng-fei, et al. Fog computing-based federated intrusion detection algorithm for wireless sensor networks [J]. Journal of Beijing University of Aeronautics and Astronautics, 2022, 48(10): 1943-1950.
[4]	陈俊杰, 邓洪高, 马谋, 等. GCN-GRU: 一种无线传感器网络故障检测模型[J]. 西安电子科技大学学报, 2022, 49(5): 60-67.
	Chen Jun-jie, Deng Hong-gao, Ma Mou, et al. GRN-GRU: a fault detection model for wireless sensor networks[J]. Journal of Xidian University, 2022, 49(5): 60-67.
[5]	特木尔朝鲁,张亚萍. 基于卷积神经网络的无线传感器网络链路异常检测算法[J]. 吉林大学学报: 工学版, 2024, 54(8): 2295-2300.
	Chao-lu Temur, Zhang Ya-ping. Link anomaly detection algorithm for wireless sensor networks based on convolutional neural networks[J]. Journal of Jilin University (Engineering and Technology Edition), 2024, 54(8): 2295-2300.
[6]	Kumar G S, Kumar S, Tyagi S. Energy efficient and effective node deployment for wireless sensor network[J]. International Journal of Communication Systems, 2022, 35(9): 21535139.
[7]	Subramani S, Selvi M. Multi-objective PSO based feature selection for intrusion detection in iot based wireless sensor networks[J]. Optik, 2023, 273: 1-15.
[8]	Wei M, Rong C M, Liang E X, et al. An intrusion detection mechanism for Ipv6-based wireless sensor networks[J]. International Journal of Distributed Sensor Networks, 2022, 18(3): 1-15.
[9]	佘锋, 杨贵永, 刘建虎, 等. 路侧多传感器融合中的联合标定优化设计 [J]. 同济大学学报: 自然科学版, 2024, 52(11): 1750-1757.
	She Feng, Yang Gui-yong, Liu Jian-hu, et al. Optimization design of joint calibration for roadside multi-sensor fusion [J]. Journal of Tongji University (Natural Science), 2024, 52(11): 1750-1757.
[10]	Gautami A, Shanthini J, Karthik J. A Quasi-newton neural network based efficient intrusion detection systemfor wireless sensor network[J]. Computer Systems Science and Engineering, 2023, 45(1): 427-443.
[11]	蔡玲如, 朱理, 黄瑞崧, 等. 基于强化学习的无线传感器网络入侵检测攻防博弈研究 [J]. 计算机应用研究, 2022, 39(12): 3786-3790.
	Cai Ling-ru, Zhu Li, Huang Rui-song, et al. Research on attack-defence game based on reinforcement learning for wireless sensor network intrusion detection[J]. Application Research of Computers, 2022, 39(12): 3786-3790.
[12]	Hussain K, Xia Y, Onaizah A N, et al. Hybrid of WOA-ABC and proposed CNN for intrusion detection system in wireless sensor networks[J]. Optik, 2022, 271: 1-16.
[13]	Zhou J J, Yang T Y, Zhang J L, et al. Intrusion detection model for wireless sensor networks based on MC-GRU[J]. Wireless Communications and Mobile Computing, 2022: 1(1): 1-11.
[14]	Liu Z, Ghulam M, Zheng J, et al. Intrusion detection in wireless sensor network using enhanced empirical based component analysis[J]. Future Generation Computer Systems, 2022, 135: 181-193.
[15]	Lu X, Han D Z, Duan L T, et al. Intrusion detection of wireless sensor networks based on IPSO algorithm and BP neural network[J]. International Journal of Computational Science and Engineering, 2020, 22(2/3): 221-232.
[16]	郁滨, 熊俊. 基于平衡迭代规约层次聚类的无线传感器网络流量异常检测方案 [J].电子与信息学报, 2022, 44(1): 305-313.
	Yu Bin, Xiong Jun. A novel WSN traffic anomaly detection scheme based on BIRCH[J]. Journal of Electronics & Information Technology, 2022, 44(1): 305-313.
[17]	Rodriguez, A, Laio A. Clustering by fast search and find of density peaks[J]. Science, 2014, 344: 1492-1496.
[18]	仝钰, 庞新宇, 魏子涵. 基于 GADF-CNN的滚动轴承故障诊断方法[J]. 振动与冲击, 2021, 40(5): 247-260.
	Tong Yu, Pang Xin-yu, Wei Zi-han. Fault diagnosis method of rolling bearing based on GADF-CNN [J]. Journal of Vibration and Shock, 2021, 40(5): 247-260.
[19]	亓慧敏, 陈昂, 杨秀春. 基于无人机高光谱的内蒙古天然牧草氮磷钾含量的反演[J]. 草地学报, 2024, 32(5): 1500-1512.
	Qi Hui-min, Chen Ang, Yang Xiu-chun. Inversion of nitrogen, phosphorus and potassium content in natural grassland in Inner Mongolia based on UAV hyperspectral data[J]. Acta Agrestia Sinica, 2024, 32(5): 1500-1512.
[20]	Li S, Chen H, Wang M, et al. Slime mould algorithm: a new method for stochastic optimization[J]. Future Generation Computer Systems, 2020, 111: 300-323.

Metrics

Viewed

Full text

Abstract

Cited

Shared

Discussed

数据集合	攻击类型	代号	标签	训练集/个	测试集/个
WSN-DS	Blackhole	B1	0	1 000	200
	Grayhole	G1	1	1 000	200
	Flooding	F1	2	1 000	200
	Scheduling	S1	3	1 000	200
	正常数据	N1	4	1 000	200
KDD CUP'99	DoS	D2	5	1 000	200
	R2L	R2	6	1 000	200
	U2R	U2	7	1 000	200
	Probe	P2	8	1 000	200
	正常数据	N2	9	1 000	200