LoRa-based Privacy-Aware and Contactless Surveillance in Next-generation Smart Homes

Abstract

There are many security and surveillance applications in which camera-based or invasive human activity recognition is not permissible due to privacy, insecurity, and discomfort issues. Moreover, vision-based surveillance systems have a few key technical challenges that limit their practicality such as bad illumination, obstacles, and occlusion. In such cases, we require a more reliable system which can work in challenging situations such as in darkness, long-range, through walls or obstacles, a rainy and smoky environment where vision-based systems do not deliver good results. Only high-energy and long-range wireless sensors which can identify targets through obstacles, walls, worst climate or surrounding conditions can serve the purpose. In this paper, we propose to use the LoRa transceiver to identify different persons from their walking patterns through bricks and a concrete wall where people other than the target person are moving freely. Moreover, we employed three different deep-learning models to evaluate the efficacy of the proposed system. Our highly accurate 99% results encourage the use of contactless LoRa sensors in many privacy-critical applications such as healthcare, security, and surveillance, where short-range and low-energy signals have limited utilisation.