logo

SCIENCE CHINA Information Sciences, Volume 63 , Issue 10 : 202306(2020) https://doi.org/10.1007/s11432-019-2850-3

Support vector machine based machine learning method for GS 8QAM constellation classification in seamless integrated fiber and visible light communication system

More info
  • ReceivedSep 5, 2019
  • AcceptedMar 25, 2020
  • PublishedSep 21, 2020

Abstract

Visible light communication (VLC) network over optical fiber has become a potential candidate in ultra-high speed indoor wireless communication. To mitigate signal distortion accumulated in optical fiber and VLC channel, we present to utilize support vector machine (SVM) for constellation classification in two kinds of geometrically-shaped 8QAM (quadrature amplitude modulation) seamless integrated fiber and VLC system. We introduce 4 sub-bands to simulate multi-user. Experimental results show that system performance can be significantly improved, and transmission at $-$2.5 dBm input optical power under 7% forward error correction (FEC) threshold can be realized employing Circular $(7,~1)$ geometrically-shaped 8QAM and SVM. At overall capacity of 960 Mbps, Q-factor increases by up to 11.5 dB.


Acknowledgment

This work was partially supported by National Key Research and Development Program of China (Grant No. 2017YFB0403603) and National Natural Science Foundation of China (Grant No. 61571133).


Supplement


References

[1] Chi N, Haas H, Kavehrad M. Visible light communications: demand factors, benefits and opportunities [Guest Editorial]. IEEE Wireless Commun, 2015, 22: 5-7 CrossRef Google Scholar

[2] Wang Y, Chi N, Wang Y. Network Architecture of a High-Speed Visible Light Communication Local Area Network. IEEE Photon Technol Lett, 2015, 27: 197-200 CrossRef ADS Google Scholar

[3] Chow C W, Yeh C H, Liu Y. Network Architecture of Bidirectional Visible Light Communication and Passive Optical Network. IEEE Photonics J, 2016, 8: 1-7 CrossRef ADS Google Scholar

[4] Komine T, Nakagawa M. Integrated system of white LED visible-light communication and power-line communication. IEEE Trans Consumer Electron, 2003, 49: 71-79 CrossRef Google Scholar

[5] Langer KD, Grubor J, Bouchet O, et al. Optical wireless communications for broadband access in home area networks. In: Proceedings of 2008 10th Anniversary International Conference on Transparent Optical Networks, 2008. 149--154. Google Scholar

[6] Tsonev D, Videv S, Haas H. Light fidelity (Li-Fi): towards all-optical networking. In: Proceedings of the SPIE, 2013. Google Scholar

[7] Chen C Y, Wu P Y, Lu H H. Bidirectional 16-QAM OFDM in-building network over SMF and free-space VLC transport. Opt Lett, 2013, 38: 2345 CrossRef Google Scholar

[8] Haigh P A, Chvojka P, Zvanovec S, et al. Experimental verification of visible light communications based on multi-band CAP modulation. In: Proceedings of Optical Fiber Communication Conference, 2015. Google Scholar

[9] Wang Y, Chi N, Wang Y. Network Architecture of a High-Speed Visible Light Communication Local Area Network. IEEE Photon Technol Lett, 2015, 27: 197-200 CrossRef ADS Google Scholar

[10] Godard D. Self-Recovering Equalization and Carrier Tracking in Two-Dimensional Data Communication Systems. IEEE Trans Commun, 1980, 28: 1867-1875 CrossRef Google Scholar

[11] Zhao J, Qin C, Zhang M. Investigation on performance of special-shaped 8-quadrature amplitude modulation constellations applied in visible light communication. Photon Res, 2016, 4: 249 CrossRef Google Scholar

[12] Thomas C, Weidner M, Durrani S. Digital Amplitude-Phase Keying with M-Ary Alphabets. IEEE Trans Commun, 1974, 22: 168-180 CrossRef Google Scholar

[13] William H, et al. The Art of Scientific Computing. 3rd ed. Cambridge: Cambridge University Press, 2011. 883--886. Google Scholar

[14] Cortes C, Vapnik V. Support-vector networks. Mach Learn, 1995, 20: 273-297 CrossRef Google Scholar

[15] Chih-Wei Hsu , Chih-Jen Lin . A comparison of methods for multiclass support vector machines. IEEE Trans Neural Netw, 2002, 13: 415-425 CrossRef Google Scholar

[16] Graf H P, Cosatto E, Bottou L, et al. Parallel support vector machines: the cascade svm. In: Proceedings of Advances in Neural Information Processing Systems, 2005. 521--528. Google Scholar

[17] Burges C J C. Data Min Knowledge Discovery, 1998, 2: 121-167 CrossRef Google Scholar

[18] Nazeer K A, Sebastian M P. Improving the accuracy and efficiency of the k-means clustering algorithm. In: Proceedings of the World Congress on Engineering, London: Association of Engineers, 2009. 1--3. Google Scholar

[19] Hsu CW, Chang CC, Lin CJ. A practical guide to support vector classification 2003. Google Scholar

[20] Wang D, Zhang M, Li Z, et al. Optimized SVM-based decision processor for 16QAM coherent optical systems to mitigate NLPN. In: Proceedings of Asia Communications and Photonics Conference, 2015. Google Scholar

[21] Chi N, Zhou Y, Liang S. Enabling Technologies for High-Speed Visible Light Communication Employing CAP Modulation. J Lightwave Technol, 2018, 36: 510-518 CrossRef ADS Google Scholar

[22] Chi N, Zhao Y, Shi M. Gaussian kernel-aided deep neural network equalizer utilized in underwater PAM8 visible light communication system. Opt Express, 2018, 26: 26700 CrossRef ADS Google Scholar

[23] Chi N, Shi M. Advanced modulation formats for underwater visible light communications. Chinese Optics Letters. 2018,16(12):120603. Google Scholar

[24] Niu W, Ha Y, Chi N. Novel phase estimation scheme based on support vector machine for multiband-CAP visible light communication system. In: Proceedings of 2018 Asia Communications and Photonics Conference, 2018. 1--3. Google Scholar

  • Figure 1

    (Color online) Overview of optical fiber based VLC network.

  • Figure 2

    (Color online) Constellations of GS 8QAM Diamond and Circular $(7,~1)$.

  • Figure 3

    (Color online) (a) Schematic diagram of linear separable SVM and (b) kernel function for linear inseparability.

  • Figure 4

    (Color online) Flow chart of SVM. (a) Training phase; (b) testing phase.

  • Figure 5

    (Color online) (a) Schematic diagram of centroid estimation and SVM classification of biased distribution; protectłinebreak (b) simulation results of Circular $(7,1)$ with 2.5% training data.

  • Figure 6

    (Color online) Experimental setup of GS 8QAM seamless integrated fiber and VLC system.

  • Figure 7

    (Color online) Spectrums of Diamond (a) and Circular $(7,1)$ (b) after optical fiber and LED channel.

  • Figure 12

    (Color online) (a) BER performance versus size of training data set of Diamond and (b) the classification results of SVM with different kernels.

  • Figure 13

    (Color online) (a) BER performance versus size of training data set of Circular $(7,1)$ and (b) the classification results of SVM with different kernels.

  • Figure 14

    (Color online) BER of 4 bands for Diamond GS 8QAM versus input optical power with different kernel SVMs or without SVM.

  • Figure 15

    (Color online) Constellation diagrams of Band1 (a) and Band4 (b) at different optical power (2 dBm, $-$4 dBm) for Diamond after CMA and the classification and phase correction performance of SVM.

  • Figure 16

    (Color online) BER of 4 bands for Circular (7,1) GS 8QAM versus input optical power with different kernel SVMs or without SVM.

  • Figure 17

    (Color online) Constellation diagrams of Band1 (a) and Band4 (b) at different optical power (2 dBm, $-$4 dBm) for Circular $(7,1)$ after CMA and the classification and phase correction performance of SVM.

  • Table 1  

    Table 1Expression of different kernel

    Kernel Expression
    Linear $~K(x_1~,x_2~)=x_1~\cdot~x_2^{\rm~T}~$
    Polynomial $~K(x_1~,x_2~)=(\gamma~(x_1~\cdot~x_2^{\rm~T})+r)^d~$
    RBF $~K(x_1~,x_2~)={\rm~exp}(-\gamma~\rVert~x_1~-~x_2~\rVert~^2)~$
    Sigmoid $~K(x_1~,x_2~)={\rm~tanh}(\gamma~(x_1~,x_2^2)+r)~$
  • Table 2  

    Table 2Expression of different kernels

    Linear Poly RBF Sigmoid
    Diamond $~C=1~$ $~C=7.55,~{\rm~gamma}=0.05,{\rm~degree}=3~$ $~C=0.31,{\rm~gamma}=0.23~$ $~C=2.38,{\rm~gamma}=0.05~$
    Circular $~C=1~$ $~C=0.17,~{\rm~gamma}=2.38,{\rm~degree}=3~$ $~C=0.07,{\rm~gamma}=1~$ $~C=1,{\rm~gamma}=0.42~$

Copyright 2020  CHINA SCIENCE PUBLISHING & MEDIA LTD.  中国科技出版传媒股份有限公司  版权所有

京ICP备14028887号-23       京公网安备11010102003388号