分布式传感器网络混合探测信号分类方法

doi:10.3969/j.issn.1000-436x.2012.z1.008

通信学报 ›› 2012, Vol. 33 ›› Issue (Z1): 53-57.doi: 10.3969/j.issn.1000-436x.2012.z1.008

分布式传感器网络混合探测信号分类方法

李侃¹,许航¹,黄忠华²

¹ 北京理工大学计算机学院，北京 100081
² 北京理工大学机电工程学院，北京 100081

出版日期:2012-09-25 发布日期:2017-08-03
基金资助:
国家自然科学基金资助项目

Classification method for mixed detection signal in the distributed sensor network

Kan LI¹,Hang XU¹,Zhong-hua HUANG²

¹ School of Computer,Beijing Institute of Technology,Beijing 100081,China
² School of Mechano-Electronics Engineering,Beijing Institute of Technology,Beijing 100081,China

Online:2012-09-25 Published:2017-08-03
Supported by:
The National Natural Science Foundation of China

摘要/Abstract

摘要：

针对分布式传感器网络的局限性特征，研究分布式传感器网络混合探测信号的分类算法。提出了基于属性重要度的贝叶斯分类算法，该算法继承了朴素贝叶斯分类算法结构简单、运算快捷的特点，同时弥补了类条件独立假设带来的缺陷，在实践中具有较高的分类精度，其特点符合混合探测信号的分类要求。实验结果表明，该算法分类效果优于同类分类算法，可以有效地完成混合探测信号的分类任务。

关键词: 朴素贝叶斯分类器, 属性重要度, 分布式传感器网络, 混合探测信号

Abstract:

Taking into account the limitations of the distributed sensor networks,a simple and efficient classification method was found.According to the main idea of na?ve Bayes classification (NBC) algorithm,a new na?ve Bayes classification based on attribute significance (NBCBAS) was proposed.The algorithm inherited the characteristics of NBC algorithm that was simple and fast computation.At the same time,the algorithm made up for the defects of conditional independence assumption.It had high classification accuracy in practice.The characteristics of the NBCBAS met the classification requirements of the mixed detection signal.At last,the NBCBAS was tested on UCI datasets and mixed detection signal datasets.The results illustrate that our algorithm improves the classification performance.

Key words: na?ve Bayes classification, attribute significance, distributed sensor networks, mixed detection signal

李侃,许航,黄忠华. 分布式传感器网络混合探测信号分类方法[J]. 通信学报, 2012, 33(Z1): 53-57.

Kan LI,Hang XU,Zhong-hua HUANG. Classification method for mixed detection signal in the distributed sensor network[J]. Journal on Communications, 2012, 33(Z1): 53-57.

图/表 5

表1

表2

表3

表4

表5

参考文献 10

[1]	GUEVARA J , BARRERO J , VARGAS F ,et al. Environmental wireless sensor network for road traffic applications[J]. Intelligent Transport Systems, 2012,6(2): 177-186.
[2]	SEKHAR V C , SARVABHATLA M . Security in wireless sensor networks with public key techniques[A]. International Conference on Computer Communication and Informatics[C]. Coimbatore, 2012. 1-16.
[3]	JAWHAR I , MOHAMED N , AQRAWAL D P . Linear wireless sensor networks:classification and applications[J]. Journal of Network and Computer Applications, 2011,34(5): 1671-1682.
[4]	BROOKS R R . Distributed target classification and tracking in sensor networks[J]. Proceedings of the IEEE, 2003,91(8): 1163-1171.
[5]	DUARTE M F , HU Y F . Vehicle classification in distributed sensor networks[J]. Journal of Parallel and Distributed Computing, 2004,64(7): 826-838.
[6]	COMMAULT C , DION J M , TRINH D H .et al. Sensor classification for the fault detection and isolation a structural approach[J]. International Journal of Adaptive Control and Signal Processing, 2011,25(1): 1-17.
[7]	GONG B , IM J , MOUNTRAKIS G . An artificial immune network approach to multi-sensor land use/land cover classification[J]. Remote Sensing of Environment, 2011,115(2): 600-614.
[8]	LIU L . A binary-classification-tree based framework for distributed[A]. target classification in multimedia sensor networks[C]. INFOCOM,Orlando, 2012. 594-602.
[9]	ABOELELA E H , KHAN A H . Wireless sensors and neural networks for intruders detection and classification[A]. International Conference on Information Networking[C]. 2012. 138-143.
[10]	GHARAIBEH K M , YAQOT A . Target classification in wireless sensor network using partical swarm optimization[A]. Sensors Applications Symposium(SAS)[C]. 2012. 1-5.

序号	数据集（样本数-属性数）	算法的准确率/%
序号	数据集（样本数-属性数）	BN	NBC	A-NBC	NBCBAS
1	Annealing(798-38)	92.427 6	75.835 2	79.064 6	92.761 7
2	Balance Scale(625-4)	72.320 0	90.400 0	91.040 0	91.200 0
3	Lenses(24-4)	70.833 3	70.833 0	70.833 3	83.833 3
4	Statlog-Heart(270-13)	81.111 1	83.703 7	82.592 6	86.296 3
5	Hayes-Roth(160-5)	84.090 9	84.090 9	76.515 2	88.636 4
6	Hores Colic(368-27)	76.550 4	76.302 1	73.828 1	77.864 6
7	Iris(150-4)	92.666 7	96.000 0	93.333 3	97.333 3
8	Liver Disorders(345-7)	56.811 6	56.811 6	67.246 4	71.884 1
9	Lymphography(148-18)	87.837 8	87.162 2	80.405 4	94.594 6
10	Lung Cancer(32-56)	93.750 0	87.500 0	65.625 0	100.000 0
11	ORH Digits(5 620-64)	91.774 9	91.794 0	91.334 5	92.437 7
12	Page Blocks(5 473-10)	93.400 0	90.846 0	90.846 0	90.969 0
13	Segmentation(2 310-19)	90.400 0	81.066 7	80.216 5	81.200 0
14	Sonar(208-60)	80.288 5	67.788 5	80.769 2	82.211 5
15	SPECT Heart(267-22)	83.750 0	83.000 0	71.250 0	90.000 0
16	Waveform(5 000-21)	79.840 0	80.000 0	80.000 0	79.960 0

序号	数据样本数量	类别数量	决策属性数量	条件属性数量	类别先验概率
1	10 000	4	4	2	0.25
2	10 000	4	4	2	0.25
3	20 000	8	4	2	0.125
4	10 000	4	4	6	0.25
5	20 000	8	4	6	0.125

序号	分类精度/%
序号	NBC	NBCBAS
1	96.85	99.95
2	92.30	96.25
3	88.15	92.75
4	94.00	98.95
5	93.70	97.50

序号	数据样本数量	类别数量	决策属性数量	条件属性数量	类别先验概率	正常级节点比例
1	10 000	4	4	2	0.25	较高
2	10 000	4	4	2	0.25	较高
3	20 000	8	4	2	0.125	较高
4	10 000	4	4	6	0.25	较高
5	20 000	8	4	6	0.125	较高
6	10 000	4	4	2	0.25	较低
7	10 000	4	4	2	0.25	较低
8	20 000	8	4	2	0.125	较低
9	10 000	4	4	6	0.25	较低
10	20 000	8	4	6	0.125	较低

序号	分类精度/%
序号	NBC	NBCBAS
1	95.35	98.95
2	92.50	96.45
3	86.45	91.60
4	94.15	98.75
5	95.20	97.65
6	90.85	98.55
7	84.60	94.15
8	79.00	86.85
9	87.05	96.95
10	90.70	94.50

分布式传感器网络混合探测信号分类方法

Classification method for mixed detection signal in the distributed sensor network

在线阅读

PDF下载

可视化

被引次数

摘要/Abstract

引用本文

使用本文

图/表 5

参考文献 10

相关文章 0

Metrics

推荐阅读 0