IoT System for Monitoring and Controlling Microclimates in a Fruit Fly Breeding Chamber ^†

Abstract

Controlling the microclimate is vital in fruit fly breeding. This project develops an automated IoT system for monitoring and controlling temperature and relative humidity in chambers, optimizing processes through accessible and flexible technology. It uses a multi-layer system starting with the application layer with the ESP32 for data acquisition and actuator control. The second layer is the network layer, and the perception layer uses VisionFive2 with MQTT and HTTP protocols for communication, as well as NodeRed for flow orchestration and MySQL for data management and storage. In the validation, there are absolute errors of ±0.434 °C and ±0.5 RH, which are values within the acceptable ranges for these applications.

1. Introduction

Controlling environmental conditions in a given setting is vital to the success of studies on fruit fly breeding, specifically Anastrepha, a pest that has had a significant impact on crops in Latin America. By controlling temperature and relative humidity, rearing chambers allow for analysis of larval stages, fertility, longevity, and host types. In other cases, they enable the development of biological control programs and sterile insect techniques (SIT). Precision control will eliminate variability in these climate chamber systems, giving greater consistency to the process [1,2,3].

In recent years, the Internet of Things (IoT) has gained relevance in the technological field, becoming a key tool in automation and enabling optimal management through intelligent systems in the areas of agriculture, livestock, and environmental control [4,5,6]. The ability to process and collect data in real time leads to automatic decision-making, resulting in the optimization of resources, reducing human intervention and, therefore, errors, as well as providing traceability of production conditions [7,8,9].

The use of platforms and development boards based on RISC-V architecture offers broad integration with microcontrollers (ESP32), stream orchestration platforms such as Node-Red, and database systems (MySQL) offers great flexibility in hardware integration as well as communication protocols [10,11,12].

Similar systems have been used in the control of smart bird cages [13], monitoring of electrical generation and distribution systems [14,15,16], smart irrigation control [17,18], among other applications, have demonstrated that the optimization of resources such as energy [19] and error reduction are feasible with the integration of processes with the cloud [20,21,22].

Lightweight protocols such as MQTT and CoAP are solid options for efficient communication between IoT applications, as they offer low energy consumption and high reliability in distributed sensor networks [23,24,25,26]. This is relevant because insect farming processes require minimal maintenance intervention and guarantee continuous operation.

Likewise, the incorporation of AI can anticipate adjustments for unwanted fluctuations in temperature and relative humidity [27,28,29,30]. Several studies, such as environmental control in aquaculture or intelligent compost management, support the technological potential of these applications, substantially improving productivity and product quality [31,32,33].

This proposal develops an IoT system for monitoring and controlling temperature and relative humidity in fruit fly breeding chambers using the ESPS32 as a data acquisition card, NodeRed for flow orchestration, MYSQL for data management, and VisionFive2 as a processor. This maintains a local and remotely accessible graphical interface, ensuring continuous operation and contributing to the automation and modernization of phytosanitary programs in Ecuador.

Unlike commercial models, this proposal integrates two control variables: temperature and relative humidity, as well as historical data management and an intuitive interface with local or remote accessibility. These features allow for a response to external factors that alter the experiment, such as power failures and climatic variations, enabling the validation of conditions and avoiding the repetition of the experiment due to external factors [5,33].

2. System Design

The design of the IoT monitoring and control system is based on three layers, presenting a functional separation and future scalability, ensuring interoperability between heterogeneous hardware, as shown in Figure 1.

Although similar solutions exist on the commercial market, this proposal tailors its development to the specific needs of the experiment with a system featuring an intuitive interface for operation and a control system that ensures minimal variation in the microclimate, as well as autonomy, which avoids constant monitoring [6,22].

2.1. Perception Layer (Edge)

This layer is at the data acquisition level, in this case temperature and relative humidity.

Sensor: SHT32, digital sensor with accuracy of ±0.3 °C and ±2% RH, long term stable, internal temperature compensation.

Data acquisition: ESP32 microcontroller, dual-core processor, Wi-Fi connectivity

Local preprocessing: the microcontroller executes filtering routines that reduce noise, improving data reliability.

This layer is defined as Edge because it allows data acquisition and preprocessing locally using the ESP32. This happens before the information is transmitted to the network, ensuring speed and reliability [10].

2.2. Network Layer

The sensor nodes and cameras are connected to the central server via a 2.4 GHz wireless Wi-Fi network.

The MQTT protocol was selected because it allows for the exchange of information, in addition to its low bandwidth consumption and its ability to tolerate temporary interruptions [24,25]. Complementing this, the HTTP protocol was also selected for tasks such as configuration, testing, and validation.

The VisionFive2 card acts as a server where a MosquittoMQTT Broker organizes the data so that information flows asynchronously and flexibly through a publish/subscribe scheme. Additionally, it incorporates a mechanism for automatic reconnection QOS1, making message integrity robust [23].

2.3. Application Layer

In this layer, logical orchestration is implemented using Node-Red, which facilitates the integration of systems for controlling temperature and relative humidity variables. It also allows for the management of automatic notifications in the event of unwanted deviations, enabling autonomous operation in real time [23].

The integration of processes for data management, variable monitoring, and control is fundamental to the operation of the system. Node-Red establishes the flow of input and output data (sensors/actuators) and operations based on data. This orchestration engine is based on node architecture, which allows for scalability, implementation of logical functions, control algorithms, and real-time analysis.

The information collected is stored in a MySQL database hosted on VisionFive2, allowing users to access historical data through graphs or CSV files. It is also designed to send notifications in case of critical deviations, ensuring greater process reliability.

The sensor nodes and cameras are connected via a 2.4 GHz wireless WiFi network.

2.4. Experimental Environment

The breeding chamber is a technological conversion of a conventional panoramic refrigerator. The modification allows for controlled microclimates to be obtained for the breeding of fruit flies. It has features that guarantee stable temperature, relative humidity, and air circulation, maintaining optimal conditions at each stage of the fly’s development.

The chamber has a capacity of 1.5 cubic meters and an adjustable working range of 0 to 30 °C with an error of ±0.3 °C, temperatures recorded in the areas affected by the pest, as well as adjustable relative humidity between 40% and 90 ± 2%, which prevents dehydration or the proliferation of fungi.

Figure 2 shows the layout of the temperature and relative humidity sensor.

2.5. Control System

The heat source is a 250 W ceramic electric resistance controlled by a solid state relay. This resistance is located on the outside and connected to the inside of the chamber by a ventilation duct.

Humidification is provided by a 2.4 MHz ultrasonic generator, which is regulated by a PID to prevent condensation peaks. This is shown in Figure 2.

3. Results

The main objective of this work is to validate the accuracy and autonomy of the IoT system in order to maintain stable microclimates.

The performance of the IoT system in monitoring and controlling the temperature and relative humidity of the breeding chamber is verified by the stability of the microclimate generated and the reliability of data transmission and storage.

The data measured and stored by the system were compared with the HTC-1 thermohygrometer. The comparison was made at different levels of temperature and relative humidity.

Table 1. Comparison of temperature readings between thermohygrometer and SHT32 sensor with respect to the reference value.

Set Point °C	Thermohygrometer °C	Sensor SHT32 °C
3	3.5	3.4/3.2
5	5.5	5.4/5.24
10	10.5	10.4/10.29
15	15.5	15.38/15.27
20	20.5	20.37/20.25
25	25.5	24.36/22.4
30	30.5	30.45/30.17

shows the comparison of readings between the desired value and the values measured with two temperature devices.

Table 2. Comparison of relative humidity readings between thermohygrometer and SHT32 sensor with respect to the reference value.

Set Point °C	Thermohygrometer °C	Sensor SHT32 °C
40	39.2	39.5
50	49.2	49.5
60	59.2	59.5
70	69.2	69.5
80	79.2	79.5
90	89.2	89.5
100	99.2	99.5

shows the comparison between the desired value and the values measured with two relative humidity devices.

The relative error is calculated using the values obtained, as shown in Figure 3. the mean absolute error is 0.434 °C in temperature and a mean relative error of 4.8%, while the mean absolute error is 0.5%RH and a mean relative error of 0.78%, data that confirm that the microclimates generated are within the accepted ranges [3].

Once the errors have been calculated, consistency tests are carried out over a period of time, showing the evolution of temperature and relative humidity in 20 s intervals, as shown in Figure 4.

Other variables necessary in the process are shown in

Table 3. Performance metrics of the IoT monitoring system.

Variable	Average Value	Standard Deviation	Range	Unit
MQTT response time	180	25	150–220	ms
Total energy consumption	12.4	0.6	11.8–13.2	Wh
Wi-Fi reconnection time	3.5	0.8	2.1–4.7	s

. The performance indicators show that the MQTT response time handles messages quickly for real-time application, averaging 180 ms. Energy consumption is 12.4 kWh, which varies depending on the desired microclimate value. However, consumption is moderate. Finally, the WiFi reconnection time averages 3.5 s, which is acceptable since it is a non-critical application.

4. Conclusions

The IoT system shows acceptable control of temperature and relative humidity variables with absolute errors of ±0.434 °C and ±0.5 HR, respectively. Based on previous studies, this variation is acceptable, guaranteeing optimal conditions and reducing the variability of the climatic environment for fly breeding.

As the hardware and software are open source, the response times in the message package are adequate, as it is not a critical process, with average responses of 180 ms in the MQTT protocol and an average of 3.5 s in WiFi reconnection times. Consumption also shows that the operation of these technological proposals is sustainable in terms of the responses they will provide to the management of certain pests.

One of the main contributions is the development of an intuitive interface that allows historical data management with local and remote access, improving the traceability of the experiment. The control systems implemented reduce the variability of the microclimates generated, ensuring optimal and consistent conditions for fruit fly breeding. Additionally, the robustness of communication and autonomy of the system

The multi-layer architecture allows for flexible and scalable integration with real-time monitoring capabilities, database management for historical monitoring, and notifications of possible critical deviations that could affect the experiment.

The system can be improved. The limitations identified focus on the dependence on a stable and continuous power supply, as well as variables such as atmospheric pressure, CO₂ concentration, and lighting levels. For future developments, we propose the incorporation of real climate curves, improving the capacity for advanced phytosanitary applications.