IOT BASED PROXIMITY TRACING CLASSIFICATION

Abstract

Since 2020, several contact tracing applications have been developed and implemented in response to the COVId-19 pandemic. This study thoroughly examines the popular contact tracing applications and highlights their security, privacy and battery-related issues. In response, the study proposes a proximity tracing solution integrated with state-of-the-art estimation and classification techniques that employ the BLE communication protocol. The first of its kind solution utilizes localization and Machine Learning classification techniques to effectively estimate the proximity between multiple users. Meanwhile, the solution significantly improves user privacy by eliminating the necessity to track the users’ locations. Furthermore, the proposed solution does not store any personal information and transfers the computational load from the users’ mobile devices to installed beacons, extending the devices’ battery life. The solution offers estimation accuracy of >97% with a 97% F1 score using localization techniques and an 89% F1 score using ML classification.

Date of Award	2021
Original language	American English
Awarding Institution	HBKU College of Science and Engineering