Christ University Institutional Repository

ACTIVITY RECOGNITION USING MACHINE LEARNING TECHNIQUES FOR SMART HOME ASSISTED LIVING.

Kavitha,, R. (2018) ACTIVITY RECOGNITION USING MACHINE LEARNING TECHNIQUES FOR SMART HOME ASSISTED LIVING. PhD thesis, Christ university.

[img]
Preview
PDF (Title) - Published Version
151Kb
[img]
Preview
PDF (Declaration) - Published Version
82Kb
[img]
Preview
PDF (Certificate) - Published Version
82Kb
[img]
Preview
PDF (Acknowledgement) - Published Version
85Kb
[img]
Preview
PDF (Abstract) - Published Version
87Kb
[img]
Preview
PDF (Contents) - Published Version
83Kb
[img]
Preview
PDF (List of tables) - Published Version
83Kb
[img]
Preview
PDF (List of figures) - Published Version
90Kb
[img]
Preview
PDF (Abbreviations) - Published Version
87Kb
[img]PDF (Chapter 1) - Published Version
Restricted to Registered users only

231Kb
[img]PDF (Chapter 2) - Published Version
Restricted to Registered users only

403Kb
[img]PDF (Chapter 3) - Published Version
Restricted to Registered users only

2019Kb
[img]PDF (Chapter 4) - Published Version
Restricted to Registered users only

403Kb
[img]PDF (chapter 5) - Published Version
Restricted to Registered users only

359Kb
[img]PDF (Conclusion) - Published Version
Restricted to Registered users only

93Kb
[img]PDF (Bibliography) - Published Version
Restricted to Registered users only

174Kb
[img]PDF (Appendix) - Published Version
Restricted to Registered users only

125Kb

Abstract

The statistical survey by United Nations Department of Economic and Social Affairs/Population Division says, "globally the number of persons aged 60 and above is expected to be more than double by 2050 and more than triple by 2100". Especially in India, 9.5 percent of the population comprises of elders above 60 years. This may reach 22.2 percent in 2050 and 44.4 percent in 2100. On one side, the population of elders are gradually increasing and on the other side there is a challenge to take care of the wellbeing of the elders when they are living alone. Smart home assisted living system can address these problems. Smart Home Assisted living System is one among the growing research areas in smart computing. Advances in sensing, communication and ambient intelligence technologies created tremendous change in smart living environment. The development in technology made smart home to support elders, disabled persons and the needy person. Activity recognition is a growing technology in recent research and it plays a vital role in smart home assisted living system. Activity Recognition is a more dynamic, interesting, and challenging research topic in different areas like Ubiquitous Computing, Smart Home Assisted Living, Human Computer Interaction (HIC) etc. It provides solution to various real-time, human-oriented problems like elder care and health care. In order to address the issue on providing support on elder care this research proposes a machine learning based activity recognition model and an enhanced communication protocol for a smart home system, which are collaborated for designing the architecture of a smart home assisted living system. This system consists of three sub phases viz., data acquisition, monitoring system, and tracking system. For evaluating the feasibility of the proposed architecture, a real smart home environment is deployed with the help of ShiB010 kit provided by CASAS project, Washington State University, USA. A new segmentation technique viz., Area-Based segmentation is proposed and implemented for segmentation of activities detected during the tracking the activities of the resident of smart home. The accuracy of the proposed segmentation approach is measured using different performance measures. In order to assess the performance of the proposed segmentation technique, two well-known segmentation techniques, viz., Activity-Based Segmentation and TimeBased segmentation are also implemented. Performance of the proposed, Area-Based Segmentation approach is compared with the performance of the other two approaches. It is observed that the proposed segmentation technique demonstrates a commendable level of accuracy and its performance is comparable or even better than the other two techniques. To check the compatibility of proposed Area-Based Segmentation on other smart homes, three more data sets which are available in public domain are also used. The comparative results are explained in detail. As a part of data acquisition, smart home network communication protocol, transfer the sensed data to the monitoring system. An energy efficient communication protocol Ad-LEACH is proposed and simulated for energy efficiency in the smart home network. The result is compared with the base protocol LEACH.

Item Type:Thesis (PhD)
Subjects:Thesis
Thesis > Ph.D
Thesis > Ph.D > Computer Science
ID Code:7870
Deposited By:Shaiju M C
Deposited On:02 Jul 2019 12:00
Last Modified:02 Jul 2019 12:00

Repository Staff Only: item control page