Enhancing Agricultural Surveillance: An Edge-A and LoRa-Based Vision Mote System for Infrastructure-Deficient Regions

Abstract

Remote areas often lack access to reliable power, internet, and surveillance infrastructure, making them vulnerable to threats such as illegal intrusion, poaching, and environmental risks. To address these challenges, the propose a self-sufficient, edge-AI-based surveillance system capable of real-time monitoring, detection, and alerting without relying on cloud connectivity. The system deploys Vision Surveillance Motes equipped with cameras, motion sensors, and acoustic inputs, and uses lightweight artificial intelligence models (MobileNet-SSD for vision and support vector machines for sound) processed locally on Raspberry Pi boards. Long-range wireless communication is enabled via LoRa (Long Range) modules, transmitting alerts to a Control Room Mote that displays data using a human-machine interface (HMI) and pushes updates to a cloud server for optional remote access. This multimodal architecture allows the system to operate in completely offline environments, with optional cloud integration for centralized visibility. The solution is field-tested and optimized for deployment in forests, disaster-prone zones, border areas, and rural locations requiring independent surveillance.