State Classification with Array Sensor Using Support Vector Machine for Wireless Monitoring Systems

Jihoon HONG  Tomoaki OHTSUKI  

IEICE TRANSACTIONS on Communications   Vol.E95-B   No.10   pp.3088-3095
Publication Date: 2012/10/01
Online ISSN: 1745-1345
DOI: 10.1587/transcom.E95.B.3088
Print ISSN: 0916-8516
Type of Manuscript: Special Section PAPER (Special Section on Medical Information Communication Technology for Disaster Recovery and Human Health Care Support)
wireless monitoring,  array antenna,  support vector machine,  

Full Text: PDF>>
Buy this Article

We have previously proposed an indoor monitoring and security system with an array sensor. The array sensor has some advantages, such as low privacy concern, easy installation with low cost, and wide detection range. Our study is different from the previously proposed classification method for array sensor, which uses a threshold to classify only two states for intrusion detection: nothing and something happening. This paper describes a novel state classification method based on array signal processing with a machine learning algorithm. The proposed method uses eigenvector and eigenvalue spanning the signal subspace as features, obtained from the array sensor, and assisted by multiclass support vector machines (SVMs) to classify various states of a human being or an object. The experimental results show that our proposed method can provide high classification accuracy and robustness, which is very useful for monitoring and surveillance applications.