A Novel Method for Inferring Person-to-person Relationship Using Wi-Fi

Abstract

Present communities are encompassed in a consistently developing telecommunication framework. This framework provides significant circumstances for detecting recording on abundance of human actions. Human adaptability patterns are distinguished example of such a conduct which has been considered deliberately grounded on Wi-Fi systems and Bluetooth signals as intermediaries for different areas. While versatility is a significant characteristic of human action, it's very pivotal to research and analyze physical interrelationship among humans. Detecting closeness that empowers social associations on a substantial scale is a practical challenge. Numerous frequently used techniques containing RFID badges and Bluetooth filtering provide only restricted scalability. This research has been conducted based on the idea to deduce the kind of dyadic relationships (friendship) between research candidates from the list of WLAN MAC addresses and detected Bluetooth devices estimated by cellular phones conveyed by the two individuals. We demonstrate our methodologies using MIT's Social Evolution (WLAN + Bluetooth) dataset. The originality of our results is demonstrated by the comparison of outcome of our analysis with the self-investigated surveys subjects issued with regard to their link and connection. Proposed methods for inferring type of dyadic relationships using WLAN dataset gives higher accuracy and F1-Score than using Bluetooth dataset as WLAN can be used for high-resolution mobility tracking of entire populations. Our results exhibit the estimation of WLAN MAC addresses as a tool for social detection and reveal how numbers of Wi-Fi information represent a potential risk to privacy.