RETSINA: Reproducibility and Experimentation Testbed for Signal-Strength Indoor Near Analysis
DocUID: 2023-003Author: Anna Baskin, Brian T. Nixon, Panos K. Chrysanthis, Christos Laoudias, Constantinos Costa
Abstract: Reproducibility is a core component of any scientific discovery. A step towards reproducibility within the IPIN community is the contribution of this paper, our software-based testbed, called RETSINA (Reproducibility and Experimentation Testbed for Signal-strength Indoor Near Analysis). RETSINA enables the repeatability, reproducibility and comparison of approaches that use machine learning to detect proximity. We demonstrate RETSINA's functionality by repeating and extending the findings of a recent case study on Wi-Fi signal strength based contact tracing accuracy. Furthermore, we leverage RETSINA to experimentally compare the results for detecting close encounters produced by the original Wi-Fi signal strength readings study and our study using Bluetooth signal strength readings.
Keywords: Reproducibility, Repeatability, Contact Tracing, Proximity Detection, Indoor Localization, Machine Learning
Published In: 13th International Conference on Indoor Positioning and Indoor Navigation (IPIN)
ISBN: 979-8-3503-2011-4
Pages: 1-6
Place Published: Nuremberg, Germany
Year Published: 2023
DOI: 10.1109/IPIN57070.2023.10332500
Project: RETSINA Subject Area: Reproducibility, Machine Learning, IoT
Publication Type: Conference Paper
Sponsor: NIH R01HL159805, NSF SES-2017614, KIOS CoE 739551