Unsupervised Proximity Detection and Pairing for Smartphones

User context, especially physical proximity, is gaining its popularity as an intelligent indicator to trigger communication between devices. Apple’s iBeacon sends out notification to iPhones when customers enter their Apple stores. It detects proximity by ranging Bluetooth beacon signals emitted from anchor nodes that are preinstalled in the store. Many short-range communication methods adopted for proximity detection demand delicate physical conditions to be met to establish a connection. Prior works in the literature require infrastructural support, special hardware on device, or training phase to obtain a signal map or model whenever to be applied on a new venue, and therefore lack scalability. Although knowing precise coordinates of users does help determining their context, existing solutions demand considerable energy consumption on device (e.g., GPS) or expensive efforts for site survey as well as infrastructural support. In this paper, we concentrate on proximity detection through a practical and low-cost approach, compensating for room-level accuracy. We achieve wide applicability and scalability by solely leveraging Wi-Fi scan information that is easily accessible with low cost on commodity mobile devices.