23 out 2020
14:00 Master's Defense Fully distance
Theme
A Comparative Study of Indoor Localization Methods based on RSS
Student
Tiago Moreira Trocoli da Cunha
Advisor / Teacher
Lucas Francisco Wanner
Brief summary
Indoor location is an important and comprehensive IoT application. In low-cost devices, methods that rely on Received Signal Strength (RSS) in radio packages to estimate the distance and position of devices are widely used, as they do not require additional hardware and provide reasonable accuracy . Although many RSS-based indoor location methods have been developed, comparative studies for these methods generally focus mainly on accuracy, ignoring important metrics such as response time and energy consumption. In devices with power restriction, in particular, accuracy cannot occur at the expense of reducing battery life and therefore maximizing accuracy alone may not be desirable. In this work, we present a comparative study of RSS-based localization techniques implemented on common hardware. We compared nine methods of location in a real-world office environment with strong signal interference and in rooms in a home with weak signal interference. In addition to accuracy, we present energy consumption and response time for each method. Our results showed that the commonly used lognormal distance estimation model performed poorly in terms of accuracy in these environments, but they had the lowest energy cost; that variations in the Fingerprinting method had more balanced and better values of energy consumption, response time and precision; and simpler implementations achieved better results than complex ones, overall. The experimental results showed that the precision was largely impacted by the position of the anchor nodes, corroborating the need for careful positioning of these devices. Finally, research has shown that the selective disposal of information from some anchor nodes can lead to better accuracy when compared to the use of all available nodes. However, selecting the optimal anchor subset for each estimate is a difficult task.
Examination Board
Headlines:
| Lucas Francisco Wanner | IC / UNICAMP |
| Gustavo Medeiros de Araujo | UFSC |
| Juliana Freitag Borin | IC / UNICAMP |
Substitutes:
| Leandro Aparecido Villas | IC / UNICAMP |
| Giovani Gracioli | UFSC |