基于深度信念网络的WSN链路质量预测

doi:10.11959/j.issn.1000-436x.2017257

Abstract

Abstract:

After analyzing the existing link quality prediction models,a link quality prediction model for wireless sensor network was proposed,which was based on deep belief network.Support vector classification was employed to estimate link quality,so as to get link quality levels.Deep belief network was applied in extracting the features of link quality,and softmax was taken to predict the next time link quality.In different scenarios,compared with the model of link quality prediction based on logistic regression,BP neural network and Bayesian network methods,the experimental results show that the proposed prediction model achieves better precision.

Key words: wireless sensor network, link quality prediction, deep belief network, link quality level

CLC Number:

TP393

Lin-lan LIU,Jiang-bo XU,Yue LI,Zhi-yong YANG. Deep belief network-based link quality prediction for wireless sensor network[J]. Journal on Communications, 2017, 38(Z2): 17-25.

Figures/Tables 19

References 21

[1]	付帅, 马建峰, 李洪涛 ,等. 无线传感器网络中匿名的聚合节点选举协议[J]. 通信学报, 2015,36(2): 88-97.
	FU S , MA J F , LI H T ,et al. Anonymous aggregator election protocol for wireless sensor networks[J]. Journal on Communications, 2015,36(2): 88-97.
[2]	BACCOUR N , KOUB , ANIS A ,et al. Radio link quality estimation in wireless sensor networks:a survey[J]. ACM Transactions on Sensor Networks, 2012,8(4): 688-722.
[3]	翁丽娜, 杨杰, 柯海舟 ,等. 基于时间序列分析的链路质量预测和稳定路由算法研究[J]. 电子与信息学报, 2011,33(4): 885-890.
	WENG L N , YANG J , KE H Z ,et al. A time series analysis-based link quality prediction algorithm and its application to reliable routing[J]. Journal of Electronics ＆ Information Technology, 2011,33(4): 885-890.
[4]	尚凤军, 龚文娟, 耿哲 . 基于链路预测和网络编码的MAC机制[J]. 通信学报, 2016,37(1): 17-27.
	SHAN F J , GONG W J , GENG Z . MAC mechanism based on link prediction and network coding[J]. Journal on Communications, 2016,37(1): 17-27.
[5]	WENG L N , ZHANG P , FENG Z Y ,et al. Short-term link quality prediction using non-parametric time series analysis[J]. Science China Information Sciences, 2015,58(8): 1-15.
[6]	BOANO C A , ZUNIGA M A , VOIGT T ,et al. The triangle metric:fast link quality estimation for mobile wireless sensor networks[C]// The 19th International Conference on Computer Communications and Networks. 2010: 1-7.
[7]	WANG J , LIU Y , HE Y ,et al. QoF:towards comprehensive path quality measurement in wireless sensor networks[J]. IEEE Transactions on Parallel and Distributed Systems, 2014,25(4): 1003-1013.
[8]	BILDEA A , ALPHAND O , ROUSSEAU F ,et al. Link quality estimation with the gilbert-elliot model for wireless sensor networks[J]. Kips Transactions Partc, 2015,88(4): 495-504.
[9]	WEN J , AASAR Z , DARGIE W . A link quality estimation model for energy-efficient wireless sensor networks[C]// The 2015 IEEE International Conference on Communications (ICC). 2015: 6694-6700.
[10]	WANG Y , MARTONOSI M , PEH L S . Predicting link quality using supervised learning in wireless sensor networks[J]. ACM Sigmobile Mobile Computing ＆ Communications Review, 2007,11(3): 71-83.
[11]	LIU T , CERPA A E . Foresee(4C):wireless link prediction using link features[C]// International Conference on Information Processing in Sensor Networks. 2011: 294-305.
[12]	LIU T , CERPA A E . Talent:temporal adaptive link estimator with no training[C]// The 10th ACM Conference on Embedded Network Sensor Systems. 2012: 253-266.
[13]	郭志强, 王沁, 万亚东 ,等. 基于综合性评估的无线链路质量分类预测机制[J]. 计算机研究与发展, 2013,50(6): 1227-1238.
	GUO Z Q , WANG Q , WAN Y D ,et al. A classification prediction mechanism based on comprehensive assessment for wireless link quality[J]. Journal of Computer Research ＆ Development, 2013,5(6): 1227-1238.
[14]	舒坚, 汤津, 刘琳岚 ,等. 基于模糊支持向量回归机的WSNs链路质量预测[J]. 计算机研究与发展, 2015,52(8): 1842-1851.
	SHU J , TANG J , LIU L L ,et al. Fuzzy support vector regression-based link quality prediction model for wireless sensor networks[J]. Journal of Computer Research ＆ Development, 2015,52(8): 1842-1851.
[15]	卢记仓, 刘粉林, 罗向阳 ,等. 基于辨识性统计特征的PQ隐密图像识别算法[J]. 通信学报, 2015,36(3): 197-206.
	LU J C , LIU F L , LUO X Y ,et al. Recognition of PQ stego images based on identifiable statistical feature[J]. Journal on Communications, 2015,36(3): 197-206.
[16]	MOHAMMAD A K , MOHAMMAD M H . Deep belief network training improvement using elite samples minimizing free energy[J]. International Journal of Pattern Recognition ＆ Artificial Intelligence, 2015,29(5): 1-18.
[17]	梁淑芬, 刘银华, 李立琛 . 基于LBP和深度学习的非限制条件下人脸识别算法[J]. 通信学报, 2014,35(6): 154-160.
	LIANG S F , LIU Y H , LI L C.Face recognition under unconstrained based on LBP and deep learning . Chinese relation extraction based on deep belief nets[J]. Journal on Communications, 2014,35(6): 154-160.
[18]	张亚军, 刘宗田, 周文 . 基于深度信念网络的事件识别[J]. 电子学报, 2017,45(6): 1415-1423.
	ZHANG Y J , LIU Z T , ZHOU W . Event recognition based on deep belief network[J]. ACTA Electronica Sinica, 2017,45(6): 1415-1423.
[19]	顾珊波, 邵枫, 蒋刚毅 ,等. 基于支持向量回归的立体图像客观质量评价模型[J]. 电子与信息学报, 2012,34(2): 368-374.
	GU S B , SHAO F , JIANG G Y ,et al. Objective stereoscopic image quality assessment model based on support vector regression[J]. Journal of Electronics ＆ Information Technology, 2012,34(2): 368-374.
[20]	RAHHAL M M A , BAZI Y , ALHICHRI H ,et al. Deep learning approach for active classification of electrocardiogram signals[J]. Information Sciences, 2016,345(C): 340-354.
[21]	刘琳岚, 樊佑磊, 舒坚 ,等. 一种基于BP神经网络的WSNs链路质量预测方法[J]. 计算机研究与发展, 2011,48(Suppl): 212-215.
	LIU L L , FAN Y L , SHU J ,et al. A link quality prediction method for WSNs based on BP artificial neural network[J]. Journal of Computer Research ＆ Development, 2011,48(Suppl): 212-215.

Metrics

Recommended 0

No Suggested Reading articles found!

链路质量等级	PRR	等级值
very good link	90%≤PRR≤100%	1
good link	70%≤PRR<90%	2
intermediate link	40%≤PRR<70%	3
bad link	10%≤PRR<40%	4
very bad link	0≤PRR<10%	5

参数属性	参数值
首层RBM输入神经元个数	49,59,69,79,89,99,109
RBM学习率	0.1
训练最大迭代次数	50
RBM的层数	2,3,4,5,6
预测评价指标	MSE

参数属性	参数值
发送功率/dBm	31
信道	26
探测分组数量/个	30
分组发送速率/（个·秒^?1）	5
发送周期/ms	200
测试周期/s	10
节点间距/m	20

n的值			MSE
n的值	2层RBM	3层RBM	4层RBM	5层RBM	6层RBM
50	0.312 500	0.322 920	0.328 130	0.343 750	0.307 290
60	0.402 120	0.333 330	0.312 170	0.349 210	0.328 040
70	0.306 450	0.295 700	0.301 080	0.317 200	0.370 970
80	0.284 150	0.311 480	0.371 580	0.377 050	0.322 40
90	0.338 890	0.288 890	0.305 560	0.366 670	0.277 78
100	0.305 080	0.338 980	0.265 540	0.378 530	0.344 63
110	0.373 560	0.344 830	0.310 340	0.327 590	0.304 60

n的值			MSE
n的值	2层RBM	3层RBM	4层RBM	5层RBM	6层RBM
50	0.407 960	0.333 330	0.358 210	0.452 740	0.313 43
60	0.459 600	0.409 090	0.439 390	0.368 690	0.247 47
70	0.271 790	0.492 310	0.394 870	0.548 720	0.338 46
80	0.390 630	0.348 960	0.385 420	0.463 540	0.317 71
90	0.222 220	0.428 570	0.460 320	0.280 420	0.174 60
100	0.559 140	0.208 600	0.268 820	0.134 410	0.408 60
110	0.196 720	0.426 230	0.431 690	0.387 980	0.524 459

Deep belief network-based link quality prediction for wireless sensor network

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 19

References 21

Related Articles 15

Metrics

Recommended 0

实验场景	n	RBM层数	MSE
场景1	100	4	0.265 540
场景2	100	5	0.134 410
场景3	60	4	0.057 143

实验场景	BP	LR	BN	LQP-DBN
场景1	0.349 500	0.339 806	0.330 023	0.265 540
场景2	0.104 200	0.098 958	0.103 482	0.057 143
场景3	0.513 900	0.148 148	0.151 005	0.134 410

模型	平均训练时间/s	平均预测时间/s	模型类别
BP	3.354 0	0.031 2	非线性
LR	0.483 6	0.001 2	线性
BN	2.277 6	0.015 6	非线性
LQP-DBN	67.813 6	0.075 4	非线性

[1]	Yue ZHAO, Zan LI, Bing LI, Benjian HAO. TDoA localization-refused area analysis and node placement strategy research [J]. Journal on Communications, 2023, 44(1): 1-13.
[2]	Cuiran LI, Xuejie WANG, Jianli XIE, Anqi LYU. Routing algorithm for railway monitoring linear WSN based on improved PSO [J]. Journal on Communications, 2022, 43(5): 155-165.
[3]	Xiujuan DU, Lijuan WANG, Jingping LIU, Long JIN. Ranging localization method for nodes in underwater wireless sensor network based on zeroing neural dynamics [J]. Journal on Communications, 2022, 43(10): 177-185.
[4]	Aijing SUN, Shichang LI, Yicai ZHANG. WSN clustering routing algorithm based on PSO optimized fuzzy C-means [J]. Journal on Communications, 2021, 42(3): 91-99.
[5]	Tianjing WANG,Xiuqin LI,Guangwei BAI,Hang SHEN. Multi-target localization algorithm based on adaptive grid in wireless sensor network [J]. Journal on Communications, 2019, 40(7): 197-207.
[6]	Xiaochen HAO,Ning YAO,Lixia XIE,Jiaojiao WANG,Liyuan WANG. Research on WSN lifetime optimization game algorithm combined power and channel [J]. Journal on Communications, 2019, 40(4): 62-70.
[7]	Linlan LIU,Shengrong GAO,Jian SHU. Link quality prediction based on random forest [J]. Journal on Communications, 2019, 40(4): 202-211.
[8]	Xiaonian WU,Chuyun ZHANG,Runlian ZHANG,Yaping SUN. Clustering routing protocol based on improved PSO algorithm in WSN [J]. Journal on Communications, 2019, 40(12): 114-123.
[9]	Haoran LIU,Heyao ZHAO,Yujing DENG,Xingqi WANG,Rongrong YIN. Coverage control algorithm for wireless sensor networks based on non-cooperative game [J]. Journal on Communications, 2019, 40(1): 71-78.
[10]	Jian SHU,Manlan LIU,Yaqing SHANG,Yubin CHEN,Linlan LIU. Link quality prediction model based on Gaussian process regression [J]. Journal on Communications, 2018, 39(7): 148-156.
[11]	Junbin LIANG,Xiang ZHOU,Taoshen LI. Energy saving proactive neighbor discovery algorithm in mobile low-duty-cycle wireless sensor network [J]. Journal on Communications, 2018, 39(4): 45-55.
[12]	Xiaochen HAO,Liyuan WANG,Jinshuo LIU,Lixia XIE,Wenhuan ZHANG. Resource allocation optimization algorithm based on double populations differential evolution in WSN [J]. Journal on Communications, 2018, 39(4): 68-75.
[13]	Tian WANG,Yang LI,Weijia JIA,Guojun WANG,Shaoliang PENG. Research progress of sensor-cloud security [J]. Journal on Communications, 2018, 39(3): 35-52.
[14]	Chan JIANG,Taoshen LI,Junbin LIANG. Low-latency algorithm for improving data persistence in mobile low-duty-cycle wireless sensor network [J]. Journal on Communications, 2018, 39(3): 53-62.
[15]	Xiangmin JI,Bo ZHAO,Jinhui LIU,Jianwei JIA,Huanguo ZHANG,Shuang XIANG. WSN key recovery attack based on symmetric matrix decomposition [J]. Journal on Communications, 2018, 39(10): 87-96.

n的值			MSE
n的值	2层RBM	3层RBM	4层RBM	5层RBM	6层RBM
50	0.106 740	0.073 034	0.067 416	0.095 506	0.106 740
60	0.114 290	0.102 860	0.057 143	0.102 860	0.108 570
70	0.122 090	0.093 023	0.081 395	0.081 395	0.087 209
80	0.100 590	0.094 675	0.088 757	0.078 313	0.088 757
90	0.072 289	0.108 430	0.090 361	0.088 757	0.114 460
100	0.797 550	0.092 025	0.110 430	0.067 485	0.098 160
110	0.087 50	0.125 00	0.093 750	0.106 250	0.093 750