An Improved HMM Model for Sensing Data Predicting in WSN

Wireless sensor networks (WSN) have been employed in numerous fields of real world applications. Data failure and noise reduction still remain tough unsolved problems for WSN. Predicting methods for data recovery by empirical treatment, mostly based on statistics has been studied exclusively. Machine learning models can greatly enhance the predicting performance. In this paper, an improved HMM is proposed for multi-step predicting of wireless sensing data given historical data. The proposed model is based on clustering of wireless sensing data and multi-step predicting is accordingly accomplished for different varying patterns using HMM whose parameters are optimized by Particle Swarm Optimization (PSO). We evaluate our model on two real wireless sensing datasets, and comparison between Naive Bayesian, Grey System, BP Neural Networks and traditional HMMs are conducted. The experimental results show that our proposed model can provide higher accuracy in sensing data predicting. This proposed model is promising in the fields of agriculture, industry and other domains, in which the sensing data usually contains various varying patterns.

[1]  Gustavo Medeiros de Araújo,et al.  An optimized Markov model to predict link quality in mobile wireless sensor networks , 2012, 2012 IEEE Symposium on Computers and Communications (ISCC).

[2]  Naixue Xiong,et al.  Data prediction, compression, and recovery in clustered wireless sensor networks for environmental monitoring applications , 2016, Inf. Sci..

[3]  A. Gupta,et al.  Stock market prediction using Hidden Markov Models , 2012, 2012 Students Conference on Engineering and Systems.

[4]  Jian Shen,et al.  A WSN-based prediction model of microclimate in a greenhouse using extreme learning approaches , 2016, 2016 18th International Conference on Advanced Communication Technology (ICACT).

[5]  Jianbo Yu,et al.  Health Condition Monitoring of Machines Based on Hidden Markov Model and Contribution Analysis , 2012, IEEE Transactions on Instrumentation and Measurement.

[6]  Ke Shi,et al.  Data Mining Techniques for Wireless Sensor Networks: A Survey , 2013, Int. J. Distributed Sens. Networks.

[7]  Ye Wei,et al.  A Self-Adaptive Hidden Markov Model for Emotion Classification in Chinese Microblogs , 2015 .

[8]  Li Li,et al.  An Improved Recommender Based on Hidden Markov Model , 2012, PRICAI.

[9]  Kumar Padmanabh,et al.  A Scalable WSN Based Data Center Monitoring Solution with Probabilistic Event Prediction , 2012, 2012 IEEE 26th International Conference on Advanced Information Networking and Applications.

[10]  Zainuddin Hassan,et al.  An adaptive HMM based approach for improving e-Learning methods , 2014, 2014 World Congress on Computer Applications and Information Systems (WCCAIS).

[11]  Teik-Toe Teoh,et al.  Hidden Markov Model for hard-drive failure detection , 2012, 2012 7th International Conference on Computer Science & Education (ICCSE).

[12]  Simon A. Dobson,et al.  Failure detection in wireless sensor networks: A sequence-based dynamic approach , 2014, TOSN.

[13]  Charu C. Aggarwal,et al.  Managing and Mining Sensor Data , 2013, Springer US.