Indoor proximity detection: The case study of a smart pet door

Abstract

The Internet of Things (IoT) era is revolving sensing and communication capabilities from industrial applications to daily life use cases. Some of them concern home automation toward the IoTs, in which smart objects help pets' owners to smartly monitor animals. These applications are made possible by low-power technologies enabling interactions based on proximity. This work demonstrates the suitability of both open source rapid prototyping solutions and Bluetooth Low Energy in proximity-based applicative scenarios. Here, received signal strength indication is enhanced with classification algorithms. By comparing neural networks and K-nearest neighbor solutions, the system effectively realizes secure access management and continuous monitoring. Classification precision reached the 80% in almost all cases and betters as distance decreases.