1. Introduction
Agriculture is the set of human practices aimed at producing food, forage, fiber and other useful products through the management of plants and animals [
1]. In a global context where the population exceeds 8 billion people and is projected to increase by another 2 billion by 2050, the United Nations [
2] estimates that food production will need to increase by 70%. This challenge requires a transformation in production and consumption methods, prioritizing sustainable agricultural practices that balance food demand with environmental conservation.
Protected agriculture refers to the action of placing structures to protect crops, with the aim of obtaining a higher-quality harvest in less time or out of season [
3]. Various techniques are used in this type of agriculture, including hydroponics.
The Nutrient Film Technique (NFT) is a soilless hydroponic cultivation method that uses inclined structures through which a thin and continuous flow of nutrient solution circulates [
4]. This technique ensures that plant roots are consistently exposed to both nutrients and oxygen, promoting efficient absorption [
5]. NFT is notable for its high water and nutrient use efficiency compared to other hydroponic techniques [
6], as well as its adaptability to controlled environments, enabling year-round crop production [
7]. Several studies have shown that this technique can accelerate plant growth by 10% to 50% compared to traditional agricultural practices [
8]. However, NFT presents limitations, such as vulnerability to power outages, unsuitability for long-cycle crops (longer than 50 days) and a high risk of rapid spread of root diseases [
9].
Indoor agriculture, also known as indoor farming, takes place in spaces with regulated environmental conditions, where factors like air temperature can be adjusted. Sunlight is replaced by artificial illumination systems. A controlled environment with artificial light is called a plant factory, growth chamber or plant production unit [
10]. The advantages of these systems are clear, as technological advancements allow crops to be produced year-round without relying on climate patterns [
11]. Currently, the most common crops in these systems include leafy vegetables (lettuce, spinach, etc.), tomatoes, cucumbers, berries and microgreens [
12].
On the other hand, the Internet of Things (IoT) is a revolutionary technology that enables the interconnections between physical objects and the Internet, establishing continuous interaction between them [
13]. In agriculture, the IoT refers to the integration of sensors, embedded devices, communication networks and digital platforms that allow for real-time data collection, analysis and monitoring. This enables automated control of various processes such as irrigation and fertilization, as well as data-driven decision-making, allowing for faster and more effective responses to adverse events, reducing resource waste and optimizing productivity [
14]. However, the impact of IoT depends on proper implementation, technical feasibility and financial accessibility for end users [
15].
Smart agriculture is based on the application of modern technologies to improve both the quantity and quality of the crops. This approach includes the use of tools such as IoT, databases, machine learning, cloud processing, machine-to-machine communication protocols, sensors, actuators and personal devices like phones and computers [
16].
In [
17], a smart agriculture framework is described as consisting of six layers. The first layer, or physical layer, contains a set of sensors, actuators and microcontrollers that collect and process data among the devices themselves. The second layer, or network layer, includes the Internet, Wi-Fi connections and wireless communication between devices. The third layer, or intermediate layer, is responsible for device management on the network, platform portability, security and authentication. The fourth layer, or service layer, provides cloud storage and software as a service for data management. The fifth layer, or analytics layer, utilizes various algorithms to process real-time data, enabling predictive analysis of various crop aspects. Finally, the sixth layer, or user experience layer, is where farmers use applications, such as social media, to share and disseminate agricultural knowledge.
Next, a review of recent studies (2020–2025) on the development and implementation of smart agricultural systems focused on small-scaled hydroponics is presented. The research explores significant advancements in technologies such as IoT and artificial intelligence (AI) applied to agriculture.
In [
18], a small-scale IoT-based aeroponic system (90 cm × 60 cm × 40 cm) is proposed for cultivating water spinach (
Ipomoea aquatica Forssk). In the physical layer of the system, a DHT-11 sensor is used to measure air temperature and relative humidity (RH) in the root chamber and control mechanisms including a Peltier cell for cooling, fans and mist generators. A Wemos D1 Mini microcontroller makes decisions based on internal air temperature and RH values. The network layer utilizes the microcontroller’s built-in Wi-Fi connectivity to transmit real-time data to the service layer, where the ThingSpeak platform [
19] is used for visualization and recording of the information. The user experience layer is implemented via QR codes, enabling manual activation and deactivation of control mechanisms. The system was tested for 16 days, demonstrating stable environmental conditions in the root chamber.
In [
20], an IoT aquaponics system architecture is proposed to monitor environmental variables for both plants and fish. For the physical layer, control mechanisms such as water pumps are used along with various sensors to monitor pH, electrical conductivity (EC), water level, water temperature,
CO2 and ammonia. A NodeMCU microcontroller (Makerfabs, Shenzhen, China) handles decision-making and also functions as a gateway, transmitting data via Message Queuing Telemetry Transport (MQTT) to the service layer. The user experience layer is implemented through a web application for remote monitoring.
In [
21], the design and development of an automated hydroponics system is presented. In the physical layer, multiple sensors are integrated to measure variables such as air temperature, RH, substrate moisture and water level in tanks. Various control mechanisms are also employed, such as Light Emitting Diode (LED) lamps, a refrigeration system, ventilation equipment and pumping devices for the nutrient solution. An ATmega microcontroller is used for decision-making.
The information collected by the monitoring system is sent via Zigbee protocol based on IEEE 802.15.4 [
22] to the service layer for storage. In the user experience layer, the operator can visualize and control the connected devices through an interface developed in Laboratory Virtual Instrument Engineering Workbench (LabVIEW).
In [
23], the development and implementation of a smart hydroponic system were carried out. The physical layer consists of several sensors for measuring environmental variables such as air temperature and light intensity, as well as nutrient solution variables such as pH, nitrogen and phosphorus and for capturing plant images. Control mechanisms are included for irrigation and lighting systems. A Raspberry Pi (Raspberry Pi Foundation, Cambridge, United Kingdom) serves as the decision-making central control unit and transmits data to the network layer. In the analysis layer, artificial intelligence is implemented using a deep learning convolutional neural network (DLCNN) prediction model for data analysis. These models estimate the nutrient values required by the plants based on reference values and environmental conditions. Additionally, in the user experience layer, a DLCNN classification system is used to send alerts via a mobile application.
In [
24], a smart monitoring and control system for hydroponics applied to lettuce cultivation was designed. The physical layer consists of various sensors capable of measuring variables such as air temperature, water level, pH and total dissolved solids (TDS) in the water. It also includes control mechanisms such as pumps for water, nutrients and saline solutions to maintain the stability of the nutrient solution. An ESP32 microcontroller (Espressif Systems, Shanghai, China) is used for data collection, decision-making and sending information to the cloud. The service layer is implemented using the Blynk platform. The user experience layer is applied through a mobile application for user data visualization.
In [
25], a vertical indoor aeroponic system architecture is proposed. The physical layer consists of various sensors for monitoring variables such as pH, EC, water tank level, air temperature, RH and photosynthetically active radiation. The system uses a Raspberry Pi as the central control unit for decision-making, managing the activation of three peristaltic pumps for precise nutrient management using feedback control and two LED lamps to provide adequate lighting. The collected data is sent to an MQTT broker. The service layer generates alerts about the system status, which are displayed to the operator through the user experience layer.
In [
26], an IoT hydroponic system was developed with the dimensions of 150 cm × 150 cm × 235 cm. For the physical layer, sensors were integrated for the measurement of environmental variables such as air temperature and RH, as well as nutrient solution variables such as pH, water temperature and EC. They include various control mechanisms such as fans, full-spectrum LED lamps, a heating system and pumps for water and nutrient circulation. A NIDO ONE central controller [
27] is used for data collection, decision-making and communication with the database. The collected information is processed and displayed to the user via the NIDO PRO platform.
In [
28], a hydroponic system with a nano bubble generator is proposed. The system’s physical layer includes sensors capable of measuring variables in nutrient solutions such as Dissolved Oxygen (DO), TDS and pH; similarly, air temperature and RH are monitored. The only control mechanism in the system is the nano bubble generator. Multiple microcontrollers (Arduino UNO, NodeMCU, ESP32-E) are used to gather sensor data. A Raspberry Pi is employed as a local server and for decision-making processes. In the service layer, the data is sent to the Antares platform. The user experience layer is supported through both a web application for real-time monitoring and a local monitor serving the same purpose.
In [
29], an automated IoT-based hydroponic system for indoor cultivation was developed. The system’s physical layer was implemented using a series of sensors to measure environmental variables such as air temperature and RH, as well as nutrient solution variables such as pH and TDS. For automation, pumps were integrated for nutrient control, along with white LED lights. An Arduino UNO microcontroller is responsible for collecting the data, which is transmitted via a Raspberry Pi to a platform built-in Node-RED [
30]. In the service layer, the data are stored in a MongoDB database and visualized through the user experience layer via a web application accessible from a computer or a mobile device connected to the same network.
In [
31], a Deep Water Culture (DWC) IoT hydroponic monitoring system is presented. The physical layer includes sensors for measuring variables such as pH and water temperature. A NodeMCU microcontroller [
32] is used to collect data, perform decision-making tasks and transmit information to the cloud. In the service layer, the Firebase platform [
33] is used to store the data. The analysis layer is implemented through a Mamdani fuzzy logic algorithm [
34] to determine whether the system conditions are within normal limits or out of range. In the user experience layer, the users receive alerts generated by the fuzzy logic algorithm via WhatsApp and view the information through a web application. The system does not include control mechanisms, as it is limited to monitoring only.
Table 1 summarizes the main characteristics of the systems presented in the studies described above. The second column describes the cultivation technique used, the third column shows the type of crop used for system validation and columns four to six indicate the type of microcontroller (MCU) used, the variables that were monitored and the control mechanisms implemented. Finally, the seventh column indicates how the end user visualizes the monitored variables.
In the present work, the design and implementation of a smart agriculture system is proposed, which is capable of perceiving its environment, processing and transmitting the collected information to the cloud and acting accordingly [
13,
17]. The system is installed in a plant growth chamber designated for hydroponic lettuce cultivation using NFT. This proposal presents an integrated solution that monitors and controls environmental conditions within the chamber, incorporates video surveillance and enables remote data visualization in the cloud, all within a closed and controlled infrastructure.
The proposed system is capable of monitoring environmental variables such as air temperature and RH, carbon dioxide concentration (CO2) and photosynthetically active photon flux (PPF). These variables are fundamental for the control of agricultural microclimates in closed environments. The system also includes control mechanisms capable of manipulating these variables and managing the pumping system of the hydroponic cultivation system. All the information collected from the monitoring and control of these variables is transmitted to the cloud and made available to users for real-time visualization through a web application. Additionally, the system includes an automatic email alert mechanism—indicating a possible system failure.
Through a case study, the proposed smart agriculture system was validated through the cultivation of lettuce using three different nutrient solution concentrations over a complete growth cycle. The experiment allowed monitoring the performance of the proposed system in terms of environmental control and crop tracking; however, conclusions regarding nutritional efficiency should be limited solely to the experiment carried out in this case study. This proposal presents an alternative approach to urban plant production that aligns with the automation of agricultural processes.
With the features demonstrated by the proposed smart system, this article contributes to the field of smart agriculture, as listed below:
The smart agriculture system presented is not limited to the sensors described here, as the use of the RS-485 communication bus allows for the connection of up to 31 additional slave devices on a single channel. This feature provides scalability and flexibility, facilitating the integration of new sensors or actuators based on the specific needs of the crop or future research.
A responsive web application and a mobile application were developed for monitoring environmental variables and modifying their reference values (setpoints). Our system represents an improvement compared to [
21,
25,
26,
28,
29], where data transmission and visualization are performed locally. The development of a customized web application allows for a user interface that is adaptable and tailored to the information displayed, unlike [
24,
26], which rely on Platform-as-a-Service (PaaS) solutions with predefined visualizations.
This article also presents a case study that describes the cultivation experiment used to validate the system, in which three treatments with different nutrient solution concentrations (T100, T75 and T50) were applied. An analysis of variance (ANOVA) is presented to compare the phenological and biochemical results obtained. Additionally, a comparison with international standards is provided, demonstrating that the biochemical characteristics of the harvested lettuce are competitive.
The article is organized as follows:
Section 2, Materials and Methods, describes the methodology used to condition the space designated as the plant growth chamber, the characteristics of the NFT hydroponic systems, the cultivation experiment conducted in the case study and the proposed smart agriculture system.
Section 3, Results, presents the findings related to the behavior of the smart agriculture system and the variables recorded during the cultivation experiment of the case study.
Section 4, Discussion, analyzes the results obtained from the proposed system and the case study used for its validation. Finally,
Section 5, Conclusions, outlines the general conclusions and the future work.
2. Materials and Methods
To evaluate the behavior of hydroponic crops under controlled conditions using smart technologies, a smart agriculture system was designed and developed and installed in a dedicated plant growth chamber. This section provides a detailed description of the infrastructure, implemented equipment and the architecture of the IoT-based smart agriculture system and the experimental methodology used for a case study conducted to verify the proper functioning of the proposed system using lettuce cultivation. It includes the specifications of the physical space, characteristics of the hydroponic system, technological integration for automation of the cultivation environment and the monitored variables considered during the experiment.
2.1. Growth Chamber
The space designated as the growth chamber, with dimensions of 3 m × 2 m × 2.5 m, is illustrated in the diagram shown in
Figure 1. An extended surface of 75 cm in depth was added to accommodate three NFT hydroponic systems used in the experiment. Additionally, the walls were coated with latex paint to prevent the development of fungi and mold. The chamber was equipped with an IoT-based monitoring and control system for environmental variables such as air temperature, RH,
CO2, PPF and photoperiod; capable of controlling up to 20 full-spectrum LED lamps, the pumps used in the NFT systems and up to six ultrasonic humidifiers; it also includes a Mirage X32 mini split air conditioning system for air temperature control, as well as a video monitoring system.
2.2. NFT Hydroponic System
The hydroponic system shown in
Figure 2 consists of two channels with a trapezoidal cross-section made of polyvinyl chloride (PVC), each featuring five planting cavities spaced 20 cm apart. The systems also include a frame constructed from chlorinated polyvinyl chloride (CPVC), designed with a 2% slope [
35], which is essential for maintaining a continuous flow and ensuring the proper operation of the Nutrient Film Technique (
Figure 2,E). Several components of this system were custom designed and manufactured using 3D printing technology, as suitable commercial alternatives were not available to meet the specific functional requirements. The dimensions of each system are 100 cm in length, 50 cm in width and 27 cm in height.
The NFT hydroponic system operates as follows: the nutrient solution flows from a 20 L tank (
Figure 2,F) through a 10 mm diameter hose (
Figure 2,D) and into a 12 V DC diaphragm water pump (model TOPINCN, 0.48 MPa, 3.5 L/min; TOPINCN, Shenzhen, China) and then into the hydroponic channels (
Figure 2,A), where the plant roots (
Figure 2,B) absorb the essential nutrients for growth while the remaining solution returns to the tank by gravity.
The proposed system was specifically designed for lettuce cultivation but also has the ability to adapt and be applied to the other vegetables such as spinach (Spinacia oleracea L), cabbage (Brassica oleracea var. Capitata) and aromatic plants like cilantro (Coriandrum sativum) due to their similar dimensions.
2.3. Monitoring and Control System
For the implementation of the monitoring and control system, based on the state of the art, continuous monitoring of critical environmental variables such as air temperature, RH,
CO2 and PPF was identified as a priority. These variables directly affect photosynthesis [
20,
21], making them a priority for the automation of the growth chamber.
The proposed smart monitoring and control system was developed based on the architecture presented by [
17,
36], which suggests a six-layer structure. However, this work simplifies the architecture into five layers: physical layer (
Figure 3A), network layer (
Figure 3B), middleware layer (
Figure 3C), service layer (
Figure 3D) and user experience layer (
Figure 3E). The analysis layer is not considered in the proposed architecture, as it is planned for future work.
The physical layer of the proposed system, shown in
Figure 4, utilizes a board based on the ATmega2560 microcontroller (Microchip Technology Inc., Chandler, AZ, USA) (
Figure 4I) to collect data every minute from sensors that monitor environmental conditions within the growth chamber, including air temperature, RH,
CO2 and PPF (
Figure 4H). Based on this data, decisions are made using an ON/OFF control scheme with hysteresis to regulate the activation of actuators. RH was stabilized using four ultrasonic humidifiers (model UH200, Bontill, Plano, Texas, USA), each with a 2.5 L capacity (
Figure 4D). Their activation is managed via a relay module (
Figure 4E), which also controls the pumps responsible for delivering the nutrient solution in the NFT system (
Figure 4B). Air temperature is maintained using a mini split air conditioner (model X32, Mirage, Monterrey, Mexico) (
Figure 4A). Additionally, a real-time clock (RTC) circuit (model DS1302, Maxim Integrated, San Jose, CA, USA) (
Figure 4F) was implemented to automate decisions based on the time of day, ensuring photoperiod regulation through full-spectrum LED lamps (model INWT80421265Ec, Barrina Inc., Livermore, CA, USA) (
Figure 4C). The program is executed automatically every minute.
The selection of the ATmega2560 microcontroller (Microchip Technology Inc., Chandler, AZ, USA) was based on its accessibility, ease of implementation and the extensive range of available documentation, which facilitates rapid development of embedded systems. Its compatibility with the MODBUS RS-485 communication protocol via the Universal Asynchronous Receiver-Transmitter (UART) module, along with the wide availability of reliable libraries, enabled the integration of various industrial sensors for environmental monitoring.
Although there are more advanced development boards such as the STM32, which offer greater processing capacity, modular architecture and support for real-time operating systems, these platforms require more specialized programming environments and are associated with higher costs.
Using the MODBUS RS-485 communication protocol, the microcontroller retrieves data from the sensors. This protocol enables the interconnection of up to 32 devices on a single bus segment, with a communication range of up to 1200 m under standard conditions. Agronomic sensors with an IP67 rating were selected, providing dust resistance and allowing them to function effectively under higher RH conditions compared to other types of sensors. The selected sensors for each variable, along with their measurement ranges, are detailed in
Table 2.
All sensors were used with their factory calibration, in accordance with the specifications provided by the manufacturer. Their proper functioning was verified by comparison with other measurement systems.
Programming logic enables automatic activation of control mechanisms, such as ultrasonic humidifiers and full-spectrum LED lamps, based on ambient conditions inside the growth chamber and predefined schedules.
As described in Listing 1, humidity control was implemented by reading the air temperature and RH sensor at one-minute intervals. The system compares the measured value with predefined maximum (hum_max) and minimum (hum_min) RH thresholds. When the RH inside the growth chamber drops below the minimum threshold, the humidifier activates; conversely, when the humidity exceeds the maximum threshold, the humidifier switches off.
Listing 1. Code snippet used to control the activation and deactivation of the humidifiers based on ambient RH. |
void pedirHumTemp() { if (node.readHoldingRegisters(0x0000, 2) == node.ku8MBSuccess) { Humedad = node.getResponseBuffer(0x00) / 10.0f; //RH Temp = node.getResponseBuffer(0x01) / 10.0f; //Air Temperature } } void control_humedad() { if (Humedad < hum_min) { digitalWrite(HUMID, LOW); // Activate humidifiers } else if (Humedad > hum_max) { digitalWrite(HUMID, HIGH); // Disactivate humidifiers }
} |
Figure 5 shows the behavior of the environmental variables—air temperature, RH,
CO2 and PPF—inside the plant growth chamber. The control over the lamps can be observed, as they are turned on at 2:00 P.M. and turned off at 4:00 A.M. the following day. Similarly, control actions to maintain air temperature and humidity within the established ranges are also evident. As a closed system capable of controlling its internal environmental variables, it offers the advantage of programming an artificial photoperiod that is offset from the natural light cycle. This allows the full-spectrum LED lamps to operate during non-working hours, providing greater convenience to the user in case access is required to perform adjustments or measurements. Furthermore, scheduling them at night represents an additional advantage, as in many regions, electricity rates are typically higher during daylight hours than at night [
37].
The network layer is implemented using an IoT module (G in
Figure 4), which transmits environmental measurements every five minutes via HyperText Transfer Protocol (HTTP) protocols to the Service Layer. In this layer, the ThingSpeak platform is used for storing and analyzing the collected data.
The user experience layer employs a responsive web application that displays real-time graphs of environmental variables (air temperature, RH, CO2 and PPF) within the growth chamber. Likewise, the use of a mobile application for the same purpose is of utmost importance for users’ accessibility to information. An alert system notifies users via email in the event of a failure that prevents data transmission to the web server for periods exceeding 30 min.
For video surveillance, a Wi-Fi video camera, model HDC10_wjmozu-gj (WJMOZU, Shenzhen, China), with an integrated microphone was used. This device is compatible with the Tuya Smart mobile application [
38], enabling users to view the crop status in real time, monitor the operation of full-spectrum LED lamps and, through the microphone, verify whether the pumps in the NFT hydroponic systems are functioning properly.
On the other hand, in order to obtain a precise estimate of the electrical energy consumption of the main devices in the proposed system, measurements were conducted using PZEM-004T v3.0 energy sensors (Peacefair, Shenzhen, China) during 24 h of continuous system operation. These sensors were temporarily connected to the main energy-consuming devices: the mini split air conditioner, full-spectrum LED lamps, recirculation pumps and ultrasonic humidifiers. This setup allowed us to record energy consumption, facilitating an evaluation of the system’s energy cost.
2.4. Case Study: Validation of the Proposed System Through Lettuce Cultivation
To validate the correct operation of the proposed system, a case study was conducted under controlled conditions using the NFT hydroponic technique.
The hydroponic lettuce experiment was conducted from 3 June (transplanting day) to 29 June 2024 (harvest day), in a controlled-environment growth chamber located at the Tecnológico Nacional de México in Celaya with the coordinates 20°32′15.422″ N, 100°48′54.604″ O. The experiment was divided into five stages, as illustrated in
Figure 6.
During the germination stage (
Figure 6A), lettuce seeds (
Lactuca sativa cultivar ‘Rodhenas’) were placed in hydroponic net pots filled with peat moss and germinated for two days. In the root growth stage (
Figure 6B), seedlings remained for 28 days in a greenhouse until they reached a height of approximately 8 cm. For the transplanting into the hydroponic systems (
Figure 6C), seedlings were inserted into foam pads with their roots protruding from the bottom and placed inside hydroponic net pots. The seedlings were distributed into treatment groups of 10 plants each. During the vegetative growth stage (
Figure 6D), the plants remained in the growth chamber for the following 26 days. Finally, during the harvest stage (
Figure 6E), on the 25th day after transplanting (DAT), the nutrient solution was replaced with tap water to prevent bitterness in the lettuce at the time of consumption.
In the plant growth chamber, the air temperature was regulated at 20 ± 2 °C, while the RH was maintained at 60 ± 10% RH. The ambient CO2 concentration was kept close to 400 parts per million (ppm). A photoperiod of 14 h per day was applied and the PPF was stabilized at 200 ± 20 µmol m−2 s−1. PPF measurements were performed by placing the quantum sensor at the height of the tallest true leaf. The height of the full-spectrum LED lamps was adjusted every three days depending on the growth stage in order to maintain the desired light intensity.
The nutrient solution used in the experiment followed a commercial Steiner formulation (Soluponics Lechuga, Inverfarms S.A. de C.V., Queretaro, Mexico) [
39], containing the essential micro- and macronutrients required for proper lettuce growth. To prepare the solution, the pH of 20 L of water was initially adjusted to 6.0 using 70% nitric acid (
HNO3), followed by the individual dissolution of mother solution bags A and B in 10 L of water each.
The treatments applied consisted of nutrient solution concentrations of 100% (T100), 75% (T75) and 50% (T50). For the T100 solution, water with an adjusted pH of 6.0 was utilized and 5 mL of Steiner A and 5 mL of Steiner B were added per liter, resulting in an EC of approximately 1970 µS cm
−1. For T75 and T50, the same procedure was followed, adjusting the concentrations to obtain an EC of approximately 1621 µS cm
−1 (T75) and 1272 µS cm
−1 (T50), by adding 3.75 mL and 2.5 mL per liter, respectively. To ensure the stability of these values, daily manual measurements of pH, EC, TDS and temperature of the nutrient solution were taken at 8:00 a.m. using a multiparameter water quality meter (model C-600, VIHELM, Shenzhen, China), as shown in
Figure 7.
2.5. Variables Obtained from the Cultivation Experiment
During the cultivation experiment, the Soil–Plant Analysis Development (SPAD) index was measured every 3 days between 8:00 AM and 9:00 AM using a portable chlorophyll meter (model MC-100, Apogee Instruments Inc., Logan, UT, USA) on the secondary leaves of the third and fourth leaf of each plant. The use of SPAD units is justified by their ability to provide fast, non-destructive and reliable measurements that correlate well with traditional destructive methods [
40].
On 29 June 2024, 26 days after transplanting, seven lettuce plants from each treatment group were harvested to evaluate phenotypic variables. The lettuces were carefully removed from both the hydroponic net pot and the foam pad. The fresh weight of the entire plant, including the root system, was recorded, followed by individual measurement of each component. A tape measure was used to determine plant height and root length. The stem diameter at the intersection point with the root was measured using a Mitutoyo (model 500-196-30, Mitutoyo Corporation, Kanagawa, Japan) digital caliper. The number of leaves was determined by manual counting. To obtain dry weight, each lettuce and its corresponding root system were individually placed in Kraft paper bags and dried in an incubator (model FE-131AD, Scientific Equipment F.E.L.I.S.A. S.A. de C.V., Ciudad de México, México) at 70 °C for 72 hrs. Subsequently, the dry mass was measured using a digital scale (model GY20002, Goyojo, Guangzhou, China).
The three remaining lettuces from each treatment were used to determine the proximal composition of the moisture percentage, protein content, fat, fiber, ash and carbohydrate levels. For the analysis, samples were collected from the upper, middle and lower sections of each plant. Moisture content was determined by drying the sample at 70 °C in a vacuum oven until it reached constant weight. Protein content was calculated using the Kjeldahl method [
41], employing a micro digester (model MDK-6, Novatech Equipos Científicos S.A. de C.V., Naucalpan, México) and a nitrogen to protein conversion factor of 6.25 to estimate crude protein. The ash content was determined through calcination in a muffle furnace (model 273, Indef S.R.L., Buenos Aires, Argentina) at an air temperature of 550 °C for 24 h. Total fat content was determined using the Soxhlet method [
42], with n-hexane as the extraction solvent. Finally, carbohydrate content was calculated by difference.
For statistical analysis of the collected data, normality was first verified using the Shapiro–Wilk test. Subsequently, an analysis of variance (ANOVA) was performed to identify significant differences among the treatments used. Mean comparisons were conducted using Tukey’s test at a 5% significance level. Statistical procedures were performed using Microsoft Excel (version 2021), with its built-in data processing tools.
4. Discussion
The purpose of a plant growth chamber is to create controlled environmental conditions that favor the growth and productivity of agricultural crops. Developing such structures requires appropriate infrastructure that ensures the maintenance of ideal conditions, as well as the implementation of an environmental monitoring and control system. In this project, the smart monitoring and control system proved effective by providing real-time data on environmental conditions, thus ensuring a favorable environment for the development of hydroponic crops.
The implementation of the NFT hydroponic technique within the growth chamber proved beneficial for the year-round cultivation of lettuce, as it created favorable environmental conditions regardless of the season. This favored uniform plant growth, as no significant differences were observed in the phenotypic characteristics of the harvested lettuce, including chlorophyll content. However, the proximal analyses revealed significant differences between treatments, particularly in protein, Sodium, Potassium and Zinc content.
The T75 treatment proved to be the most effective of the treatments used in the growth chamber, as it showed the fewest significant differences in the proximal analyses compared to T100 and T50, promoting lettuces with acceptable phenotypic characteristics while achieving 25% nutrient savings.
The developed system offers benefits that contribute to a sustainability-oriented approach in agriculture. First, by employing the NFT hydroponic technique, it achieves a 90% reduction in water usage compared to traditional farming methods [
49]. Second, supplying nutrients in a controlled manner helps minimize waste. Third, as the system operates within a closed and controlled environment, plant exposure to external factors is minimized significantly reducing the need for agrochemicals and promoting cleaner, more environmentally friendly practices.
Although the current energy consumption per plant (0.46 kWh/day) is relatively high, this value can be lowered by optimizing space usage through increased plant density (plants per m3), integrating renewable energy sources and applying energy optimization algorithms. Therefore, the system shows strong potential for sustainability.
According to the studies reviewed in the state of the art, when comparing environmental management IoT technologies with our proposal, similarities were found in the microcontroller used, as reported in [
21,
28], as well as in the monitored variables, which align with those reported in [
18,
21,
23,
26]. Similarities were also found across all studies regarding the use of mobile or web applications for data visualization. The main differences are highlighted in
Table 6.
In comparison with systems described in previous studies, the system proposed in this work presents several advantages. One of its main strengths lies in its ability to simultaneously monitor multiple environmental variables: air temperature, RH,
CO2 concentration, photoperiod and PPF. In most of the reviewed works, only two or three variables were monitored, as seen in [
18,
21], with [
20] being the only one comparable to the system described in this article.
Similarly, the described system stands out by offering complete microclimate control in a fully enclosed indoor environment, including automatic regulation of air temperature via air conditioning, ultrasonic humidification and photoperiod control using full-spectrum LED lamps. This level of automation was only partially addressed in [
18], which focused on aeroponic systems with more limited environmental control.
Another significant contribution is the implementation of a responsive web application and a mobile app for visualizing and modifying the setpoints of environmental variables, unlike [
21,
25,
28,
29], the proposed system allows remote access to data outside the local network. This feature greatly enhances the system’s accessibility, enabling users to monitor and adjust crop parameters from any location with Internet access.
An additional contribution is the automatic email alert system triggered by data transmission failures, which strengthens the system’s operational reliability. This is only comparable to the WhatsApp alert system used in [
31].
Regarding experimental validation, this study not only presents the behavior of environmental variables but also evaluates phenotypic and nutritional characteristics of the harvested crops like fresh and dry weight, number of leaves, plant height, root length and content of proteins, fiber and carbohydrates. This comprehensive approach sets it apart from studies such as [
20,
23,
24,
25,
28,
31], which do not assess the direct impact of environmental conditions on final crop quality.
Despite the proper performance of the proposed system, one of the main limitations identified compared to [
24] is the absence of an automated control loop for adjusting nutrient solution variables such as pH, EC and TDS. In its current version, these adjustments must be made manually. It is anticipated that future implementations will incorporate control mechanisms and algorithms capable of regulating these variables in real time, thereby improving nutrient solution management.
Additionally, although an IoT architecture based on five layers is proposed, the current implementation does not incorporate artificial intelligence for predictive analysis, as seen in the system developed by [
23].
5. Conclusions
This article presents the development of a smart agriculture system that monitors and controls various environmental variables, supported by a web application and a mobile app for data visualization. Additionally, this study analyzed the growth of lettuce (Lactuca sativa) in hydroponic systems, evaluating three treatments with different nutrient concentrations: 50%, 75% and 100% of the Steiner solution, respectively.
The presented system demonstrates effective integration of sensors and actuators in an indoor hydroponic environment, using an ATmega2560 microcontroller and the MODBUS RS-485 communication protocol. This configuration enabled real-time monitoring and control of environmental conditions, maintaining stable key variables for plant development such as air temperature, RH, PPF and photoperiod, thereby promoting uniform plant growth. Additionally, the proposed architecture supports expansion of up to 31 slave devices, including sensors and control mechanisms, providing high adaptability for deployment in scenarios that require a greater number of variables to be monitored and controlled.
The implementation of the smart agriculture system enabled the evaluation, through an experimental design using 50%, 75% and 100% nutrient concentration treatments of the phenotypic and nutritional development of lettuce under controlled conditions. The finding of few significant differences between treatments suggests that controlled environmental variables contribute to crop uniformity, regardless of variations in the applied nutrient solution.
The integration of the smart agriculture system not only ensured efficient control of environmental conditions, maintaining stability over 26 days without technical failures, but also enabled continuous access to crop data via electronic means and offered opportunities for improvement, such as the integration of alarm systems.
The inclusion of a responsive web application and a mobile app for environmental monitoring and control represents a substantial improvement over local or limited-access solutions. This design facilitates remote interaction, interface customization and efficient management of crop parameters, providing a user-centered approach that is key for adoption in diverse contexts.
The presented system is scalable and adaptable with potential applications in both commercial production and educational and research environments. Furthermore, its implementation promotes sustainable agricultural practices by reducing environmental impact, as hydroponic systems can save up to 90% of water compared to traditional farming methods [
49] and contributes to the technological advancement of the agricultural sector.
However, it is important to note that the proposed system does not yet incorporate an automated control loop for adjusting nutrient solution variables such as pH, EC and TDS. This limitation represents a key aspect to be considered in future research.
Future Work
Although the development board based on the ATmega2560 microcontroller proved to be suitable for managing various sensors and control mechanisms, and for maintaining communication with the IoT module to send and receive data from the cloud server, the proposed architecture has been designed with scalability and adaptability in mind.
This design allows for easy migration to more robust microcontrollers such as the STM32 in future implementations. This capability will be crucial for integrating artificial intelligence modules that require greater processing power.
As future work, it is planned to expand the system’s implementation toward different leafy crops such as spinach and Swiss chard to evaluate its effectiveness. Parallel to this, automated monitoring and control sub-systems for the nutrient solution will be integrated to ensure homogeneity at all crop stages.
To address the need for intelligent management highlighted by current research, the integration of low-complexity supervised machine learning models, such as decision trees or logistic regression, is proposed. These models will enable the identification of patterns in historical data and the generation of alerts regarding crop conditions, such as deviations in pH or EC ranges [
50], thereby reducing agronomic risks and improving the final product’s quality.
Finally, the implementation of a cloud-based recommendation system is proposed, using rule-based and trend-driven logic to support decisions such as adjusting irrigation cycles or modifying reference values for environmental variables [
51]. This enhancement will enable more precise and timely adaptation of the cultivation environment, improving overall performance, reducing unnecessary resource consumption and supporting better long-term management.