March 23, 2026
Energy efficiency in lighting via IoT and sensors
Adriano Caversan; Diogo Alfieri Palma
Summary prepared by the ResumeAI tool, an artificial intelligence solution developed by Instituto Pecege focused on synthesis and writing.
The contemporary energy scenario demands technological solutions that integrate efficiency and automation, especially given the progressive increase in electricity consumption. According to data from the statistical yearbook of the Energy Research Company (2024), electricity consumption in the Brazilian residential sector jumped from 23.64% in 2004 to 31.49% in 2024. Within this amount, lighting represents a significant portion, corresponding to approximately 5.81% of the total household consumption. Faced with this reality, social awareness about energy saving becomes fundamental, as reducing individual consumption alleviates system overload, decreases the need for investments in heavy infrastructure, and results in direct savings for the end consumer. One of the most effective strategies to promote this saving is the provision of lighting strictly linked to real demand, regulating luminous flux according to the presence of individuals and the availability of natural light.
The Internet of Things emerges as the technological pillar for enabling this intelligence in everyday devices. The application of concepts such as Selective and Dynamic Lumen Output allows for automated adjustment of light intensity. As pointed out by Pinto (2016), this technique enables flux regulation that can generate energy savings of up to 40%, ensuring that lamps do not remain on at maximum power unnecessarily. Beyond the economic aspect, maintaining adequate lighting levels is a regulatory and ergonomic requirement. The Brazilian Association of Technical Standards establishes rigorous guidelines, such as NBR 5413 and NBR 8995-1, which define parameters for visual comfort and occupational health. Non-compliance with these standards in workplaces can even constitute a criminal offense according to the precepts of the Consolidation of Labor Laws, which in its article 175 requires lighting appropriate to the nature of the activity. Therefore, the development of autonomous systems that control lighting on demand not only serves a sustainability purpose but also legal compliance and human well-being.
The theoretical foundation of the project was based on the integration of hardware and software to create an intelligent environment. The use of low-cost microcontrollers, such as Arduino, allows for the prototyping of complex systems that capture electrical signals from sensors and transform them into logical actions. In the context of lighting, the system must be capable of interpreting ambient luminosity and the presence of people to decide the voltage level to be applied to the lamp. This decision-making process is what characterizes the device’s intelligence. Communicating this data to the cloud, using protocols such as Message Queuing Telemetry Transport, enables remote monitoring and historical consumption analysis, allowing for fine-tuning of the control algorithm to maximize energy efficiency.
For the project’s execution, the Design Thinking methodology was adopted, structured in five fundamental stages that guided the process from understanding the problem to validating the solution. In the empathy phase, the aim was to understand the needs for economy, ergonomics, and sustainability in the residential context. In the definition stage, the central problem was consolidated: the significant increase in electricity consumption in the last decade and the inefficiency of conventional lighting systems. Ideation resulted in the proposal of an Internet of Things-based system for automating these environments. Prototyping consisted of building a functional apparatus capable of collecting experimental data in real-time. Finally, evaluation involved rigorous analysis of the results obtained through the sensors, comparing the performance of the intelligent system with that of a traditional installation.
The experimental apparatus was meticulously constructed in two distinct fronts: a control module and a research module. The control module consisted of an Arduino Uno microcontroller, a light sensor to measure natural light, a passive infrared presence sensor to detect human movement, and an electronic dimmer responsible for regulating the intensity of dimmable LED lamps. The central objective of this module was to automate lighting to achieve the ideal lumen level without waste. In parallel, the research module was electrically isolated to ensure that its own consumption did not interfere with the lamp’s measurements. This second module included a voltmeter, an ammeter, a real-time clock module for precise timing, and a wireless communication module for sending data to the cloud.
The technical operation of the system begins with capturing the intensity of the electrical signal emitted by the sensors through the Arduino’s analog and digital ports. This signal is processed by the firmware developed in C++ language, which acts as a programmable logic circuit. The luminosity sensor provides a value that determines the need for artificial supplementation, while the presence sensor acts as a safety trigger, blocking the system if the environment is empty. Consumption, voltage, and current data are packaged and sent via the Message Queuing Telemetry Transport protocol to a cloud server. For information persistence, a Linux instance was used, hosting a Python script that consumes messages from the broker and stores them in a non-relational Firebase database. This architecture allowed for continuous tracking of the system’s operating profile under real domestic use conditions.
The component details reveal the complexity of the integration. The dimmer operates by controlling the phase of the alternating current, allowing the power delivered to the lamp to be adjusted from zero to 100%. The presence sensor has an internal timer and sensitivity adjustment, ensuring that the system does not suffer abrupt interruptions due to short periods of user immobility. The luminosity sensor, based on a light-dependent resistor, was calibrated to identify different brightness ranges, from total darkness to direct sunlight. On the software side, the firmware was programmed to take readings at millisecond intervals, applying moving average filters to avoid unwanted fluctuations in lamp intensity caused by sensor noise.
Data collection was carried out in two distinct scenarios for comparative purposes. In the first scenario, the system operated autonomously with all sensors and the dimmer active. In the second scenario, the same lamp was connected directly to the electrical grid, operating as a conventional lamp controlled by a manual switch. The experiment was repeated for several consecutive days, between May 16 and 29, 2025, to validate the consistency of the obtained data. During the testing period, the system faced real climatic variations, including sunny days and rainy afternoons with low natural light, which provided a rich dataset for performance analysis under different environmental conditions.
The results obtained revealed that the average electricity savings were 28.6% when comparing the intelligent system with the conventional one. However, a detailed analysis by periods of the day showed distinct behaviors in the apparatus’s efficiency. During the night, the efficiency was significantly higher, achieving savings of 50.1%. This is because, in the absence of natural light, the system can optimize power delivery according to the specific needs of the environment and the detected presence. In contrast, during the morning and afternoon periods, the results were unexpected. In the morning, the system consumed 112.8% more energy than the lamp turned off, and in the afternoon, the savings were only 25.1% in some measurements, even showing higher consumption on very bright days.
The technical analysis of this phenomenon identified that the dimmer and the microcontroller have a constant residual consumption, even when the lamp is virtually off or operating at 0% of its power. This maintenance consumption, estimated at approximately 0.1 W, although seemingly insignificant in isolation, becomes relevant when accumulated over hours when artificial lighting would not be necessary. In a conventional system, the physical switch completely cuts off the current flow, resulting in zero consumption. In the tested prototype, the control electronics remained energized to ensure system readiness, which ended up generating excess consumption during periods of high natural light availability.
To refine the analysis, a more consistent time series was selected, discarding days with atypical behaviors or outside the standard deviation. In this adjusted series, it was observed that in the morning period, the system consumed 296.6% more energy than the traditional model, while in the afternoon, consumption was 51.7% higher. Real and effective savings were concentrated in the nighttime period, with a 43.4% reduction in consumption. When consolidating all periods of this adjusted series, the final average savings were only 2.7%. This value is considerably below the initial expectation of 40% suggested by technical literature. This discrepancy highlights the importance of considering the idle consumption of automation components in the overall energy efficiency calculation.
The discussion of the results indicates that, although the dimming logic by presence and luminosity is effective in reducing the lamp’s consumption itself, the support infrastructure required for this intelligence may compromise energy savings if not adequately designed. The system proved to be extremely useful in environments where usage frequency is irregular and where there is a need to maintain minimum lighting levels for safety or comfort. However, for the apparatus to reach its maximum savings potential, improvements in the hardware architecture are necessary. One proposed solution is the inclusion of an electromechanical or solid-state relay that completely cuts off power to the dimmer and the lamp when the luminosity sensor detects sufficient natural light or when the presence sensor indicates that the environment has been empty for a long period.
In addition to technical factors, climatic conditions directly influenced the data. On rainy days, the low incidence of natural light forced the system to operate at higher power during the afternoon, which increased consumption compared to clear days. This demonstrates that the efficiency of intelligent lighting systems is dependent on the environment’s architecture and its geographical location. Environments with large openings for natural lighting tend to favor savings, as they reduce the activation time of the artificial system. On the other hand, in confined spaces or those with poor light ventilation, the automation system becomes even more vital to avoid wasting energy with lamps on at maximum power unnecessarily.
The experience with the prototype also allowed for an evaluation of the robustness of communication via the Internet of Things. The Message Queuing Telemetry Transport protocol proved to be extremely efficient for sending small data messages with low bandwidth consumption, which is ideal for home automation devices. Integration with the Firebase database facilitated real-time data visualization and export to statistical analysis tools. The use of Python scripts for cloud data processing proved to be a scalable solution, allowing multiple sensors to be monitored simultaneously without performance loss.
The analysis of the study’s limitations indicates that the testing period, although sufficient to validate the concept, could be expanded to cover different seasons of the year. Variations in solar inclination and day length throughout the year would certainly bring new nuances to the consumption data. Furthermore, the use of different types of LED lamps could reveal variations in response to the dimmer, as not all LED drivers have the same efficiency curve at low power. Sensor sensitivity is also a point of attention; the infrared presence sensor can suffer interference from heat sources, while the luminosity sensor can be affected by momentary shadows or internal reflections.
Despite the challenges encountered, the implementation of the Internet of Things system for lighting control represents a significant advancement in the pursuit of smarter and more sustainable homes. The ability to collect, process, and act on environmental data in real-time opens doors for much more precise energy management than traditional manual methods. The study showed that artificial intelligence applied to hardware should not only focus on the main task, which is lighting, but also on managing its own energy consumption. Firmware optimization to enter low-power modes (sleep mode) and the use of components with lower leakage current are essential paths for future iterations of this project.
The integration between sensors and actuators mediated by accessible microcontrollers democratizes access to home automation. What was once restricted to expensive and complex proprietary systems can now be developed with open-source tools and off-the-shelf components. This accessibility allows energy-saving solutions to be customized for the specific needs of each user or environment, enhancing results. The discussion on energy efficiency should, therefore, go hand in hand with the evolution of software and hardware engineering, ensuring that technology is an ally in preserving natural resources and reducing operational costs.
It is concluded that the objective was achieved, as the prototype was built, tested, and the statistical data were collected and analyzed rigorously. The average savings of 28.6% demonstrate the system’s viability, although detailed analysis revealed that the idle consumption of electronic components can drastically reduce efficiency during periods of high natural light. The identification of this technical limitation points to the need to implement total power cut mechanisms, such as automatic relays, to avoid residual consumption by the dimmer. The study confirmed that the integration of presence and luminosity sensors via the Internet of Things is a powerful tool for energy management, but it requires a hardware design that minimizes the consumption of the control infrastructure itself for the economic and environmental benefits to be fully realized.
Bibliographic References:
Associação Brasileira de Normas Técnicas [ABNT]. Disponível em: <https://abnt.org.br/>. Acesso em: 31 mar. 2025
BRASIL, lei nº 6.514, de 22 de dezembro de 1977 da Consolidação das Leis do Trabalho. Disponível em: <https://www.planalto.gov.br/ccivil_03/leis/l6514.htm>. Acesso em: 31 mar. 2025
Empresa de Pesquisa Energética, EPE – Anuário Estatístico de Energia Elétrica. Disponível em: <https://www.epe.gov.br/pt/publicacoes-dados-abertos/publicacoes/anuario-estatistico-de-energia-eletrica>. Acesso em: 02 abr. 2025
Pinto, D. M. de O. 2016 – Sistemas de Controlo de Iluminação Pública – Universidade de Coimbra. Disponível em: <https://hdl.handle.net/10316/40438>. Acesso em: 31 mar. 2025
Executive summary from the Final Project of the Specialization in Software Engineering of the MBA USP/Esalq
Learn more about the course; click here:
