This article presents the design, development, and implementation of a public lighting management system for the University of Campinas (Unicamp). The system integrates a communication infrastructure with a server application to enable real-time monitoring, control, and telemetry of light points across the university campus. Using technologies such as the Internet of Things, scalable NoSQL databases, and secure web interfaces, the platform facilitates energy efficiency, predictive maintenance, and enhanced operational reliability. In addition, the system incorporates a strategy for estimating the energy consumption of legacy lighting assets lacking communication capabilities, enabling the management of heterogeneous installations. The proposed solution complies with Brazilian regulatory frameworks and cybersecurity best practices, and its modular architecture supports seamless integration with commercial equipment. This approach demonstrates the potential for scaling to municipal public lighting systems, providing a model for smart city applications.
This paper presents a case study on the implementation of an open-source LoRaWAN network at the University of Campinas (Unicamp), Brazil, supporting educational and scientific Internet of Things (IoT) applications. The architecture is based on the BitDogLab platform with the RP2350 microcontroller (RISC-V) and a modular wireless board featuring an ESP32-C3 and RFM95W LoRa transceiver. All components follow open hardware and software principles to promote scalability, reproducibility, and hands-on learning. Solar-powered sensor nodes transmit environmental data every 15 seconds to a gateway connected to The Things Network (TTN), with real-time visualization via a ThingsBoard CE dashboard. Field tests demonstrated reliable communication using Spreading Factor 7 (SF7), with observed SNRs as low as -7 dB and RSSIs down to -114 dBm, confirming link quality under challenging conditions. The system achieved packet delivery ratios up to 79% and operated autonomously, consuming about 1 W per node. With a production cost of approximately USD 20 per unit (for batches over 200), the infrastructure qualifies as low-cost and offers a replicable model for open-source IoT deployments in academic and experimental settings.
Forest preservation faces significant threats from illegal logging and fires, demanding effective real-time monitoring solutions in remote environments. Conventional approaches often struggle with heavy data transmission burdens or compromise model performance due to limited edge device resources. This paper proposes a low-cost, real-time forest acoustic monitoring system that leverages on-device machine learning for efficient event detection. Our solution employs a Convolutional Neural Network (CNN) with the YAMNet model as a feature extractor, fine-tuned on a curated subset of the ESC-50 dataset to classify six critical forest-related sounds, including 'chainsaw', 'footsteps', 'crackling fire', 'rain', 'engine', and 'hand saw'.The system acquires audio segments, processes them in 1-second windows, and classifies acoustic events directly on a Raspberry Pi 3. To ensure reliable data transmission in challenging remote conditions, the classified event outputs are encoded into compact messages and transmitted via LoRaWAN. The fine-tuned model achieved promising performance with an 88.3% accuracy, 88.5% precision, 88.3% recall, and an F1-score of 87.9% on the test set. This approach demonstrates the feasibility of deploying high-level audio classification on resource-constrained edge platforms, enabling real-time alerting and enhancing forest preservation efforts in real-world scenarios.
This paper proposes a novel architecture for agricultural monitoring that integrates drone-mounted airborne LiDAR (ALS) systems with long-range wireless communication over TV White Space (TVWS) links. The proposed system addresses connectivity challenges in rural areas by combining high-resolution 3D terrain and crop data acquisition with cost-effective, scalable broadband connectivity. Key components include a LiDAR module with onboard processing, a remote base station for data pre-processing and transmission, and a cognitive radiobased TVWS transceiver for reliable backhaul. An overview of LiDAR technology, including state-of-the-art ALS systems, is also presented to contextualize the application.
Public Safety Diving is challenged by environmental factors like water depth, pollution, and temperature, which jeopardize diver safety and operational effectiveness. This study introduces a low-cost Internet of Things (IoT) prototype, integrated with an Unmanned Surface Vehicle (USV), to mitigate these risks through real-time water temperature monitoring. The system, utilizing an RC boat with an ESP8266 microcontroller and DS18B20 sensor, demonstrated functional success in a controlled environment, providing stable temperature and location data. Key findings include an average temperature of 22.67150 degrees C (+/- 0.00001 degrees C) over 7.5 hours of operation, significant battery autonomy, and a total cost of approximately R$258.00 (approx. USD 46.57). This technology offers a promising solution for improving diver safety and optimizing resource allocation in underwater public safety missions, with potential for future sensor expansions.
