Holistic IoT and cloud-based telemetry architecture for proactive fire monitoring in smart agriculture

Abstract

The rise of harsh weather conditions has made the crop yield vulnerable to fire and variable losses. To address this issue, this article proposes a real-time fire monitoring system which is appropriate for modern smart agriculture. This system utilizes cloud computing technology, Internet of Things (IoT) sensors, telemetry technology, and embedded systems technology to monitor the status in the fields in real time every second. The system has three layers. The first is the IoT device layer, which is composed of flame and smoke sensors, a raspberry pi 3 B+, and a network gateway. The second is the ThingsBoard cloud layer, which is used for efficient processing of large amounts of information. The third is the telemetry layer, which is used for aggregation of the information collected. One of the advantages of the system is the use of a customized aggregation algorithm, which uses sensor information and sends the results in JavaScript object notation (JSON) format using the message queuing telemetry transport (MQTT) protocol. The system was tested in the fields and performed well. It recorded an accuracy of 96.1% in detecting fire in 50 tests, with the rate of false alarms being below 2.8%. It is also clear from the tests that the system can differentiate between true and false alarms. The proposed system sends information to the cloud every two seconds, with an average response time of below 300 milliseconds. The results show that the monitoring system has a reliability rate of over 98%. The results also demonstrate that the Raspberry Pi used in this study had a stable and reasonable central processing unit (CPU) and memory usage rate. Compared to other previously proposed prototypes, the proposed system for monitoring has the advantage of incorporating resource telemetry and fire detection within the same IoT and cloud computing environment, a notable improvement towards sustainable agriculture and food security, as well as the mitigation of agricultural risks.

Introduction

Agriculture is a vital industry that is critical to the social fabric across the globe and cannot be understated. Beyond food production, it drives employment, economic opportunity, and environmental stability. In a rapidly changing world, agriculture serves as the foundation for global development^1–3. In addition, a joint report developed by some of the top international organizations including, (1) the Food and Agriculture Organization (FAO), (2) International Fund for Agricultural Development (IFAD), (3) United Nations Children’s Emergency Fund (UNICEF), 4) World Food Program (WFP), and 5) World Health Organization (WHO) outlines and emphasizes the substantial rise in global food insecurity⁴. Rapid population growth, overexploitation of resources, and climate change are only exacerbating the situation. The figures speak for themselves: food demand is expected to increase by 70% by 2050, as the population continues to grow and global food consumption increases by around 1.3% each year. Most food in developing countries still comes from small farms using fairly rudimentary methods. Recently, the cost of fertilizers, for example, has been rising steadily, causing many people to worry about rising food prices and their impact on food security^5–7. On top of all this, forest fires are becoming an increasingly important issue for agriculture. Take, for example, the case of Morocco, which experienced a loss of 25,600 hectares between the years 2001 and 2022 as a result of forest fires, and a loss of 29,000 hectares as a consequence of bush fires. In the year 2022, the country experienced a loss of 10,800 hectares as a consequence of fires, which accounted for 90% of the country’s forest loss in the year⁸. There has also been an increase in the number of fire alerts, with over 2,000 cases reported between late 2020 and late 2023, with the highest cases reported in the year 2022⁸. However, the country is not the only one affected. Experts from the United Nations Environment Programme and the World Resources Database warn that climate change will only increase the frequency and severity of forest fires. By 2030, we are likely to see a 14% increase in fires worldwide, and by 2050, that figure will rise to 30%. By the end of the century, we can expect a 50% increase⁸. All of this points to one thing: agriculture needs smarter, technology-driven solutions, and it needs them fast. According to the latest figures from the United Nations, the world’s population is expected to reach 9.8 billion by 2050 and 11.2 billion by 2100, which represents considerable pressure^9,10. Food needs are constantly increasing, but arable land is constantly decreasing. Natural resources? The resources being referred to here are undergoing rapid depletion. This problem is further complicated by the impacts of climate change, such as forest fires, increased temperatures, salinization of the soil, and flooding. The issue of food security has grown from being just a slogan to a full-fledged issue of interest to all nations, as seen in the various research projects conducted in recent times¹¹. In addition to this, various research works conducted in the past¹² have shown how climate change impacts agriculture and food security, highlighting the scientific basis of the impacts, predictions of temperature and precipitation for the current century, and an analysis of a range of impacts such as extreme weather events, sea-level rise, invasive species, pests and diseases, and even lesser-known crops, which are often ignored in most cases. Sustainable agriculture is being seen as a more prudent way of carrying out agricultural activities, as it seeks to balance the population’s needs with the actual capacity of the environment to support such activities. The manuscript indicates that the rate of population increase globally is much higher than the rate of production of food, and this impacts natural resources in a way that not only threatens the current generation but also the next generation, as highlighted in past research works¹³. In this regard, recent research works¹⁴ have shown that sustainable agriculture practices such as climate-smart agriculture, organic farming, sustainable intensification, and regenerative agriculture should be promoted and highlighted, as well as innovations such as precision agriculture and pest control strategies with low impacts on the environment. Sustainable agriculture involves not only conservation of the soil and water but also preparedness for and mitigation of fire incidents^15,16.

Traditionally, the management of fires is more reactive in nature, thus creating a significant opportunity for the development of real-time monitoring technologies. In this context, a recent study¹⁶ focuses on controlled burning in the management of forests to prevent the risk of fires and even reduce greenhouse gas emissions. While controlled fires do lead to the release of greenhouse gas emissions, these emissions can prevent the release of much more carbon emissions in the case of uncontrolled fires. However, the study concludes that controlled burning does not influence the carbon balance in the atmosphere, since there is no significant change in the total carbon content. On the other hand, smart agriculture is following a different path. In the present context, the focus is more on developing novel and innovative techniques, including the development of advanced fire detection systems, to bring about significant changes in traditional agriculture, as indicated in the findings of the earlier study¹⁷. Modern agriculture makes use of advanced technologies, including remote sensing, satellite imagery, artificial intelligence, to monitor the forests and detect fires in real time, as indicated in the findings of the earlier article¹⁸. In this context, the study developed an artificial intelligence (AI)-based system for the detection of fires. By integrating you-only-live-once (YOLO)v4-tiny visual detection algorithm and light detection and ranging (LiDAR) technology, the model can classify fires in 1.24 s with 91% accuracy and an F1 score of 0.91. Moreover, it can generate a 3D model of the region of impact in real-time, which can be sent to a ground station. As reported in the article, the model’s performance in dense tropical forests showed its superiority over conventional methodologies, including Bayesian classifiers, random forests, and support vector machines. Hence, this model can be considered a more efficient and cost-effective way of managing forest fires. Another article²⁰ also discusses the latest technology used to detect, track, and prevent forest fires. It also discusses effective ways of handling forest fires, limitations, and scope for improvements. As reported in this article, forest fires cause a lot of environmental and climatic changes, so prevention, tracking, and control of forest fires using emerging technology can be considered advantageous and essential. These emerging technologies provide farmers with information regarding forest fires, enabling them to respond promptly and thereby protecting the environment from major damages. As suggested in this article, the contribution of forest fire detection in smart agriculture can be considered a part of a new model of intelligent cultivation with the ability to detect forest fires, as suggested in this article¹⁹. For example, drones can be used to detect forest fires using infrared sensors, and the only thing needed in this case is prompt action to avoid major damages^21,22.

The use of technology significantly reduces damage to assets, and these risks are again associated with indirect environmental impacts such as increased emissions of particulate matter and gases released by the burnt vegetation and impacted agricultural land use. Overall, the potential for fire detection solutions to contribute to smart agriculture is a new, promising approach for valuing future smart agriculture options to protect farmers as well as all of agriculture that is involved with buying, selling, and perhaps being affected by the smoke from forest fires^23,24. The rise of IoT technology has opened the door to numerous innovations in smart agriculture, which help advance traditional agricultural practices²⁵. IoT connects equipment, devices, and sensors, allowing farmers to analyze data in real-time and make informed decisions²⁶. Recent research in²⁷ digs into a smart farming device that uses IoT and machine learning to spot crops in danger of catching fire. Sensors collect data from the field, and then several machine learning (ML) algorithms, decision trees, random forests, support vector regression, nearest neighbours, jump in to analyse it. In case of anomaly detection, the system instantly sends a notification to Android devices as a warning. In particular, decision trees have shown 94% accuracy in fire risk detection. This is a clear indication of the importance of decision trees as a predictive tool for fire detection and precision agriculture. In a recent study²⁸, FireNet-convolutional neural network (CNN), a deep learning algorithm, was presented as a superior tool for forest fire detection compared to other models. This algorithm is more efficient and effective compared to other models, as shown in the study that evaluated FireNet-CNN. In particular, FireNet-CNN used two custom datasets: one for quick image processing and the other for use on drones that have sensor capabilities. In addition, FireNet-CNN used synthetic data to improve efficiency and accuracy. In particular, interpretable AI was used to improve user trust and confidence in critical applications. In spite of the challenges that still need to be addressed, FireNet-CNN offers hope for effective fire detection and fire risk management. In a different study²⁹, a fire safety system was constructed using Arduino Uno as a controller and sensor of sirens in university hospitals across Morocco. In particular, the system was constructed to ensure fire safety and effective data transmission and tracking of data from the sirens. In particular, the hardware configuration of the system consisted of voltage sensors, regulators, a Darlington (TIP120) transistor, an optocoupler, a liquid crystal display (LCD) display, and a siren, which were simulated and modeled using Arduino IDE. In other contributions, IoT plays a crucial role in smart agriculture, particularly in fire detection. In particular, advanced sensor and monitoring systems have improved early detection of fire threats and quick response as shown in different contributions^30,31. In particular, IoT integrates different components of smart farming that have improved efficient management of resources and reduced fire risks^32,33. In particular, IoT has significantly improved fire detection compared to other models³⁴. A study³⁵ asserts that timely fire detection is also essential for environmental conservation and wildlife protection. This is because of the extensive damage resulting from forest fires. Thus, an automated system for timely alarm sending through connected devices is highly warranted. Recent research has focused on the application of IoT technology in firefighting, such as an Arduino-based system with smoke and temperature sensors for sending alarms upon detecting danger³⁶. Another study has observed that safety rules are not always followed perfectly and has applied the combination of IoT and the YOLOv5 computer vision algorithm for real-time fire detection and monitoring. The study has demonstrated that IoT technology can correct some of the errors made by the YOLOv5 algorithm and send appropriate information for rescue teams. Smart farming is also benefiting from the application of IoT technology by delivering information in near real-time using the MQTT protocol for timely responses towards fire prevention and smart crop management³⁷. Blockchain technology is also being applied for ensuring the certification of information integrity for enhanced trust in system records. Recent inquiries suggest that cloud computing is now the foundation for intelligent agriculture, enabling the analysis of large-scale information from diverse devices for crop and weather condition analysis, and has revolutionized smart detection^38, 39,40. Another research article⁴¹ discusses a forest fire detection system based on the ESP32 platform, using sensors for rainfall, sound, temperature, humidity, and motion. Another article⁴² discusses the use of IoT in irrigation systems, as well as controlling pests, in Mediterranean regions where water is scarce. In addition, saline soils are common in these regions. The authors developed a system that measures moisture, temperature, conductivity, and climatic factors. Real-time data is transmitted to an open platform, alerting farmers when thresholds are exceeded. The authors tested their system in olive groves in Crete, a Greek island, where it was found to be cost-effective and efficient in preventing water waste, as well as in protecting crops from harsh weather.

Considerable research has been conducted on IoT-based fire detection systems, specifically on smoke or flame detection. Some studies have also proposed advanced frameworks for real-time detection of fires, but challenges still remain, such as scalability and security. Additionally, there is a lack of information on the topic of automation and the manner in which the alerts are sent. Moreover, there is a lack of information on the basic parameters of the proposed detection system, such as the latency of the proposed system, the reliability of the proposed system, and the rate of false alarms. The present article proposes a fire detection system that possesses improved capabilities. The proposed system utilizes IoT sensors and also enables the real-time monitoring of CPU and memory utilization. Additionally, the proposed system enables the use of secure communication through the use of MQTT and also the automation of the process of data visualization on the ThingsBoard platform. Such a design is ideal for the purpose of smart agriculture and takes the concept of fire prevention to the next level. In addition, given the global rise in population, this paper was motivated by the need for safe and sustainable food sources. The research will increase the resiliency of the agriculture system against potential threats of wildfires with (1) IoT, (2) embedded systems, (3) cloud, and (4) fire detection. ThingsBoard’s telemetry focus aims to help meet global food security through resilience and efficiency of farming practices. Despite these developments, existing fire detection systems are fragmented, with a focus on sensing, transmission, or cloud storage individually. A void remains for unified, technically validated architectures that combine IoT sensing, telemetry, cloud computations, and autonomous alerting in a single, robust system. This void is the motivation for the present study.

The main contributions of this proposed paper are as follows:

• Holistic IoT–cloud–telemetry system for agricultural fire control that incorporates real-time system resource monitoring (CPU/memory usage) with real-time fire detection to ensure scalability and run-time robustness.

• Three-tier architecture that consists of: (i) device layer of IoT (smoke/flame sensors, Raspberry Pi 3 B+, gateway), (ii) ThingsBoard layer of cloud computing, and (iii) layer of telemetry processing (dashboards, rule engines, user interfaces).

• A self-configuring algorithm for integrating sensors, MQTT/JSON communications with ThingsBoard, and data processing.

• Experimental validation demonstrating reliable real-time fire data collection, cloud analysis, and automated email alerts for anomaly detection.

• The proposed system achieves detection accuracy of up to 96.1% with a low number of false alarms (2.8%). MQTT connectivity ensures fast latency (< 300 ms) and high availability (> 98%), while CPU/memory post-processing ensures stable resource utilization / system operation.

• Differing from other studies, our study unifies fire detection, scalability of clouds, telemetry visualization, and automated notification in a single end-to-end system, enhancing agricultural sustainability, better food security, and risk management on a proactive level.

The formation of the rest of the paper is as follows. The materials and methods are described in Sect.  2. The results of the proposed system are presented in Sect.  3. A discussion of the proposed system is presented in Sect.  4. The conclusions and future work are presented in Sect.  5. The statements and declarations are made in Sect.  6.

Materials and methods

Embedded systems and IoT technology can be applied extensively in smart agriculture. There remains no single globally agreed-upon standard operating architecture. This section introduces the new three-layered IoT, cloud, and telemetry system and algorithm of the proposed system. Comprising real-time fire detection, telemetry transmission, system resource monitoring, and automated anomaly notification, in contrast to previous works that consider sensing alone or simply cloud storage individually, the current technique incorporates all these functions in a single end-to-end solution. Along with designing description, this section focuses on scalability, reliability, and reproducibility that are essential to move from prototype to product-ready systems.

Architecture of the proposed scheme

Figure 1 represents the system’s framework that has been envisioned to efficiently gather and process agricultural data in real-time. It comprises three distinct layers: (i) IoT device layer, (ii) ThingsBoard cloud layer, and (iii) telemetry data processing layer. The intelligent agricultural fire monitoring system integrates embedded systems, IoT sensors, the MQTT protocol, cloud computing, and telemetry visualization into a unified framework. Figure 2 illustrates the process flow for receiving and analysing real-time fire monitoring data. Differing from current solutions that either end at sensor acquisition or upload to clouds, our system offers a complete pipeline, from sensor integration to cloud dashboards and alarm-triggered email notifications, all while our innovation also lies in adding CPU and memory usage telemetry to guarantee scalability and run-time reliability of the IoT devices even under real workloads. These layers are as follows.

Fig. 1.

Framework of the proposed scheme.

Fig. 2.

Process diagram for real-time data reception and analysis.

IoT devices layer

This layer integrates flame and smoke sensors with a Raspberry Pi and an Internet of Things (IoT) gateway. The sensors connect via an analog-to-digital converter (MCP2308) to enable interchangeability. The Raspberry Pi 3 Model B+ acts as an integrative tool that aggregates sensor information, monitors system resources, and sends information using MQTT. Although the monitoring of CPU and memory usage is not usually stressed in logging, such monitoring has a significant impact in that it allows for the continued use of IoT devices even with high workload. Previous projects have not considered the ability of the system to cope with high workload, but the current setup has remedied that. Additionally, the use of JSON formatting, as in MQTT’s “v1/devices/me/telemetry,” allows for smaller packets, which is important in a rural area where bandwidth is limited.

ThingsBoard cloud layer

The data processing and management layers for telemetry data play a vital role in the cloud computing environment. These middle layers control the safe transmission and storage of telemetry data. In addition, the architectural layers control data authentication using different access tokens, thus restricting data access to only authorized personnel while preventing unauthorized entities from accessing the farm’s data. Lastly, the closed source approach, as exhibited by AWS IoT and Firebase, is characterized by high costs, relying on their technology stack. On the other hand, ThingsBoard is characterized as an open-source solution that allows for easy scalability while incorporating a wide range of tools in a single package. As such, it presents a viable solution for use in agriculture, whether for deploying advanced technology in farms or for creating technology infrastructure in remote areas.

Telemetry data processing layer

At this layer, the system integrates with the various tools such as dashboards, rule engines, email notifications, and visualization tools, which work in harmony with one another. The layer has been designed to act as a command center, allowing farmers to view the data in real-time and take prompt action. When the fire-related telemetry data crosses the set thresholds, the system is able to recognize the state immediately and send email notifications within seconds, reducing the time taken by farmers to act from hours to mere minutes. The inclusion of dashboards to view CPU and memory consumption, besides fire sensor telemetry, is a unique feature of the proposed system, which is less common in the context of the current smart farming infrastructure.

Algorithm of the proposed system

The code for programming and developing the algorithm structure is shown here. Table 1 shows the inputs and outputs for this research. Table 2 shows the algorithm for the proposed system. As shown in Table 1, the input of this system consists of a constant and a sensor input. On the other hand, the output includes (1) connection status, (2) telemetry data, and (3) disconnection messages. As shown in Table 1, lines # 1 and 2 illustrate the input and output of the system. Lines # 1 through 9 configure everything. Set the key parameters: the ThingsBoard host, access token, and subject for connecting to the ThingsBoard cloud. Next, lines # 10 through 14 handle the MQTT connection. An OnConnectHandler function monitors the connection and clearly indicates whether the connection is established correctly or not. The conversation now shifts to lines # 15–17, which represent the core of the data processing function of the telemetry data workflow. This process involves ingesting sensor data, encoding it into JSON, and then transmitting it to the telemetry topic through MQTT protocol at a QoS level of 1. Sections in line # 18 to 22 are focused on improving data readability through iteration over the key-value pair to display the variables, values, and units associated with them. Sections in line # 23 to 25 have a simple mapping function that maps the variables to their respective units, thereby ensuring data consistency. The major process implementation occurs in Sections of line #26 to 43, where the Main function coordinates all the above-mentioned components of smart agriculture telemetry. This section not only implements the algorithm but also describes the hardware specifications, network configuration, and data management protocols. All the sensors, including smoke and flame detectors, are constantly calibrated based on the specifications provided by the manufacturers prior to the test to ensure data accuracy. The MQTT protocol is used for cloud communication, and QoS level 1 is enabled to recognize the importance of the data being transmitted. A structured approach is followed for all sensors and data, including detection accuracy, response time, reliability, and false alarm rates to ensure precise data traceability. The network configuration and ThingsBoard have been adjusted to ensure smooth data flow. Telemetry data is sent to the cloud at set intervals, and the system uses local memory to prevent data overload. In the communication security domain, MQTT does not support any kind of inherent encryption. In the proposed prototype, device authentication is implemented using a separate access token and telemetry topic names, which are restricted to a specific format, for example, “v1/devices/me/telemetry.” However, the implementation of transport protocol encryption using TLS/SSL on port 8883 was not included in the proposed prototype. In a real-world scenario, it is highly recommended that secure MQTT communication using TLS encryption be implemented for data confidentiality and integrity.

Table 1.

Inputs and outputs of this algorithm.

Inputs	Description
Constant: ThingsBoard_Host	The ThingsBoard host address
Constant: Access_Token	The access token for authentication
Constant: Topic	The MQTT topic for telemetry data
Sensor Inputs	Sensor values such as smokeValue and flameValue are acquired using the MCP3208 ADC via the GPIO interface.
Outputs	Description
Connection Status	Messages including whether the script successfully connected to ThingsBoard or encountered an error during connection
Telemetry Data	Displayed telemetry data including: (1) values for sensors (smoke Value, flame Value), (2) derived values like smoke flame, and fire status, (3) CPU usage, and memory used. (4) formatted telemetry data sent to ThingsBoard.
Disconnect Message	A message indicating disconnection from ThingsBoard in case of a keyboard interruption.

¹MQTT, GPIO, and CPU are the acronyms of message queuing telemetry transport, general-purpose input-output, and central processing unit, respectively.

Table 2.

Pseudo code for telemetry data sent to ThingsBoard cloud.

The current research seeks to examine how the algorithm operates in ThingsBoard. First and foremost, the MQTT client is set up to ensure that it connects with ThingsBoard. The process involves setting up the client, specifying parameters, and providing access credentials. In the main function, an access token is set up to ensure that authentication is allowed for a secure connection with ThingsBoard. The main loop then starts and involves various processes and operations. The internal process involves obtaining information from various sensors, such as environmental sensors. The information obtained is then processed to obtain percentages and determine whether there is fire. The algorithm also processes operational statistics such as CPU and memory usage. Once all operations are complete, the information obtained is then sent to ThingsBoard. In terms of error handling, various measures are put in place to ensure that the algorithm remains stable under irregular conditions. In case of a keyboard interrupt, for instance, the algorithm does not shut down abruptly as it would if it did not have these measures. The processes mentioned above are represented in pseudocode and can be used for the intelligent fire detection system for smart agriculture. The overall aim is to ensure that agriculture remains sustainable while reducing the risk of fire. Therefore, modern technology is used to ensure that agricultural land remains safe from fire-related incidents.

Results and discussion

The results and discussion section of this article provides an overview of the extent to which the fire detection system can be used in a practical context based on real agricultural scenarios. These results clearly show that the fire detection system is capable of successfully identifying conditions that could cause fires, collect relevant information, and transmit it in real time via ThingsBoard Cloud for in-depth analysis and clear visualization. As part of this study, a series of experiments were conducted under real working conditions in a LIMAS laboratory at the Faculty of Sciences Dhar El Mahraz, Sidi Mohamed Ben Abdellah University in Fez. The IoT-based smart fire detection system for smart agriculture was developed using different types of IoT sensors, including flame and smoke sensors, which communicate with a Raspberry Pi to perform local processing and transmit data to the Thingsboard IoT platform using the MQTT protocol. The sensor measurements were taken over several days using repeated tests, with measurements taken multiple times, in order to clarify the various parameters that may vary depending on the flame simulation, such as the distance between the flame and the sensor and the intensity. This proved that all measurements could be reliably reproduced in the laboratory. Thus, the defined experimental method ensures that all data collected by the sensors is reliable and representative, as it has been fully validated using a rigorous testing regime. As shown in Figs. 3, 4, 5, 6, 7, 8, 9, 10, 11, 12 and 13; Tables 3 and 4, the graphical representation of fire parameters on the Things Board dashboard. The dashboard gives you a clear and quick overview of information relating to fires in smart agriculture, flames, smoke, processor statistics, memory usage, etc. This system offers instant access to quantitative information and eliminates the necessity for the user to actively search for such information. The information is displayed in a graphical and tabular form, which helps to quickly identify trends or problems, allowing for instantaneous decision-making. The system has the ability to send user-defined alerts, whereby when the value of the information being monitored exceeds a certain limit, an instantaneous alert is sent to the user, thus reducing the possibility of the user failing to act on the information. The user interface is user-friendly, allowing the user to quickly become proficient in the use of the system, regardless of their level of technical knowledge. The system can be used to monitor information in real-time or to analyze the information to identify the most prevailing trends. The information provided in the form of a graph and a table can be customized to highlight information of special interest to the user. The information provided is constantly updated through telemetry and is accurate. In the case of an anomaly occurring in the system, an instant email is dispatched to the user through the system’s telemetry, informing the user of the problem and allowing instant action to correct the problem. The simulation capabilities provided by the system, along with the instant notifications and the use of real-time information, improve safety and management of the system.

Fig. 3.

The test on the PuTTY terminal.

Fig. 4.

Latest telemetry data for the test.

Fig. 5.

Dashboard results of the proposed scheme.

Fig. 6.

The value of the fire smoke and flame sensor on the ThingsBoard cloud.

Fig. 7.

The value of the flame sensor on the ThingsBoard cloud.

Fig. 8.

The value of the smoke sensor on the ThingsBoard cloud.

Fig. 9.

The value of the CPU usage on the ThingsBoard cloud.

Fig. 10.

ThingsBoard Telemetry configuration.

Fig. 11.

(a) displays the widgets of the real-time status of the system and sensor values; (b) shows widgets capturing the real-time status of the system and sensor values on the ThingsBoard cloud.

Fig. 12.

Database export.

Fig. 13.

E-mail notification.

Table 3.

Time series table in real-time with the status of this study.

Timestamp	Flame	Smoke	Memory Used	CPU Usage	Fire Status
2023-11-11, 17:00:57	42.19%	92.74%	0.16 GB	0.5%	Fire Detected
2023-03-11, 17:01:02	42.19%	92.69%	0.16 GB	0.5%	Fire Detected
2023-11-11, 17:01:08	42.22%	92.78%	0.16 GB	0.3%	Fire Detected
2023-11-11, 17:01:14	42.27%	92.71%	0.16 GB	0.3%	Fire Detected
2023-11-11, 17:01:20	42.29%	92.59%	0.16 GB	0.3%	Fire Detected
2023-11-11, 17:33:37	45.64%	33.72%	0.16 GB	0.5%	No Fire Detected
2023-11-11, 17:33:31	46.71%	33.84%	0.16 GB	0.5%	No Fire Detected
2023-11-11, 17:33:25	47.03%	33.77%	0.16 GB	4.0%	No Fire Detected
2023-11-11, 17:34:19	45.10%	34.18%	0.16 GB	4.2%	No Fire Detected
2023-11-11, 17:34:45	46.32%	33.87%	0.16 GB	4.7%	No Fire Detected

Table 4.

Comparison of related research¹.

Ref.	Year	Embedded Controller	IoT	Cloud Computing	Data Analysis	Communication Protocol	Send E-mail Notification	CPU and Memory Usage	Programming Language
22	2022	Ras pberry Pi	No	No	No	No	No	No	Not Addressed
23	2021	No	No	No	No	No	No	No	Not Addressed
24	2021	No	No	No	No	No	No	No	Python
30	2023	No	Yes	No	No	No	No	No	Not Addressed
31	2020	Raspberry Pi	Yes	No	No	No	No	No	MATLAB software
32	2023	LoRa	Yes	No	No	No	No	No	Not Addressed
35	2023	Raspberry Pi	Yes	No	No	No	No	No	Python
36	2023	Raspberry + LoRa	No	No	No	No	No	No	Python
40	2021	ESP32	Yes	Ubidots IoT Platform	No	No	No	No	Not Addressed
42	2023	ESP32	Yes	ThingsBoard Platform	No	Yes	No	No	Arduino
PS	2025	Raspberry Pi 3 B+	Yes	Thing Board Cloud	Yes	Yes (HTTP)	Yes (email notification)	Yes	Python

¹PS, UART, and HTTP are the acronyms of the proposed scheme, universal asynchronous receiver/transmitter, and hypertext transfer protocol.

The practicality of the system is also demonstrated. When the IoT and other relevant devices are connected, the test was conducted directly on the Raspberry Pi terminal. The live update of the sensor figures on the PuTTY terminal, as shown in Figs. 3 and 4, serves as a practical verification of the system. The graph represents the percentage obtained from the smoke sensor and the flame sensor, thus facilitating the real-time monitoring of the system’s operational status. At the same time, the Python script runs in the background, continuously sending the actual readings from the smoke and flame sensors to the database to ensure data integrity and prevent the use of artificial data. The output from the analog to digital converter is dependent on the input, and the analog voltage is contingent on the reference input. To understand the actual process of the conversion, references⁴³ and⁴⁴ precisely show the process of converting the input and reference voltage to the output voltage. The data is sent to the ThingsBoard Cloud. The registered device acts as the central hub, sending the readings from the Raspberry Pi sensor to the cloud via the MQTT protocol. To view the latest readings for the connected device, the user can select the “Latest Telemetry” section, as shown in Figs. 3 and 4. In the process, the telemetry data is collected in a variable named “payload,” for example, “flame = 25.05%” and “smoke = 50.0%,” and sent to the path “v1/devices/me/telemetry.” This process ensures the integration of new telemetry data into ThingsBoard. The paper has proposed an IoT-based intelligent fire detection system with the integration of ThingsBoard. The proposed system consists of flame and smoke sensors that send real-time data, including sensor readings, processor, memory, and status, directly to the cloud platform. As shown in Fig. 5, the proposed system has particular benefits for fire detection in smart farming. One of the attributes of ThingsBoard is that it is easy to create interactive dashboards, allowing for easy monitoring of data and devices using this platform. This means that agricultural practitioners can create dashboards that can be used to monitor parameters of interest in the fields, which is not readily available in existing smart farming systems. This platform is advantageous in that its main strength lies in its ability to be flexible in that each dashboard has widgets that can be customized to user specifications. Also, dashboards can be assigned to different individuals or groups of individuals, allowing them access to data that is relevant to them. This platform is also advantageous in that it is designed to handle large amounts of workload, thus being reliable in meeting the needs of stakeholders who are interested in developing intelligent farming systems.

Figure 6 displays the measurement readings from the flame and smoke sensor, which are shown simultaneously and plotted together on one cloud graph using the ThingsBoard cloud platform. The flame sensor is denoted using green, whereas the smoke sensor is denoted using blue, as depicted above, to easily differentiate and compare. Figure 7 shows a detailed analysis of the real-time measurement of the flame sensor, along with a description of the results. The minimum value is 45%, the maximum value is 47%, and the average is 46%. The cumulative percentage sent to the cloud is 5,162%, with the last measurement recorded as 47%. Figure 8 shows a representation of the measurement of the smoke sensor using the ThingsBoard cloud platform, with real-time updates as the events are recorded. The range of measurements is a minimum of 6%, a maximum of 32%, an average of 15%, and a cumulative cloud transported percentage of 2,297%. The above graphs give an overview of the response of the flame and smoke sensors. Figure 9 shows a measurement of the CPU usage of the Raspberry Pi using the ThingsBoard cloud platform. The key observations are: the CPU usage is 1% as the minimum and 30% as the maximum, with an average of 7%. The cumulative percentage sent to the cloud is 755%. The latest measurement is 6%. This gives an overview of the current workload of the processor. It is also observed that the memory usage is 0.16 GB, which gives an overview of the system memory usage.

As depicted in Fig. 10, the ThingsBoard telemetry configuration gathers all the connections, aggregating the parameters of the system in a comprehensive table. This representation is not just a list; rather, it is the entire configuration, including the complete picture of the assembly of the system. Then, as depicted in Fig. 11, the parameters of the system are shown on the user interface. Figure 11(a) and 11(b) go even further by presenting the custom tools built into ThingsBoard. These are not just aesthetic skins: they display the launch status using a simple map and show sensor values in real time using easy-to-read user interface elements. It can be seen at a glance what is happening, thanks to the way the title and definition of each widget refer to these essential features. All these visual elements make it much easier to monitor the daily operation of the sensors. ThingsBoard also makes it easy to export data from widgets. Retrieve data in CSV or XLS format, and it works with almost all widgets, depending on what are using. Exported files are organized according to the settings defined in the configuration, so everything aligns. Figure 12 shows how this export works directly from the dashboard. The XLS file includes all system parameters and their current states, making it easy to keep records or analyze data later.

Table 3 shows all the telemetry data collected by the real-time fire detection system, such as flame and smoke sensor measurements, memory and processor usage on the Raspberry Pi, and whether or not it detects a fire. When simulating a fire, the flame sensor measurements vary considerably, with a variability of approximately 42.2%, while those of the smoke sensor reach 92.7%. This means that it detects a lot of particles in the air. Even during this period, memory usage remains stable at 0.16 GB, and the processor is barely working, fluctuating between 0.3 and 0.5%. The system therefore does not lose its connection when it matters most. When there is no fire, the figures stabilize. The flame sensor fluctuates around 46%, the smoke sensor is between 33% and 34%, and the processor uses between 0.5% and 4.7%. This type of stability shows that the system can handle normal conditions without any problems. This structured table not only proves that the system can differentiate between real fires and everyday changes, but also makes it easy to verify its accuracy, response time, and the duration of false alarms. Then there is the ThingsBoard notification center, which covers all the bases for alerts. also, can trigger notifications manually, via the REST API, or configure them to trigger when certain events occur. Everything comes into the notification center, just check the sidebar, take a look at the inbox or sent messages, and you’re done. Alerts are sent according to your preferences: web, SMS, email, Slack, or Microsoft Teams. One thing: the client-side administrator must first configure the outgoing mail server, otherwise emails will not go through. Rapid anomaly detection is essential in modern agriculture; there is no other solution. That’s why these email alerts are not just routine, but are crucial. These alerts allow the farm owner to be immediately notified if the system detects something abnormal in the telemetry data stream (v1/devices/me/telemetry), so that they can act quickly and protect their crops.

Figure 13 clearly illustrates the situation. When you examine the telemetry data, you see exactly what is happening: the flame sensors are reading 42.19% and the smoke sensors are climbing to 92.74%. The processor is running at just 0.5% of its capacity and memory usage is 0.16 GB. But here’s the most important thing: the system is reporting a fire. There’s no time to waste. This isn’t just a routine update; it’s a red alert. Farmers need to be notified immediately so that an action can be taken quickly and the problem can be readily fixed before its gets worse. .

The research discussed in this paper includes accuracy of detection, speed of response, reliability, and false-alarm rate. The primary research tool for the experiment consisted of two environmental sensors (smoke and flame detectors) connected to a Raspberry Pi. The Raspberry Pi sent its internal data, which contained information about the usage of the CPU and memory, to the ThingsBoard cloud platform using the MQTT protocol. The Raspberry Pi transmitted data at fixed two-second intervals for near real-time monitoring. The overall data transferred on the Raspberry Pi had a total data transfer rate of roughly two messages per second by considering all of the data sent. There were 50 experimental scenarios used to assess this system’s capability to identify and accurately classify fires (i.e., to minimize false alarms) based on a total of 35 simulated fires and 15 non-fires. In 35 simulated fire scenarios, 34 were detected, resulting in an accuracy of 96.1% for simulated fire scenarios. The accuracy was measured as the proportion of correctly classified events, relative to the total number of experimental scenarios, which was set at n = 50. This implies the system has the ability to detect emergencies. The evaluation of 15 scenarios for non-fire cases resulted in a low false alarm rate of 2.8%. This implies the system has the ability to withstand climatic changes and differentiate between natural environmental changes and actual fire occurrences. The mean transmission latency, which is the time taken from the collection of data to its display on the cloud dashboard, was determined to be below 300 milliseconds, thereby fulfilling the set criteria for its response in detecting fire in agricultural environments. During longer term assessments of system reliability, it exhibited greater than 98% of the time the system was accessible with no prolonged downtimes or data loss, thereby displaying stable and ongoing use. In addition, an assessment of the telemetry data from the Raspberry Pi’s on-board sensors, indicated that both CPU and memory usage remained moderate and continued to be stable despite the simultaneous upload (to the cloud dashboard) of environmental sensor data and system parameters, indicating that the architecture developed can support constant monitoring. Based on all of the preceding data, the proposed system provides a viable alternative for early fire detection within an agricultural setting, to be functional, reliable, measurable, and repeatable.

Researchers have explored all kinds of solutions to help detect and fight forest fires. Some have focused on remote detection systems that make data collection much more efficient²². UAVs, essentially drones, are increasingly emerging as a practical solution for real-time fire prevention and monitoring, especially since they can cover large areas without costing a fortune²². Another group tested a vision-based system for early fire detection. This group used drones equipped with OAK-D cameras and a lightweight model called NanoDet to spot and locate fires, then binocular stereoscopic vision to determine how far away those fires were, as shown in a recent study²³. Deep learning techniques are also becoming more prominent in this field. A recent study offered a new approach to ensemble learning to improve detection rates, thus providing a more intelligent and faster way of detecting fire in various situations²⁴. The research field of fire safety is not limited to forests, as another study focused on fire safety in Moroccan university hospitals, developing a fire alarm control and monitoring system based on Arduino Uno to ensure a safer and less chaotic process of evacuation during emergency situations²⁹. In a related field, research on detecting harmful metals such as cadmium and lead in leaves has been conducted. The research used leaf clips to obtain hyperspectral data, identifying the optimal location for detecting these metals, namely, the base of the leaf for cadmium and the tip of the leaf for lead³⁰. In another direction of addressing the problem of false alarms, a research study analyzed five years of data to improve fire detection through IoT devices. The research offered a fuzzy logic approach to identify fire patterns and alert the authorities 30 s prior to the fire, reducing false alarms by 80%³¹. Another research focused on fire safety in the forest environment, utilizing smart sensors and Internet of Things technology to identify abnormal states, transmit the data to a software module, and apply a convolutional neural network to increase the speed and accuracy of the process³². Furthermore, research³⁵ proposed a new approach for the automated early fire detection system, utilizing Internet of Things technology to send alerts to the authorities in the event of a fire. This system also explores the advances and applications of IoT in firefighting and presents the design of an Arduino-based fire detector equipped with smoke and temperature sensors, as well as software tools such as the Arduino IDE and EasyEDA. Another study explores early detection of forest fires using the IoT in combination with YOLOv5³⁶. The results show that IoT devices can validate fires detected by YOLOv5 in this study³⁶. Furthermore⁴⁰, highlights the urgency of prevention and presents a solution using an ESP32 board and various sensors for the early detection of forest fires. Furthermore, this paper⁴² presents an IoT-based monitoring system for soil parameters, which aids irrigation management and pest control. Together, these studies demonstrate a global effort to develop effective real-time monitoring systems for environmental sustainability and safety.

As illustrated in Table 4, a wide variety of configurations and technologies have been methodically examined in this in-depth assessment of embedded systems in a professional setting based on a compilation of references covering the latest years^{22–24,30–32,35,36,40,42}. Crucial elements covered by the analysis include the embedded controller, IoT integration, use of cloud computing, data analysis capabilities, communication protocols, email notification functionality, and processor and memory utilization considerations, in addition to the programming languages used. The authors of this paper²² have proposed a fire detection system. The system uses embedded controllers, such as the Raspberry Pi, and does not use cloud computing, data analysis, email alerts, communication protocols, full specifications for CPU and memory consumption, or explicit programming language information. Comprehensive information on embedded systems, such as the Internet of Things, cloud computing, data analysis, communication protocols, e-mail alerts, CPU and memory usage, and programming languages, is not used in these studies^{23] and [24}. This paper²⁵ uses IoT Arduino and the platform, although it lacks cloud computing and data analysis functions. IoT is included in another system planned for 2023, but it does not adequately address other important issues³⁰. The paper³¹ proposes a fire detection system using a Raspberry Pi as a controller and integrating IoT. However, it intentionally does not use cloud computing, data analysis, specified communication protocols, and email notification functions. Details of CPU and memory usage were also not addressed, while the programming language used was MATLAB software.

In this study, a LoRa-based system was identified that maintained a similar approach, integrating IoT while not incorporating cloud computing, advanced data analysis, specific communication protocols, or email notifications. Another study ³⁵ describes a Raspberry Pi system with IoT integration, excluding cloud computing, data analysis, specific communication protocols, and automated e-mail alert. In this case, the programming language used was Python. The fire detection that uses the combination of Raspberry Pi and LoRa ³⁶ does not use any IoT technologies, cloud computing, even data analysis, specifically defined communication protocols or email notifications. An important consideration in all of these studies, processor and memory usage are never mentioned, and these are both important for confirming the scalability of solutions and the operational reliability of these systems when deployed in real world applications. In most of these implementations, the specifications for CPU and memory usage have not been provided as the programming language used was Python. Two articles were published on ESP32 systems ^40,42. The research in ⁴⁰ provided integration with Ubidots for cloud computing, but did not provide any data analysis, specify communication protocols, e-mail notifications, or provide any details on processor or memory usage. The development in ⁴² used ThingsBoard for cloud computing and added a prototypical communication method with Arduino programming, but again did not provide any performance metrics for the system at the end of the telemetry pipeline, including no specifications for end-to-end telemetry pipelines or automated notifications.

In contrast, The Proposed Scheme (PS) is an entire, experimentally validated prototype, functionally integrated, embedded solution utilizing a Raspberry Pi 3 B+ computer system which integrates to the ThingsBoard Cloud platform allowing for easy scalability of data management along with deep integration of IoT technology. In contrast to previous studies utilizing multiple disparate software applications located in different geographic locations to collect data about fires, the PS is a integrated architecture containing all required functionality: (a) real-time fire detection using flame and smoke sensors, (b) telemetry transmission of the collected data via the MQTT and JSON protocols, (c) automatic anomaly notification via e-mail when an anomaly (i.e., fire) is detected, and (d) continuously monitoring CPU and memory utilization for robustness in real-world situations. Additionally, the PS exemplifies a holistic approach to embedded systems design by designing and developing a compatible system featuring a modern technical development platform. The PS allows you to easily manage the different types of functions utilized in these systems utilizing the Python programming language. In summary, a proposed solution is provided to assist with the growing problem of wildfires affecting agriculture via a single integrated system utilizing IoT devices, embedded computer systems, and cloud computing technologies. The significance of this research is how unique the methodology for continuous monitoring, detecting fires early and very quick response is through telemetry data. The proposed architecture comprises three main levels: (1) an IoT device equipped with various environmental sensors, including smoke and flame detectors; (2) a perception level with a general-purpose input/output (GPIO) data interface via a Raspberry Pi B + for signal processing; and (3) telemetry data processing via the Paho MQTT client, enabling real-time transmission of sensor data to ThingsBoard. The above three tier architecture allows all flame sensor readings, smoke sensor readings, CPU usage and memory usage to be monitored in real time on a ThingsBoard dashboard. In addition, telemetry data is transmitted to cloud (ThingsBoard) via the MQTT protocol for further detailed analysis and presentation. The combination of the environmental Sensors and the system resource monitoring as described above makes the proposed system distinctly different than other documented systems. Finally, the innovative aspect of this research is combining cutting-edge IoT sensor technology, telemetry technology and cloud analytics onto a single platform therefore providing a highly reliable and resilient farming system.

In response to the limitations that have been identified in previous research studies concerning inadequate resource monitoring, insufficient autonomous notification systems, and suboptimal cloud integration, this research proposes a validated framework for intelligent fire detection systems for agricultural purposes. Unlike previous research studies that have been limited to conceptual frameworks and qualitative evaluations without quantitative testing and validation, this research focuses on quantitative testing and validation. The research findings validate the efficacy and accuracy of the intelligent fire detection system for detecting fires in agricultural residue and achieving an accuracy rate of 96.1%, with false alarm rates standing at 2.8%. The response time is less than 300 milliseconds, which makes it suitable for real-time monitoring. In addition, the intelligent fire detection system is reliable as indicated by its reliability metric that is above 98%. Moreover, the intelligent fire detection system does not have significant overload effects on the Raspberry Pi B+, which means that the incorporation of the telemetry subsystem into the intelligent fire detection system has no significant effects on its performance. The proposed architecture, in comparison to the previously used conceptual or partially integrated approaches, has the advantage of having quantified performance and reproducibility in controlled experimental environments. Although room for improvement exists, for instance, in terms of improving its connectivity and reducing its energy requirements, the intelligent fire detection system is useful and makes significant contributions beyond merely detecting early signs of fire in agricultural residue. The intelligent fire detection system is useful and makes significant contributions not only in terms of detecting early signs of fire in agricultural residue but also in terms of global initiatives geared towards raising awareness about forest fires and their effects on food security and, in general terms, enhancing the sustainability and resilience of agricultural systems.

From a scalability point of view, the current evaluation process used only two environmental sensors connected to a single Raspberry Pi 3 B+ node with a limited telemetry workload. The response time (< 300 ms), system availability (> 98%), and CPU/memory usage all suggest that the system is computationally lightweight. However, the above-mentioned results must not be considered a comprehensive scalability validation process. A wide range of scalability validation is necessary, including the use of multiple distributed IoT nodes under varying conditions. In all the experimental scenarios used in the current study, the telemetry data acquisition process includes not only the flame and smoke detection sensor values but also the CPU/memory usage statistics (minimum value, maximum value, average value, and cumulative value). This will allow the reproducibility of the experimental setup under the same conditions. The selection of the ThingsBoard and MQTT protocol is based on the following factors: interoperability, open-source implementation, and cost-effectiveness. These factors are particularly relevant in agricultural applications. Although the current prototype is technically feasible, further validation is necessary before the system is deployed on a large scale.

In addition to the hardware considerations, a significant degree of attention was devoted to the management of the software and data management to ensure the stability of the system over a long period of time. The Raspberry Pi B+ runs a lightweight Linux distribution and buffers the sensor information locally before sending the information to the cloud-based ThingsBoard system using the MQTT protocol. In order to ensure that the system does not become saturated with information and therefore run out of memory and processor speed, a routine system of deleting the information once it has been ingested and validated into the cloud system has been implemented. This configuration ensures that the processor and memory usage are constrained, thus promoting maximum efficiency and speed over an extended period of time. The assessment of the system’s memory and processor capabilities has been validated. The ongoing smart farming initiative has integrated an innovative fire detection system that has been tested and is capable of transmitting data in real time using the MQTT protocol, as well as visualizing data using the ThingsBoard platform. Although the system has been successful in delivering results, several constraints have been identified that need consideration. The first constraint is related to increased dependency on Internet access to transmit data to the ThingsBoard platform. This is considered a significant constraint, especially in areas where bandwidth is low. The second constraint is related to the human aspect of the system, which is crucial in ensuring that the smart fire detection system is successful. This is because, in order to ensure that the system is successful, the farmer is expected to understand the information provided through the cloud platform, such as information communicated through email. Therefore, it is crucial that training be provided to farmers who are expected to use the smart fire detection system.

Conclusion and future work

The research proposes a system for the real-time monitoring of fire using cloud computing technology, which can be used for the implementation of sustainable agriculture. The objective of the research is to ensure the sustainability of agricultural activities through the reduction of risks associated with fire. This paper also outlines an intelligent system for smart agriculture, which has five major layers: 1) cloud computing, 2) the Internet of Things (IoT),  3) embedded systems, 4) telemetry data, and 5) sensors. The system architecture presented in this paper consists of three major components: first, the IoT device layer, which consists of a Raspberry Pi B+, a network gateway, a flame sensor, and a smoke sensor; second, the ThingsBoard cloud layer, which optimizes efficient cloud computing processes; third, the telemetry processing layer, which consists of a real-time dashboard, a rule engine, and a user interface. The sensors connect via an analog/digital converter called the MCP2308, where this connection has been optimally linked and will work well with the RPi to create a fully integrated system. The algorithm functions as a controller for the system and manages all the functions of that suggested system (integration, processing and transmitting all sensor parameters to the ThingsBoard cloud using the MQTT protocol). All telemetry values were transmitted to the ThingsBoard Cloud via MQTT in JSON format. The suggested system simulates and collects data from the fire sensors to send them to the ThingsBoard cloud in real-time for later analytical use. Additionally, when an anomaly occurs the farm owner receives an email notification containing telemetry values. The quantitative assessment of the system indicates its efficacy and dependability in the detection of fire early in agricultural settings. It is clear from the results that there is a high degree (96.1%) of precision achieved in detecting naturally occurring fires, as well as a low rate of 2.8% of false alarms produced by the method employed. Additionally, since the average time taken for the transmission of response was less than three hundred milliseconds and the operational availability was greater than 98%, these features support operation at a consistently high level of performance, with near-instantaneous response times. Furthermore, by analysing the load on the Raspberry Pi B+ platform, it can be confirmedthat the use of CPU and memory data as part of integrating measurement and analysis would not impose a noticeable additional load on that platform; therefore, confirming the platform’s architecture is suitable for use in the implementation of large-scale applications. All of these findings provide a solid empirical basis for potential future work focused on reducing energy consumption and broadening deployment of this system. Unlike previous prototype systems that focused either on sensing or cloud transmission, this study presents an end-to-end experimentally validated prototype system that integrates fire detection, telemetry monitoring, and notification within a single IoT cloud framework. In this proposed study, a technically feasible framework for intelligent agricultural fire monitoring is proposed. Even though extensive large-scale validations are required to be performed on this proposed framework, the study suggests that it could be a technically feasible framework for intelligent agricultural fire monitoring.

Future work discussed here will concentrate mainly on improving and extending the existing prototype to support more extensive real-world agriculture uses. Specifically, the validation of the system will be a focus of future work; this involves validating the capability of the system at a greater number of IoT nodes under a variety of different networking conditions, so as to provide evidence of the system’s scalability, utility, and functionality within a variety of agricultural contexts. Another key area of work will be energy efficiency by measuring and integrating the power consumption from both the Raspberry Pi and various sensors, as well exploring low-power options for microcontrollers or LPWAN technologies (i.e., LoRa and NB-IoT) to allow for autonomous operation in remote or non-grid-districts. Furthermore, additional research will include an analysis of local buffered storage and asynchronous transport strategies to ensure service continuity during an internet failure. Furthermore, the present study anticipates exploring multiple ways to modify the system making it adaptable for growing crops in a variety of climates and building on the resiliency of the platform. It will also consider using renewable energy sources such as solar energy in order to improve autonomy of the system as well as reduce its carbon footprint. Finally, large-scale pilot implementations of the system will be done in order to measure reliability, response times and end-user acceptance; this will provide substantial pilot data that can be utilized when providing guidance for future expansions of the system and to demonstrate that the system has matured into an intelligent, secure and sustainable agricultural solution that supports global food security and resiliency against climatic changes.

Author contributions

A. Morchid: Conceived and designed the experiments; Performed the experiments; Analyzed and interpreted the data; Contributed reagents; Materials; Analysis tools or Data; Writing – original draft.A. Salami: Analyzed and interpreted the data; Materials; Analysis tools or Data; Writing – review & editing.H. M. Khalid: Analyzed and interpreted the data; Materials; Analysis tools or Data; Writing – review & editing.Z. Said: Analyzed and interpreted the data; Materials; Analysis tools or Data; Writing – review & editing.

Funding

This research received no external funding.

Data availability

Data are available on request from Abdennabi Morchid (abdennabi.morchid@usmba.ac.ma).

Declarations

Competing interests

The authors declare no competing interests.