L-PLC Channel Characteristics Prediction Based on SVM

Time series prediction can be a very useful tool in communication to predict the behavior of system. In this paper, support vector machine (SVM) is applied to predict the channel characteristics of low-voltage power line communication (L-PLC), which is a vast infrastructure of power distribution and offers an alternative and cost-effective Internet access technology. Firstly, the largest Lyapunov exponent and the saturated correlation dimension of time series measured from power line are calculated. According to the results, the L-PLC channel is manifested to be a chaotic system. Then, a prediction model of L-PLC channel characteristics is proposed based on phase space reconstruction theory and SVM algorithm. In this paper, the parameters of SVM algorithm are chosen by the minimum mean square error (MSE) principle and the selection principles of parameters are particularly discussed. The actual time series are used in experimental simulations. The simulation results indicate that the prediction model based on SVM can be used to predict L-PLC channel characteristics accurately. In addition, the relations among sampling interval, predicted step and prediction precision are discussed.

[1]  D. Sabolic,et al.  Signal propagation modeling in power-line communication networks , 2005, IEEE Transactions on Power Delivery.

[2]  Xueqin Zhang,et al.  Support Vector Machines for Anomaly Detection , 2006, 2006 6th World Congress on Intelligent Control and Automation.

[3]  Yi Pan,et al.  Clustering support vector machines for protein local structure prediction , 2007, Expert Syst. Appl..

[4]  F. Takens.  Detecting strange attractors in turbulence , 1981 .

[5]  Gan Yu-tao.  Research on modeling of the low-voltage power line communication channel based on chaos theory , 2007 .

[6]  Shu Hui.  GA multi-parameter identification of low-voltage powerline carrier communication channel model , 2005 .

[7]  Shijie Cheng,et al.  A novel multicarrier signal transmission system over multipath channel of low-voltage power line , 2004 .

[8]  Zhu Changchun,et al.  Prediction of the chaotic time series using support vector machines for fuzzy rule-based modeling , 2005 .

[9]  Francis Eng Hock Tay,et al.  Support vector machine with adaptive parameters in financial time series forecasting , 2003, IEEE Trans. Neural Networks.

[10]  Chunru Wan,et al.  Classification using support vector machines with graded resolution , 2005, 2005 IEEE International Conference on Granular Computing.

[11]  Robert A. Lordo,et al.  Learning from Data: Concepts, Theory, and Methods , 2001, Technometrics.

[12]  Ke Wang,et al.  Localization site prediction for membrane proteins by integrating rule and SVM classification , 2005, IEEE Transactions on Knowledge and Data Engineering.

[13]  Klaus Dostert,et al.  A multipath model for the powerline channel , 2002, IEEE Trans. Commun..

[14]  Johan A. K. Suykens,et al.  Least Squares Support Vector Machines , 2002 .

[15]  Vladimir N. Vapnik,et al.  The Nature of Statistical Learning Theory , 2000, Statistics for Engineering and Information Science.

[16]  Zhang Xuegong,et al.  INTRODUCTION TO STATISTICAL LEARNING THEORY AND SUPPORT VECTOR MACHINES , 2000 .

[17]  Wang Qinruo,et al.  Communication reliability design of low-voltage power line communication system , 2003, SMC'03 Conference Proceedings. 2003 IEEE International Conference on Systems, Man and Cybernetics. Conference Theme - System Security and Assurance (Cat. No.03CH37483).

[18]  Johan Suykens.  Least Squares Support Vector Machines : an Overview , 2002 .

[19]  Farhad Rachidi,et al.  Analysis of power line communication networks using a new approach based on an efficient measurement technique , 2003, 2003 IEEE Symposium on Electromagnetic Compatibility. Symposium Record (Cat. No.03CH37446).