The Use of Support Vector Machines in Wi-Fi Based Indoor Localization Detection and the Effect of Kernel Function Selection on Classification Performance: An Example of Finding the Location of Kindergarten Students

Abstract

In recent years, the importance of locating children indoors has increased due to the problems in child safety. In this study, the locations of children in different rooms were determined by using wireless signal strength and Support Vector Machines classification algorithm. In addition, the effects of the algorithm on the classification performance were examined using different kernel functions and cross-validation technique was used for performance analysis. In the performance analysis, Confusion matrix, Precision, Sensitivity, F-Score, AUC, Accuracy, Kappa statistical value and Root Mean Square Error value were examined. As a result of the analysis, when linear kernel functions are preferred, the highest performance metrics were obtained and 97.9% success was observed

Keywords

• WiFi
• Support vector machines
• Kernel
• Indoor localization

Destek Vektör Makinelerinin Wi-Fi Tabanlı İç Mekan Lokalizasyon Tespitinde Kullanımı ve Çekirdek Fonksiyon Seçiminin Sınıflandırma Performansına Etkisi

Öz

Son yıllarda çocuk güvenliğinde yaşanan kazaların artması nedeniyle iç mekanlarda çocukların yerini tespit etme çalışmaları önem kazanmıştır. Bu çalışmada kablosuz sinyal gücü ve Destek Vektör Makineleri sınıflandırma algoritması kullanılarak iç mekanlarda farklı odalarda bulunan insanların konumları tespit edilmiştir. Algoritmanın performansının arttırılması için farklı çekirdek fonksiyonları denenmiş ve çekirdek fonksiyonu seçiminin algoritmanın sınıflandırma performansına etkisi incelenmiştir. Performans ölçüm yöntemi olarak 10 kat çapraz doğrulama yöntemi kullanılmıştır. Performans değerlendirmesi, çapraz doğrulama öncesi ve sonrası sınıflandırma performansları karşılaştırılarak yapılmıştır. Yapılan performans değerlendirmesi sonucu iç mekanda konum belirlemede Destek Vektör Makineleri algoritması kullanılırken doğrusal çekirdek fonksiyonunun seçimi uygun görülmüştür.

Anahtar Kelimeler

• Kablosuz Sinyal Gücü
• Destek Vektör Makineleri
• Çekirdek Fonksiyon
• İç Mekan Konum Belirleme
• WiFi
• Support vector machines
• Kernel
• Indoor localization

Kaynakça

1. Arlot S., Celisse, A survey of cross-validation procedures for model selection. Statistics surveys, A. 2010; 4, 40-79.
2. Bayıroğlu H., Ayan K. Android üzerinde web tabanlı çocuk takip sistemi. Sakarya Üniversitesi Fen Bilimleri Enstitüsü Dergisi 2014; 18(2): 87-91.
3. Bejuri W., ve Mohamad M. M. Wireless LAN/FM radio-based robust mobile indoor positioning: an initial outcome. Int J Softw Eng Appl 2014; 8(2): 313-324.
4. Chandra A., Jain, S., Qadeer MA. Implementation of location awareness and sharing system based on GPS and GPRS using J2ME, PHP and MYSQL. Paper presented at the 2011 3rd International Conference on Computer Research and Development.
5. Chen R.-C., Huang SL. A new method for indoor location base on radio frequency identification. Paper presented at the WSEAS International Conference. Proceedings. Mathematics and Computers in Science and Engineering 2009.
6. Cortes C., Vapnik, V. Support-vector networks. Machine learning 1995; 20(3):273-297.
7. Frank A. UCI machine learning repository 2010. http://archive. ics. uci. edu/ml.
8. Kamath U., De Jong, K., & Shehu, A. Effective automated feature construction and selection for classification of biological sequences. PloS one, 2014; 9(7), e99982.

9. Li X., Nsofor, GC., Song L. A comparative analysis of predictive data mining techniques. International Journal of Rapid Manufacturing, 1(2), 2009;150-172.
10. Liao YC., Jeng JT., Chuang CC., Chen JC. Systematic design for the global positional systems with application in intelligent google android phone. Paper presented at the Proceedings 2011 International Conference on System Science and Engineering 2011.
11. Orozco-Arias S., Isaza G., Guyot R., Tabares-Soto R. A systematic review of the application of machine learning in the detection and classification of transposable elements. PeerJ, 2019;7, e8311.
12. Rawal K., Ramaswamy R. Genome-wide analysis of mobile genetic element insertion sites. Nucleic acids research, 2011;39(16), 6864-6878.
13. Rida ME., Liu F., Jadi Y., Algawhari AA. A., Askourih, A. Indoor location position based on bluetooth signal strength. Paper presented at the 2015 2nd International Conference on Information Science and Control Engineering.
14. Rohra JG., Perumal B., Narayanan SJ., Thakur P., & Bhatt R. B. User localization in an indoor environment using fuzzy hybrid of particle swarm optimization & gravitational search algorithm with neural networks. Paper presented at the Proceedings of Sixth International
15. Conference on Soft Computing for Problem Solving 2017. Sabanci K., Yigit E., Ustun D., Toktas A., Aslan MF. Wifi based indoor localization: application and comparison of machine learning algorithms. Paper presented at the 2018 XXIIIrd International Seminar/Workshop on Direct and Inverse Problems of Electromagnetic and Acoustic Wave Theory (DIPED).
16. Schietgat L., Vens C., Cerri R., Fischer CN., Costa E., Ramon, J., . . . Blockeel, H. A machine learning based framework to identify and classify long terminal repeat retrotransposons. PLoS computational biology, 2018;14(4), e1006097.
17. Seco F., Jiménez, AR., Zampella F. Joint estimation of indoor position and orientation from RF signal strength measurements. Paper presented at the International Conference on Indoor Positioning and Indoor Navigation 2013.
18. Shu X., Du Z., Chen, R. Research on mobile location service design based on Android. Paper presented at the 2009 5th International Conference on Wireless Communications, Networking and Mobile Computing 2009.
19. Yasir M., Ho SW., Vellambi BN.. Indoor position tracking using multiple optical receivers. Journal of Lightwave Technology, 2015; 34(4): 1166-1176.