Smart Objects and Technologies for Social Good. Third International Conference, GOODTECHS 2017, Pisa, Italy, November 29-30, 2017, Proceedings

Research Article

Statistical Features for Objects Localization with Passive RFID in Smart Homes

Download
154 downloads
  • @INPROCEEDINGS{10.1007/978-3-319-76111-4_3,
        author={Kevin Bouchard},
        title={Statistical Features for Objects Localization with Passive RFID in Smart Homes},
        proceedings={Smart Objects and Technologies for Social Good. Third International Conference, GOODTECHS 2017, Pisa, Italy, November 29-30, 2017, Proceedings},
        proceedings_a={GOODTECHS},
        year={2018},
        month={3},
        keywords={Smart environment Passive RFID Indoor localization Machine learning Data mining},
        doi={10.1007/978-3-319-76111-4_3}
    }
    
  • Kevin Bouchard
    Year: 2018
    Statistical Features for Objects Localization with Passive RFID in Smart Homes
    GOODTECHS
    Springer
    DOI: 10.1007/978-3-319-76111-4_3
Kevin Bouchard1,*
  • 1: Université du Québec à Chicoutimi
*Contact email: Kevin.bouchard@uqac.ca

Abstract

Smart homes offer considerable potential to facilitate aging at home and, therefore, to reduce healthcare costs, both in financial and human resources. To implement the smart home dream, an artificial intelligence has to be able to identify, in real-time, the ongoing activity of daily living with a fine-grained granularity. Despite the recent and ongoing improvements, the limitation of the literature on this subject primarily concerns the quality of the information which can be inferred from standard ubiquitous sensors in a smart home. Passive Radio-Frequency Identification is one of the technology that can help improving activity recognition through the tracking of the objects used by the resident in real-time. This paper builds upon the literature on objects tracking to propose a machine learning scheme exploiting statistical features to transform the signal strength into useful qualitative spatial information. The method has an overall accuracy of 95.98%, which is an improvement of 8.26% over previous work.