SMART-CYPS: An Intelligent Internet of Things and Machine Learning Powered Crop Yield Prediction System for Food Security

SMART-CYPS: An Intelligent Internet of Things and Machine Learning Powered Crop Yield Prediction System for Food Security | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Short Report SMART-CYPS: An Intelligent Internet of Things and Machine Learning Powered Crop Yield Prediction System for Food Security Martin Kuradusenge, Eric Hitimana, Kambombo Mtonga, Antoine Gatera, and 5 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3834903/v1 This work is licensed under a CC BY 4.0 License Status: Published Journal Publication published 05 Nov, 2024 Read the published version in Discover Internet of Things → Version 1 posted 11 You are reading this latest preprint version Abstract The sub-Saharan Africa region continues to experience food insecurity, a consequence of the less productive agricultural sector that has dragged to adapt to the effects of climate change. As the region’s population continues to grow, there is a need to modernize the region’s agricultural sector to meet the increasing food demand. Although extreme atmospheric conditions cannot be entirely mitigated, however, the integration of technologies such as the Internet of Things (IoT) and Machine Learning (ML) can increase the quantity and quality of production from the crop fields. These technologies have potential to empower agricultural management systems to handle both climatic and farm data in an orchestrated manner, informing formulation of effective strategies. This study presents the design and development of a system for predicting crop yields that integrates IoT and ML. The system combines current weather data and historic crop yield data to predict seasonal crop yields. The weather parameters including, rainfall, temperature, humidity and soil moisture are collected by IoT sensors and transmitted to the cloud for crop yield forecasting. The system is used to analyze seasonal yields of Irish-Potato and Maize in Musanze District of Rwanda. Using data over different agricultural seasons, the system achieved favorable predictive accuracy with mean absolute percentage error (MAPE) values of 0.339, 0.309, and 0.177 for two seasons of Irish potatoes and one season of maize, respectively. Such predictive yield systems can reduce food insecurity risks and enhance harvest efficiency by enabling early awareness of crop production, fostering effective strategies shared among decision-makers and stakeholders. While maize and Irish potatoes were the initial case studies, expansion to include other crops is envisioned. IoT machine learning prediction weather climate change crop yield maize irish potatoes Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Figure 7 Figure 8 Figure 9 Figure 10 Figure 11 1. Introduction According to the World Bank, the number of people facing food insecurity increased from 135 million to 345 million from 2019 to June 2022, due to the Ukraine war and the economic fallout from the COVID-19 pandemic that caused food prices to rise extremely [ 1 ]. Food insecurity is a result of several causes, including weather-related hazards, a lack of input resources, crop diseases, regional or international wars, and threats from both intentional and inadvertent activities [ 2 , 3 ]. Weather-related threats include, high temperatures, water shortages, floods, droughts, and higher atmospheric CO2 concentrations. For instance, Corn and wheat production have gone down recently because of bad weather, plant diseases, and a lack of water around the world [ 4 ]. Declining crop yields, particularly in the areas with the greatest food insecurity on the globe, will cause more people to live in poverty if there are no remedies. With climate change, more extreme weather occurrences are anticipated, which could paralyze agricultural productivity [ 5 ]. It is projected that by 2030, 43 million people in Africa alone will be living in poverty [ 1 ]. On a global scale, investments in technology development, adaption, and transfer might significantly lessen these effects [ 6 ]. To cope with the issue of food insecurity, the government of Rwanda has established different strategies including the Crop Intensification Programme (CIP), Livestock Intensification Program (LIP), Land Use Consolidation Program (LUCP), and others [ 7 ]. The CIP and LUCP among others were prioritized programs focusing on specialized crops with a target of increasing the agricultural production and food security in Rwanda [ 8 , 9 ]. However, because most of the farming activities are predominately rain-fed, Rwandan agriculture is extremely vulnerable to climate change and extreme weather events like droughts, floods, and severe storms [ 10 ]. To counter the impact of the extreme weather events on food production, the Rwanda government has put in place different measures such as the promotion of drought-relieving irrigation methods. Under this initiative, local communities receive training in low-cost irrigation techniques such as rainwater collection, water storage, water pumping from water bodies, drought-resistant seeds, and food storage, where communities preserve extra harvest that can be used during drought periods [ 7 ]. However, rainfall deficits and rainfall induced disasters strike different regions in Rwanda leading to food insecurity being experienced in multiple regions of the country. For instance, eastern province is vulnerable to droughts due to irregular rainfall and long dry season that result in the shortage of food [ 11 , 12 ]. On the other hand, the hilly and mountainous regions of Rwanda (namely north and west) are characterized by plenty of rain especially during the March – May rain season. As a result, floods and landslides are frequent, leading to poor crop harvest [ 13 ]. Even though it would be difficult to fully control climate conditions, technology use can reduce the magnitude of associated risks including food crisis. Historical weather and crop yields data can be used to predict the future production levels of different crops. This knowledge can be useful to policy makers enabling them to set some strategies aiming at reducing the risks of food insecurity. At present, fully documented studies done on the prediction of crop production using combined weather and crop yield data are scarce in Rwanda. Also, in some cases where irrigation is done, there is no control of how much water content the soil needs according to the growth stage of the crops. Therefore, there is need to use IoT systems to monitor the water content in the soil so that irrigation is done when needed. The use of IoT would promote proper use of water but also minimizes risks of water lodging due to excessive irrigation. In this work, a model integrating IoT and ML is proposed that is informed by policy maker’s perspective but also integrates farmers’ context thereby setting to contribute to a more efficient strategy for technology testing. The model uses historical yield data and current weather data to predict crop yield. The IoT system captures real time weather data which is fed into the Machine learning model for prediction purposes. A prototype of the system is developed and deployed on the farm for real-time data capturing and crop yield prediction using a robust prediction model. This system responds directly to the need for data driven approaches to policy making in the agricultural sector to ensure food security, especially in low-to-middle income countries. By capturing real-time weather parameters, the system has potential for use beyond crop prediction as it can also be used to optimize irrigation and water use in Rwanda. Furthermore, by using weather and crop yield data, the model makes it possible to analyze the relationship between crop production and climate change with time. Specifically, the following are the contributions of this work: Presents the designing and development of a crop yield prediction system called SMART-CYPS that integrates IoT and machine learning. Using collected real-time weather data and historical crop yield data, we analyze Irish-Potato and Maize seasonal yields for Musanze District in Rwanda. 2. Related works Emerging technologies continue to play a central role in solving different problems, including those related to the basic needs of people, such as food security. Various studies used IoT and artificial intelligence (AI) to provide solutions to the problems related to insufficient crop yield. The optimization of crop production may start with matching of crops to specific weather condition. For instance, the study conducted by Amna I. et al. [ 14 ] used IoT and ML for smart selection of crops to be planted in a specific farm by monitoring different parameters such as rainfall temperature, humidity, nitrogen, potassium, pH, and carbon dioxide (CO2). Since each crop has different nutrient requirements, the designed system gathers data from the soil, and with ML algorithms, the appropriate type of crop is recommended. Similar studies have been conducted using the same parameters in [ 15 – 17 ]. The quantity and quality of crop production can also be enhanced by predicting crop diseases using IoT and AI. In their study, Ravesa A. et al. [ 18 ] proposed a prediction model for apple orchard diseases using data analytics, and ML in an IoT system. They fed a linear regression model using real-time data collected by wireless sensor nodes and IoT nodes. The design consists of several IoT sensor nodes placed across the apple orchard and a nearby gateway for gathering data from the nodes' mesh. A network was created between the nodes, and a nearby gateway. The fog node collects data from a certain number of nodes in the network. For each location, data analysis is performed in real-time to enable prompt responses as necessary. The application might provide farmers with real-time updates on the health of their apple orchards, allowing them to take immediate action against any potential illness like scab if it is accurately and properly forecasted. On the other hand, optimizing crop yield can be done through monitoring various influencing parameters, such as those related to weather and soil properties. The system designed by [ 19 ] monitored daily rainfall and temperature for the prediction of the crop harvest. The architecture of the system consists of a peasant farming layer where the farmer uses a mobile application to input the seasonal crop yield into the system for storage. The same application is used by the farmer to visualize the predicted crop yield as well as weather data that are remotely collected by IoT sensors. The second layer of this system is a wireless sensor network that consists of ultra-low power transceiver modules that are used for communication among the sensor nodes. IoT-based hardware is the third layer of the system. It was made of a general-purpose microcontroller connected to the transceiver module, which is the Global System for Mobile (GSM). The last layer was the cloud computing platform, which comprises four cloud computing services: an IoT message hub for enabling two-way communication between system components; data storage service for storing data from sensors; data analytics service for data mining; and a machine learning service for the implementation of predictive analytics for crop production. An IoT ecosystem for smart agriculture framework proposed by Vu K et al. [ 20 ] consists of four main components. The first component which consists of IoT devices for monitoring the farming environment, comprises sensors to gather information such as weather-related data and actuators to enable physical action as a result of decisions based on the information coming from sensors. The second component of this framework is the communication technology, which consists of infrastructure and communication protocols that are responsible for data transfer from the sensor node to the cloud. The third component of this framework is the cloud, which comprises database and server applications responsible for data processing and storage, big data analytics, and decision making. The fourth section is the end user (farmer or system owner) who interacts with the system using an application. Other studies looked at how to anticipate agricultural yields using remote sensing. Those that have been in practice for a while use satellite and drone images for crop harvesting decisions [ 20 – 24 ]. The monitoring of crops on a global and regional scale is currently greatly aided by better new technologies [ 21 ]. As highlighted by the above cited works, IoT and ML are key technologies in the realization of food security. IoT supports real time monitoring of crops and weather parameters, whilst ML algorithms can use both sensed and historical data to predict crop yields. Knowledge of anticipated harvest empowers the farmers to plan ahead, e.g., by searching for markets. Famine early warning is another application of crop production prediction, which allows decision-makers like the government to establish a strategic plan to address the impending food insecurity. The above highlighted studies have made attempt to use IoT and ML technologies for addressing various aspects of crops and plants growth. However, unlike previous studies, this work focuses on crop yield prediction using current weather data and historical crop yield data. The IoT sensors are responsible for capturing weather data and feeding it into the ML model which combines it with historical meteorological and crop yield data for yield prediction purposes. Considering the impact of climate change on Rwanda’s agricultural sector, we believe the develop system will support various initiatives being undertaken by the government to ensure food security in the country. 3. SMART-CYPS – System Design and Development 3.1 Methodology Data collection and modeling This study builds on the findings reported in [ 25 ]. In [ 25 ], Kuradusenge, et al. conducted a study that analysed the impact of climate conditions on crop production. As part of the study, historical data about meteorology and crop yields (i.e., for Irish Potato and maize) were collected and using this data modelling was performed to determine the best ML model to be implemented for crop yield prediction. Furthermore, the study identified the optimum values of weather parameters for optimum yields. Therefore, this study uses the values from that study to develop a crop yield prediction system that integrates IoT and ML. 3.2 System archtecture The architecture of the developed crop yield prediction system is shown in Fig. 1 . The data used by the system for prediction comprises secondary data and primary data. The secondary data includes, crop yield data and climatic data, i.e., rainfall, temperature and humidity but also crop yield data. The secondary data can be accessed from various databases whilst the primary data is continuously being collected by the IoT sensors and being fed into the system. As can be seen from the Fig. 1 , the sensors collect data for two purposes, namely; for prediction and storage. All real-time data is fed into the deployed model for harvest prediction. Informed by the findings in [ 15 ], the ML model trained for prediction is the Random Forest regressor. The model is configured in a such way that, it predicts the harvest basing on the crop growing stage or monthly data batches and not basing on single data prediction. 3.3 SMART-CYPS – IoT sub-system System design and functionality The IoT sub-system of the SMART-CYPS system comprises various sensors that collect data related to climatic conditions including, humidity, temperature, rainfall, and soil moisture. The system further has sensors that monitors system Current which help in determining the battery status. Other components include, the GSM module used for communication, and GPS module used for locating the node. Table 1 provides the details of all the electronic components of the IoT sub-system. Table 1 Description of components used for the development of the CYPS Component Description Function Operating Voltage ATMEGA328P-PU Microcontroller IC Performs data processing function 1.8–5.5 V DHT22 Digital Temperature and Humidity Sensor module Measuring the temperature & humidity of the surrounding air 5V FST100-2007 RS485 Modbus 24V DC Rain and Snow Sensor Detects rainwater Measures whether there is rain or snowfall outdoors 10-30V-DC Soil moisture sensor Soil moisture VWC water content test sensor probe with 3 pins Measuring the soil water content 5-30V-DC Voltage Sensor Voltage detection module Monitoring the status voltage saved by the battery powered by the solar panel 5V GY-NEO6MV2 GPS Module Detecting the location of the sensor node 3-5V SIM800L GSM/GPRS module Connecting the device to the GSM network 4.2V All components described in the table above are mounted on the printed circuit board (PCB) allowing the data collection and processing. Figure 2 shows the circuit design. The powering system is made of three main components: solar panel (20 W), Battery, and charge controller. Figure 3 shows the system prototype including the powering components. System communication Figure 4 shows the communication architecture of the SMART-CYPS IoT subsystem. As pointed out above, the IoT data are collected using sensors (sensor nodes labelled as Edge Layer in this context) that transmits the data to the cloud through GPRS protocol within the telecommunication cellular network. To ensure security of the data, the sensors transmit the data to the cloud with associated tokens. To manage the data ingestion from heterogenous devices, the data is sent and received in JavaScript Object Notation (JSON) format, the sample is shown in the diagram. The past production information and weather conditions are also stored in the server where data mining techniques will be used to produce a report showing the future production predicted by the system. To ease the computation by avoiding model testing overhead, the cloud services were developed under container paradigms as microservices. This scheme proved to be efficient while in Docker environment. An authenticated farmer/policymaker can be able to interact with the system to see those reports and predictions. The developed web solution can predict the crop yield depending on the current data from the sensor nodes and the historical harvest. This prediction is due to the integrated model in the cloud services. 3.4 SMART-CYPS – Machine learning model performance evaluation This work implements the Random Forest regressor based on its best performance in crop yield forecasting as reported in [ 15 ]. We adopt the Mean Absolute Percentage Error (MAPE) as the performance evaluation metric, which is calculated using the following formula: . $$MAPE=\frac{1}{n}{\sum }_{i=1}^{n}\left|\frac{{A}_{i}-{P}_{i}}{{A}_{i}}\right|100$$ 1 , where MAPE is the Mean Absolute Percentage Error, n is the total number of data points (or observations), Ai represents the actual or observed values, and Pi represent the predicted values for the corresponding observations. 4. Results 4.1 Real time data gathering and visualization The designed and developed system was deployed in the study area after six months of testing. The IoT system daily collects weather and soil humidity data from the farms and sends it to the cloud for long-term storage and future analysis. Figure 5 is one of an IoT node deployed on the farm. The system dashboard displays the real time data coming from three of the sensor nodes. In the back end the rest of the data are stored on daily basis. The frequency of data transmission from sensor node is 15 minutes. As shown in Fig. 6 , in real-time, the data can be visualized on the system dashboard. The figure also shows the data summary and at the end of the day the minimum, maximum and average data are saved in the database. At the end of the season, the rest of the data together with actual crop yield are added to the historical data. As shown in Fig. 6 , the system has different modules. Below is a description of each of the module: Managing users: used to manage users such as adding, removing, and assigning roles. Cooperative module: used to visualize or add the name of farmers cooperatives that will have access to the information in the system. Roles: used to add the visualize or add the managing roles. Devices: module is used to track the device: it provides the GPS coordinates and location on the map. This module also provides the battery (voltage) level. Crop category: provides the option to select or add crop types for yield prediction. Data visualization: comprises historical data for weather (rainfall, temperature, and humidity), and harvest. Since the system’s deployment, sensors have continuously captured and recorded the data in the database. Figure 7 shows the variation of data over time since the deployment of the system prototype. It is further possible to track location of the deployed devices and monitor battery life. As illustrated in Fig. 3 , the system relies on a solar panel system for power, a necessity in remote areas without access to the main power grid. The web application is used to keep tabs on the status of the battery power. Figure 8 portrays the voltage variations for the period of one year and half. The figure illustrates a decrease in voltage during the rainy seasons in Rwanda, spanning from March to May and September to December. It was common to observe interruptions in data transmission during these periods. 4.2 Correlation of weather parameters To identify the correlation among the weather variables, we utilized the data collected the system throughout the study's duration. The correlation matrix depicted in Fig. 9 , reveals several significant correlations. Specifically, there is a moderate positive correlation of 0.51 between rainfall and soil moisture, a moderate positive correlation of 0.41 between rainfall and humidity, and a small negative correlation of − 0.25 between rainfall and maximum temperature. Furthermore, maximum temperature exhibits a moderate negative correlation of − 0.36 with soil moisture, and a moderate negative correlation of − 0.58 between humidity and maximum temperature. Additionally, a positive correlation of 0.48 was observed between soil moisture and humidity. 4.3 Prediction results and the actual yield for Irish potatoes and maize Currently, SMART-CYPS monitors the cultivation of two crops: Irish potatoes and maize, which are rotated between seasons A and B in the study area. Typically, maize is planted during season A, and Irish potatoes are cultivated in season B. Since the system's deployment, we have monitored one season of maize and two seasons of Irish potatoes. In this area, maize has a growth period of six months, while Irish potatoes are harvested after four months. The subsequent tables (Table 2 , Table 3 , and Table 4 ) present the predicted and actual production for each crop during three agricultural seasons. The system forecasts crop yields at various growth stages. Table 2 Comparison of the yield predicted by CYPS and the actual yield for Irish potatoes (Season B 2022) Prediction time (growing stage) Predicted yield (kg ha-1) Actual yield (kg ha-1) Accuracy (MAPE) 21 days (sprout development) 12,109 8,717 0.339 2nd month (vegetative & tuber initiation) 11,373 3rd month (tuber bulking/filling) 11,639 4th month (maturity) 11,594 The growing stages considered for Irish potatoes are sprout development, vegetative, tuber initiation and bulking, and maturity [ 26 ]. The initial prediction was conducted at the end of the sprout development phase of Irish potatoes, while the last occurred during maturity. As shown in the Table 2 . Comparison of the yield predicted by CYPS and the actual yield for Irish potatoes (Season B 2022), the actual yield for Irish potatoes in the 2022 season B was 8,717 kg ha -1, whereas the predicted yield ranged from 11,373 kg ha -1 to 12,109 kg ha -1. The prediction error (MAPE) for this season was 0.339. In the agricultural season B of 2023, the crop yield was 9,837 kg ha -1, and the predicted yield varied from 11,985 kg ha -1 to 13,529 kg ha -1, with a prediction error of 0. 309.. Table 3 Comparison of the yield predicted by CYPS and the actual yield for Irish potatoes (Season B 2023) Prediction time (growing stage) Predicted yield (kg ha-1) Actual yield (kg ha-1) Accuracy (MAPE) 21 days ADP (Sprout development) 13,529 9,837 2nd month (vegetative & tuber initiation) 12,898 0.309 3rd month (tuber bulking/filling) 11,985 4th month (maturity) 13,130 Table 4 Comparison of the yield predicted by CYPS and the actual yield for maize (Season A 2023) Prediction time (growing stage) Predicted yield (kg ha-1) Actual yield (kg ha-1) Accuracy (MAPE) 1st month (establishment & vegetative) 1,758 1,837 0.177 2nd month (vegetative) 1,222 3rd month (tasseling) 1,580 4th month (cob setting) 1,345 5-6th month (grain-filling & maturity) 1,651 Six growing stages were considered to predict maize production. Those are establishment, vegetative, tasseling, cob setting, grain-filling, and maturity [ 27 ]. The results in Table 4 are for agriculture season A of 2023 that started in September 2022. The predicted maize yield varied between 1,222 kg ha-1 and 1,758 kg ha-1 whereas the actual maize yield was 1,837 kg ha-1. The performance of the prediction was 0.177 MAPE. 5. Discussion Crop yield is a consequence of a range of parameters, which can be broadly classified into two categories: factors within human control and those influenced by natural forces. Human-controllable factors encompass various elements, including the choice of crops or seed variety tailored to the specific region, planting schedules, soil preparation, the application of fertilizers, irrigation methods, weed management, and the use of pesticides, among others. In the study area, these variables fall under the supervision of local agronomists who collaborate closely with farmers, offering guidance and support as needed. Therefore, the natural factors affecting crop yields are the subject of monitoring in this study. Weather data is collected for two primary purposes: predicting crop yields and creating a historical record for future reference. Predictions are conducted monthly, aligning with the various growth stages of the crops, as indicated in Table 2 through Table 4 . The prediction process commences following planting and is initiated by the responsible individual, such as the agronomist or the leader of the farmers' cooperative. 5.1 Rainfall and soil moisture The SMART-CYPS incorporates a rain sensor designed for rain detection. This sensor furnishes binary values (0 or 1) to signify the presence or absence of rain. The preference for a rain sensor over a rain gauge was driven by the unavailability of a rain gauge equipped with network capabilities for real-time rainfall data acquisition. To overcome this limitation, we adopted a strategy that involved utilizing rain and soil moisture sensors. This approach enables us to observe shifts in soil moisture levels during rainfall events and supplement this data with secondary rainfall information obtained from the meteorological office. During consecutive rainy days, the soil moisture increases even when the amount of rainfall is not substantial. This indicates that soil moisture depends not only on the daily rainfall but also on antecedent rainfall. The correlation between rainfall and soil moisture is shown in Fig. 10 5.2 Rainfall data from rain gauge vs rain detected by sensor node Upon comparing the rainfall data obtained from the rain gauge with that from the rain sensor, we discovered discrepancies between the two sources. For example, there were some instances where the system signaled rain, but the recorded rainfall data was zero (0). Contrariwise, there were dates when the rainfall data exceeded zero, yet the system did not detect any rain. These inconsistencies are depicted in Fig. 11 . The inconsistency is due to the locations of the device versus the location of the rain gauge. That means there are some cases where rain does not reach the entire area and this shows the need of increasing the distribution of rain gauges for more accurate information about rainfall in the study region. 5.3 Prediction performance During every growth stage, the SMART-CYPS computes yield forecasts using cumulative rainfall, average minimum temperature, and average maximum temperature from the start of that stage. It is important to note that humidity and soil moisture were excluded due to insufficient historical data. These parameters will be incorporated into the system once enough data is recorded. Table 2 to Table 4 demonstrates that predicted data varied across growth stages in response to weather conditions, while actual data remained consistent since it is recorded at the conclusion of the agricultural season. The results in Table 2 indicate an accurate prediction of Irish potato yields with a precision of 0.339, which is a commendable outcome. A similar level of model performance was maintained in the subsequent seasons, with a MAPE of 0.309 for predicting the yield of the same crop. The maize yield predictions for the 2023 agricultural season A demonstrated a significantly improved performance, as indicated by a MAPE of 0.177. Predictions were conducted monthly, a total of five times, utilizing the same parameters as those applied to Irish potatoes. In our study, maize is typically harvested six months after planting. During the initial five months, we conducted predictions because during the sixth month, the maize crop is no longer developing, as it is intended for grain drying purposes. The higher frequency of maize yield predictions might be attributed to the enhanced performance compared to Irish potatoes, which could also be attributed to the data quality used in the prediction process. The disparities between the forecasted and real crop yields may stem from additional factors, such as fertilizer usage, planting timing, and more. Consequently, enhancing the prediction capability may be achieved by incorporating these variables in future research. 6. Conclusions Predicting crop yields can significantly contribute to reducing food insecurity since such information empowers farmers and other agricultural stakeholders to take appropriate steps well before harvest. In this work, two sources of data have been used to predict crop yields, i.e., real time information provided by the IoT sensor nodes and historical meteorological and crop yields data. The developed SMART-CYPS system has been deployed in Maize and Irish potato farms, however, the system can be adapted for other crop types subject to availability of historical data on climate and crop yields, along with comprehensive information on the crop's weather requirements. In future endeavors, the SMART-CYPS will be further improved by incorporating real-time rain gauge data and expanding its parameters to include soil properties, fertilizer usage, and more. Declarations Funding: This research was funded through the grant offered by the University of Rwanda in the partnership with Sida (Swedish International Development Agency) under the programme UR-Sweden program. The grant supported all research activities such as data collection, purchase of equipment and materials, fieldwork, etc. The APC was also funded by the same grant. Conflicts of Interest: The authors declare no conflict of interest. Data Availability Statement: All data used in this study are accessible online via the portal of the African Centre of Excellence in IoT (ACEIoT) at the University of Rwanda. The data link can be found in the reference section [28]. Code availability: Not applicable Author Contributions: Conceptualization, M.K.; methodology, M.K. and E.H.; software, E.H. and J.N; validation, K.M.; formal analysis, E.H.; resources, J.H., A; data curation, J.N.; writing original draft preparation, M.K.; writing review and editing, K.M., J.H, P.R, and A.G; visualization, A.G.; project administration, M.K.; funding acquisition, M.K. All authors have read and agreed to the published version of the manuscript. References Worldbank, Climate Explainer: Food Security and Climate Change, Worldbank.Org. (2022). https://www.worldbank.org/en/news/feature/2022/10/17/what-you-need-to-know-about-food-security-and-climate-change (accessed February 28, 2023). A. Djurle, B. Young, A. Berlin, I. Vågsholm, A.L. Blomström, J. Nygren, A. Kvarnheden, Addressing biohazards to food security in primary production, Food Secur. 14 (2022) 1475–1497. https://doi.org/10.1007/s12571-022-01296-7. V. Gitz, A. Meybeck, L. Lipper, C. Young, S. Braatz, Climate change and food security: Risks and responses, 2016. https://doi.org/10.1080/14767058.2017.1347921. Worldwide Concern, How climate change increases hunger — and why we’re all at risk | Concern Worldwide, (2022) 1–7. https://www.concern.net/news/climate-change-and-hunger (accessed February 28, 2023). T.J. Troy, C. Kipgen, I. Pal, The impact of climate extremes and irrigation on US crop yields, Environ. Res. Lett. 10 (2015). https://doi.org/10.1088/1748-9326/10/5/054013. D. Tilman, C. Balzer, J. Hill, B.L. Befort, Global food demand and the sustainable intensification of agriculture, Proc. Natl. Acad. Sci. U. S. A. 108 (2011) 20260–20264. https://doi.org/10.1073/pnas.1116437108. N. Cantore, The crop intensification program in Rwanda: a sustainability analysis, Environ. Progr. United Nations Dev. Progr. United Nations. 44 (2011) 1–35. https://www.odi.org/sites/odi.org.uk/files/odi-assets/publications-opinion-files/7094.pdf. K. Kariyasa, Y.A. Dewi, Land consolidation, specialization and household diets: Evidence from Rwanda, J. Gender, Agric. Food Secur. 1 (2011) 1–22. A. Nsabimana, F. Niyitanga, D.D. Weatherspoon, A. Naseem, Land Policy and Food Prices: Evidence from a Land Consolidation Program in Rwanda, 19 (2021) 63–73. https://doi.org/doi:10.1515/jafio-2021-0010. L. Muneza, Droughts and Floodings Implications in Agriculture Sector in Rwanda: Consequences of Global Warming, Nature, Causes, Eff. Mitig. Clim. Chang. Environ. (2021) 13. http://dx.doi.org/10.1039/C7RA00172J%0Ahttps://www.intechopen.com/books/advanced-biometric-technologies/liveness-detection-in-biometrics%0Ahttp://dx.doi.org/10.1016/j.colsurfa.2011.12.014. M. Lydie, Droughts and Floodings Implications in Agriculture Sector in Rwanda: Consequences of Global Warming, in: S.A. Harris (Ed.), IntechOpen, Rijeka, 2022: p. Ch. 18. https://doi.org/10.5772/intechopen.98922. L. Nahayo, G. Habiyaremye, A. Kayiranga, E. Kalisa, C. Mupenzi, D.F. Nsanzimana, Rainfall Variability and Its Impact on Rain-Fed Crop Production in Rwanda, Am. J. Soc. Sci. Res. 4 (2018) 9–15. A.J. Van Bodegom, B. Satijn, Climate Change Profile, (2015). A. Ikram, W. Aslam, R.H.H. Aziz, F. Noor, G.A. Mallah, S. Ikram, M.S. Ahmad, A.M. Abdullah, I. Ullah, Crop Yield Maximization Using an IoT-Based Smart Decision, J. Sensors. 2022 (2022). https://doi.org/10.1155/2022/2022923. P.K. Priya, N. Yuvaraj, An IoT based gradient descent approach for precision crop suggestion using MLP, J. Phys. Conf. Ser. 1362 (2019). https://doi.org/10.1088/1742-6596/1362/1/012038. K. Bakthavatchalam, B. Karthik, V. Thiruvengadam, S. Muthal, D. Jose, K. Kotecha, V. Varadarajan, IoT Framework for Measurement and Precision Agriculture: Predicting the Crop Using Machine Learning Algorithms, Technologies. 10 (2022). https://doi.org/10.3390/technologies10010013. S. Bevinakoppa, S.K. Padaganur, V. Nidagundi, IoT Based Smart Prediction System for Crop Suitability, in: 2021 IEEE Int. Conf. Comput., 2021: pp. 174–179. https://doi.org/10.1109/ICOCO53166.2021.9673499. R. Akhter, S.A. Sofi, Precision agriculture using IoT data analytics and machine learning, J. King Saud Univ. - Comput. Inf. Sci. 34 (2022) 5602–5618. https://doi.org/10.1016/j.jksuci.2021.05.013. L.O. Colombo-Mendoza, M.A. Paredes-Valverde, M.D.P. Salas-Zárate, R. Valencia-García, Internet of Things-Driven Data Mining for Smart Crop Production Prediction in the Peasant Farming Domain, Appl. Sci. 12 (2022). https://doi.org/10.3390/app12041940. M. Dhanaraju, P. Chenniappan, K. Ramalingam, Smart Farming : Internet of Things ( IoT ) -Based Sustainable Agriculture, (2022) 1–26. O.N. Lungu, L.M. Chabala, C. Shepande, Satellite-Based Crop Monitoring and Yield Estimation — A Review, 13 (2021) 180–194. https://doi.org/10.5539/jas.v13n1p180. N. Torbick, D. Chowdhury, W. Salas, J. Qi, Monitoring Rice Agriculture across Myanmar Using Time Series Sentinel-1 Assisted by Landsat-8 and PALSAR-2, Remote Sens. 9 (2017). https://doi.org/10.3390/rs9020119. A.M. Ali, M. Abouelghar, A.A. Belal, N. Saleh, M. Yones, A.I. Selim, M.E.S. Amin, A. Elwesemy, D.E. Kucher, S. Maginan, I. Savin, The Egyptian Journal of Remote Sensing and Space Sciences Crop Yield Prediction Using Multi Sensors Remote Sensing ( Review Article ), Egypt. J. Remote Sens. Sp. Sci. 25 (2022) 711–716. https://doi.org/10.1016/j.ejrs.2022.04.006. T.S. Abdul-Jabbar, A.T. Ziboon, M.M. Albayati, Crop yield estimation using different remote sensing data: literature review, in: IOP Conf. Ser. Earth Environ. Sci., 2023. https://doi.org/10.1088/1755-1315/1129/1/012004. M. Kuradusenge, E. Hitimana, D. Hanyurwimfura, P. Rukundo, K. Mtonga, A. Mukasine, C. Uwitonze, J. Ngabonziza, A. Uwamahoro, Crop Yield Prediction Using Machine Learning Models: Case of Irish Potato and Maize, Agriculture. 13 (2023) 225. https://doi.org/10.3390/agriculture13010225. J.E. Obidiegwu, Coping with drought: stress and adaptive responses in potato and perspectives for improvement, Front. Plant Sci. 6 (2015). https://doi.org/10.3389/fpls.2015.00542. B.A.A. Zemba, S.Z. Wuyep, A.A. Adebayo, C.J. Jahknwa, Growth and Yield Response of Irish Potato ( Solanum Tuberosum ) to Climate in Jos-South , Plateau State , Nigeria Growth and Yield Response of Irish Potato Solanum Tuberosumto Climate in Jos-South , Plateau State , Nigeria Strictly as per the compliance a, Int. J. Plant Res. 2019 (2013) 1–7. http://journal.sapub.org/plant. Research data, (2024). https://aceiot.ur.ac.rw/SMART-CYPS_data.zip. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Published Journal Publication published 05 Nov, 2024 Read the published version in Discover Internet of Things → Version 1 posted Reviews received at journal 23 Jul, 2024 Reviewers agreed at journal 15 Jul, 2024 Reviews received at journal 10 Jun, 2024 Reviews received at journal 27 May, 2024 Reviewers agreed at journal 27 May, 2024 Reviewers agreed at journal 26 May, 2024 Reviewers agreed at journal 26 May, 2024 Reviewers invited by journal 19 Apr, 2024 Editor assigned by journal 21 Mar, 2024 Submission checks completed at journal 15 Mar, 2024 First submitted to journal 04 Jan, 2024 You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3834903","acceptedTermsAndConditions":true,"allowDirectSubmit":false,"archivedVersions":[],"articleType":"Short Report","associatedPublications":[],"authors":[{"id":279866544,"identity":"2ac57c6e-b337-4d10-809c-b8df01efdd92","order_by":0,"name":"Martin Kuradusenge","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA5klEQVRIiWNgGAWjYBACCRDBA2EzPmBgOECaFmYDkrWwSRClRXJG7sMPb9vq5Mz5Fx+r5qm5I8fPwPzw0Q08WqQl0o0l57YdNrac8SztNs+xZ8aSDWzGxjl4tMhJpDFI87YdSNxw44zZbR62w4kbDvCwSRPQwvybt62ufsON89+Kef4RoUVaIo0NaAtzgsH5HjZm3jYitEj2PGOznHPusOGGG2xAT/UdNpZsJuAXieNpzDfelNXJG5w//PDDm2+H5fjZmx8+xqcFSXMCAxM4gpiJUg4C/AcYGH8QrXoUjIJRMApGEgAAyJtM4wZGDxoAAAAASUVORK5CYII=","orcid":"","institution":"University of Rwanda","correspondingAuthor":true,"prefix":"","firstName":"Martin","middleName":"","lastName":"Kuradusenge","suffix":""},{"id":279866545,"identity":"7bd11b90-737a-4588-b9e8-585a1ff08f7e","order_by":1,"name":"Eric Hitimana","email":"","orcid":"","institution":"University of Rwanda","correspondingAuthor":false,"prefix":"","firstName":"Eric","middleName":"","lastName":"Hitimana","suffix":""},{"id":279866546,"identity":"0a10ee72-64ba-484a-9e81-aa8c41e003f1","order_by":2,"name":"Kambombo Mtonga","email":"","orcid":"","institution":"Training \u0026 Education Development Consulting","correspondingAuthor":false,"prefix":"","firstName":"Kambombo","middleName":"","lastName":"Mtonga","suffix":""},{"id":279866547,"identity":"0c0e561e-0667-4283-bd0d-1eb60dadc9c6","order_by":3,"name":"Antoine Gatera","email":"","orcid":"","institution":"University of Rwanda","correspondingAuthor":false,"prefix":"","firstName":"Antoine","middleName":"","lastName":"Gatera","suffix":""},{"id":279866548,"identity":"ada9bf1e-d5f7-4257-83e8-058990dac595","order_by":4,"name":"Joseph Habiyaremye","email":"","orcid":"","institution":"Narada Electronics","correspondingAuthor":false,"prefix":"","firstName":"Joseph","middleName":"","lastName":"Habiyaremye","suffix":""},{"id":279866549,"identity":"d481fbfa-9f54-49b7-8226-75eb522d277f","order_by":5,"name":"Jackson Ngabonziza","email":"","orcid":"","institution":"Bank of Kigali Plc","correspondingAuthor":false,"prefix":"","firstName":"Jackson","middleName":"","lastName":"Ngabonziza","suffix":""},{"id":279866550,"identity":"84e5fc6d-16ee-475b-a1c9-1becc5e314c5","order_by":6,"name":"Damien Hanyurwimfura","email":"","orcid":"","institution":"University of Rwanda","correspondingAuthor":false,"prefix":"","firstName":"Damien","middleName":"","lastName":"Hanyurwimfura","suffix":""},{"id":279866551,"identity":"c4cd7b76-12ab-4080-9595-4fefd330a6dd","order_by":7,"name":"Placide Rukundo","email":"","orcid":"","institution":"AfricaRice","correspondingAuthor":false,"prefix":"","firstName":"Placide","middleName":"","lastName":"Rukundo","suffix":""},{"id":279866552,"identity":"78f4235c-3166-4f65-a68c-551ff535399d","order_by":8,"name":"Angelique Mukasine","email":"","orcid":"","institution":"University of Rwanda","correspondingAuthor":false,"prefix":"","firstName":"Angelique","middleName":"","lastName":"Mukasine","suffix":""}],"badges":[],"createdAt":"2024-01-04 15:47:37","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3834903/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3834903/v1","draftVersion":[],"editorialEvents":[{"content":"https://doi.org/10.1007/s43926-024-00079-0","type":"published","date":"2024-11-05T15:58:11+00:00"}],"editorialNote":"","failedWorkflow":false,"files":[{"id":53123802,"identity":"7db2f355-014d-409b-be7c-5dedc62dbd53","added_by":"auto","created_at":"2024-03-20 22:47:56","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":54228,"visible":true,"origin":"","legend":"\u003cp\u003eSMART-CYPS – crop yield prediction system architecture\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-3834903/v1/9d7e350e64aedad685a88aaf.png"},{"id":53123803,"identity":"20b3c1e8-63db-4f8d-8bd2-483d513351fb","added_by":"auto","created_at":"2024-03-20 22:47:56","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":108046,"visible":true,"origin":"","legend":"\u003cp\u003eCircuit diagram\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-3834903/v1/c30874460bb921966533420d.png"},{"id":53123801,"identity":"0a1e1690-9383-430b-a79c-59343c3cadf8","added_by":"auto","created_at":"2024-03-20 22:47:56","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":312562,"visible":true,"origin":"","legend":"\u003cp\u003ePrototype system with the powering components\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-3834903/v1/992b0c908287120b12f5fbbc.png"},{"id":53123804,"identity":"a275d7a5-9fd2-4803-8cf7-0cadac33d430","added_by":"auto","created_at":"2024-03-20 22:47:56","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":181688,"visible":true,"origin":"","legend":"\u003cp\u003eIoT-based data collection process\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-3834903/v1/bfe0727adba5e0405e544538.png"},{"id":53123805,"identity":"8578290f-2ea8-4207-824b-2f8945e01b69","added_by":"auto","created_at":"2024-03-20 22:47:57","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":571693,"visible":true,"origin":"","legend":"\u003cp\u003eDeployed sensor node\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-3834903/v1/34909dbb996a801b35eb91ce.png"},{"id":53123808,"identity":"d8dca441-6af5-4872-87b9-0e4763f75d2a","added_by":"auto","created_at":"2024-03-20 22:47:57","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":131670,"visible":true,"origin":"","legend":"\u003cp\u003eReal time data visualization\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-3834903/v1/9568eb620950d0ba87504f89.png"},{"id":53123809,"identity":"246e0492-2806-4354-9b8e-a164d221c050","added_by":"auto","created_at":"2024-03-20 22:47:57","extension":"png","order_by":7,"title":"Figure 7","display":"","copyAsset":false,"role":"figure","size":243264,"visible":true,"origin":"","legend":"\u003cp\u003eWeather data captured by one of the deployed sensor nodes\u003c/p\u003e","description":"","filename":"7.png","url":"https://assets-eu.researchsquare.com/files/rs-3834903/v1/1a0bab954db27b7643961751.png"},{"id":53123811,"identity":"b4598ef2-701e-4e3c-b410-538fa66c563a","added_by":"auto","created_at":"2024-03-20 22:47:57","extension":"png","order_by":8,"title":"Figure 8","display":"","copyAsset":false,"role":"figure","size":165756,"visible":true,"origin":"","legend":"\u003cp\u003eBattery life data from the voltage sensor\u003c/p\u003e","description":"","filename":"8.png","url":"https://assets-eu.researchsquare.com/files/rs-3834903/v1/0f81dce309cbca155a629bd9.png"},{"id":53123939,"identity":"5c3799bd-2ed5-415e-a780-2bf8f29a539b","added_by":"auto","created_at":"2024-03-20 22:55:57","extension":"png","order_by":9,"title":"Figure 9","display":"","copyAsset":false,"role":"figure","size":53032,"visible":true,"origin":"","legend":"\u003cp\u003eCorrelation among weather parameters\u003c/p\u003e","description":"","filename":"9.png","url":"https://assets-eu.researchsquare.com/files/rs-3834903/v1/feb558c9c230eddae222c1d7.png"},{"id":53123806,"identity":"7786b973-106a-4667-9da3-91048036e1d4","added_by":"auto","created_at":"2024-03-20 22:47:57","extension":"png","order_by":10,"title":"Figure 10","display":"","copyAsset":false,"role":"figure","size":107989,"visible":true,"origin":"","legend":"\u003cp\u003eCorrelation among weather parameters\u003c/p\u003e","description":"","filename":"10.png","url":"https://assets-eu.researchsquare.com/files/rs-3834903/v1/f7166975b6b253ab7ee5d0c3.png"},{"id":53123810,"identity":"8ff77667-1f63-49a1-9caf-1d81628d95d1","added_by":"auto","created_at":"2024-03-20 22:47:57","extension":"png","order_by":11,"title":"Figure 11","display":"","copyAsset":false,"role":"figure","size":39680,"visible":true,"origin":"","legend":"\u003cp\u003eInconsistences between rainfall data from rain gauge and rain sensor\u003c/p\u003e","description":"","filename":"11.png","url":"https://assets-eu.researchsquare.com/files/rs-3834903/v1/d81e649d88c696db429c3c52.png"},{"id":68750154,"identity":"bc9e5c48-2bfb-493e-a619-9b90319f36ab","added_by":"auto","created_at":"2024-11-11 16:11:14","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2709480,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3834903/v1/4f199b95-7348-4afe-9f88-51fabaed562e.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"SMART-CYPS: An Intelligent Internet of Things and Machine Learning Powered Crop Yield Prediction System for Food Security","fulltext":[{"header":"1. Introduction","content":"\u003cp\u003eAccording to the World Bank, the number of people facing food insecurity increased from 135\u0026nbsp;million to 345\u0026nbsp;million from 2019 to June 2022, due to the Ukraine war and the economic fallout from the COVID-19 pandemic that caused food prices to rise extremely [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e]. Food insecurity is a result of several causes, including weather-related hazards, a lack of input resources, crop diseases, regional or international wars, and threats from both intentional and inadvertent activities [\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e, \u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e]. Weather-related threats include, high temperatures, water shortages, floods, droughts, and higher atmospheric CO2 concentrations. For instance, Corn and wheat production have gone down recently because of bad weather, plant diseases, and a lack of water around the world [\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e]. Declining crop yields, particularly in the areas with the greatest food insecurity on the globe, will cause more people to live in poverty if there are no remedies. With climate change, more extreme weather occurrences are anticipated, which could paralyze agricultural productivity [\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e]. It is projected that by 2030, 43\u0026nbsp;million people in Africa alone will be living in poverty [\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eOn a global scale, investments in technology development, adaption, and transfer might significantly lessen these effects [\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e]. To cope with the issue of food insecurity, the government of Rwanda has established different strategies including the Crop Intensification Programme (CIP), Livestock Intensification Program (LIP), Land Use Consolidation Program (LUCP), and others [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. The CIP and LUCP among others were prioritized programs focusing on specialized crops with a target of increasing the agricultural production and food security in Rwanda [\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e, \u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e]. However, because most of the farming activities are predominately rain-fed, Rwandan agriculture is extremely vulnerable to climate change and extreme weather events like droughts, floods, and severe storms [\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e]. To counter the impact of the extreme weather events on food production, the Rwanda government has put in place different measures such as the promotion of drought-relieving irrigation methods. Under this initiative, local communities receive training in low-cost irrigation techniques such as rainwater collection, water storage, water pumping from water bodies, drought-resistant seeds, and food storage, where communities preserve extra harvest that can be used during drought periods [\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e]. However, rainfall deficits and rainfall induced disasters strike different regions in Rwanda leading to food insecurity being experienced in multiple regions of the country. For instance, eastern province is vulnerable to droughts due to irregular rainfall and long dry season that result in the shortage of food [\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e, \u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e]. On the other hand, the hilly and mountainous regions of Rwanda (namely north and west) are characterized by plenty of rain especially during the March \u0026ndash; May rain season. As a result, floods and landslides are frequent, leading to poor crop harvest [\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eEven though it would be difficult to fully control climate conditions, technology use can reduce the magnitude of associated risks including food crisis. Historical weather and crop yields data can be used to predict the future production levels of different crops. This knowledge can be useful to policy makers enabling them to set some strategies aiming at reducing the risks of food insecurity. At present, fully documented studies done on the prediction of crop production using combined weather and crop yield data are scarce in Rwanda. Also, in some cases where irrigation is done, there is no control of how much water content the soil needs according to the growth stage of the crops. Therefore, there is need to use IoT systems to monitor the water content in the soil so that irrigation is done when needed. The use of IoT would promote proper use of water but also minimizes risks of water lodging due to excessive irrigation.\u003c/p\u003e \u003cp\u003eIn this work, a model integrating IoT and ML is proposed that is informed by policy maker\u0026rsquo;s perspective but also integrates farmers\u0026rsquo; context thereby setting to contribute to a more efficient strategy for technology testing. The model uses historical yield data and current weather data to predict crop yield. The IoT system captures real time weather data which is fed into the Machine learning model for prediction purposes. A prototype of the system is developed and deployed on the farm for real-time data capturing and crop yield prediction using a robust prediction model. This system responds directly to the need for data driven approaches to policy making in the agricultural sector to ensure food security, especially in low-to-middle income countries. By capturing real-time weather parameters, the system has potential for use beyond crop prediction as it can also be used to optimize irrigation and water use in Rwanda. Furthermore, by using weather and crop yield data, the model makes it possible to analyze the relationship between crop production and climate change with time. Specifically, the following are the contributions of this work:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003ePresents the designing and development of a crop yield prediction system called SMART-CYPS that integrates IoT and machine learning.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eUsing collected real-time weather data and historical crop yield data, we analyze Irish-Potato and Maize seasonal yields for Musanze District in Rwanda.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e"},{"header":"2. Related works","content":"\u003cp\u003eEmerging technologies continue to play a central role in solving different problems, including those related to the basic needs of people, such as food security. Various studies used IoT and artificial intelligence (AI) to provide solutions to the problems related to insufficient crop yield. The optimization of crop production may start with matching of crops to specific weather condition. For instance, the study conducted by Amna I. et al. [\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e] used IoT and ML for smart selection of crops to be planted in a specific farm by monitoring different parameters such as rainfall temperature, humidity, nitrogen, potassium, pH, and carbon dioxide (CO2). Since each crop has different nutrient requirements, the designed system gathers data from the soil, and with ML algorithms, the appropriate type of crop is recommended. Similar studies have been conducted using the same parameters in [\u003cspan additionalcitationids=\"CR16\" citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe quantity and quality of crop production can also be enhanced by predicting crop diseases using IoT and AI. In their study, Ravesa A. et al. [\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e] proposed a prediction model for apple orchard diseases using data analytics, and ML in an IoT system. They fed a linear regression model using real-time data collected by wireless sensor nodes and IoT nodes. The design consists of several IoT sensor nodes placed across the apple orchard and a nearby gateway for gathering data from the nodes' mesh. A network was created between the nodes, and a nearby gateway. The fog node collects data from a certain number of nodes in the network. For each location, data analysis is performed in real-time to enable prompt responses as necessary. The application might provide farmers with real-time updates on the health of their apple orchards, allowing them to take immediate action against any potential illness like scab if it is accurately and properly forecasted.\u003c/p\u003e \u003cp\u003eOn the other hand, optimizing crop yield can be done through monitoring various influencing parameters, such as those related to weather and soil properties. The system designed by [\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e] monitored daily rainfall and temperature for the prediction of the crop harvest. The architecture of the system consists of a peasant farming layer where the farmer uses a mobile application to input the seasonal crop yield into the system for storage. The same application is used by the farmer to visualize the predicted crop yield as well as weather data that are remotely collected by IoT sensors. The second layer of this system is a wireless sensor network that consists of ultra-low power transceiver modules that are used for communication among the sensor nodes. IoT-based hardware is the third layer of the system. It was made of a general-purpose microcontroller connected to the transceiver module, which is the Global System for Mobile (GSM). The last layer was the cloud computing platform, which comprises four cloud computing services: an IoT message hub for enabling two-way communication between system components; data storage service for storing data from sensors; data analytics service for data mining; and a machine learning service for the implementation of predictive analytics for crop production.\u003c/p\u003e \u003cp\u003eAn IoT ecosystem for smart agriculture framework proposed by Vu K et al. [\u003cspan citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e] consists of four main components. The first component which consists of IoT devices for monitoring the farming environment, comprises sensors to gather information such as weather-related data and actuators to enable physical action as a result of decisions based on the information coming from sensors. The second component of this framework is the communication technology, which consists of infrastructure and communication protocols that are responsible for data transfer from the sensor node to the cloud. The third component of this framework is the cloud, which comprises database and server applications responsible for data processing and storage, big data analytics, and decision making. The fourth section is the end user (farmer or system owner) who interacts with the system using an application.\u003c/p\u003e \u003cp\u003eOther studies looked at how to anticipate agricultural yields using remote sensing. Those that have been in practice for a while use satellite and drone images for crop harvesting decisions [\u003cspan additionalcitationids=\"CR21 CR22 CR23\" citationid=\"CR20\" class=\"CitationRef\"\u003e20\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e]. The monitoring of crops on a global and regional scale is currently greatly aided by better new technologies [\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e]. As highlighted by the above cited works, IoT and ML are key technologies in the realization of food security. IoT supports real time monitoring of crops and weather parameters, whilst ML algorithms can use both sensed and historical data to predict crop yields. Knowledge of anticipated harvest empowers the farmers to plan ahead, e.g., by searching for markets. Famine early warning is another application of crop production prediction, which allows decision-makers like the government to establish a strategic plan to address the impending food insecurity.\u003c/p\u003e \u003cp\u003eThe above highlighted studies have made attempt to use IoT and ML technologies for addressing various aspects of crops and plants growth. However, unlike previous studies, this work focuses on crop yield prediction using current weather data and historical crop yield data. The IoT sensors are responsible for capturing weather data and feeding it into the ML model which combines it with historical meteorological and crop yield data for yield prediction purposes. Considering the impact of climate change on Rwanda\u0026rsquo;s agricultural sector, we believe the develop system will support various initiatives being undertaken by the government to ensure food security in the country.\u003c/p\u003e"},{"header":"3. SMART-CYPS – System Design and Development","content":"\u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003e3.1 Methodology\u003c/h2\u003e \u003cp\u003eData collection and modeling\u003c/p\u003e \u003cp\u003eThis study builds on the findings reported in [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e]. In [\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e], Kuradusenge, et al. conducted a study that analysed the impact of climate conditions on crop production. As part of the study, historical data about meteorology and crop yields (i.e., for Irish Potato and maize) were collected and using this data modelling was performed to determine the best ML model to be implemented for crop yield prediction. Furthermore, the study identified the optimum values of weather parameters for optimum yields. Therefore, this study uses the values from that study to develop a crop yield prediction system that integrates IoT and ML.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003e3.2 System archtecture\u003c/h2\u003e \u003cp\u003eThe architecture of the developed crop yield prediction system is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e. The data used by the system for prediction comprises secondary data and primary data. The secondary data includes, crop yield data and climatic data, i.e., rainfall, temperature and humidity but also crop yield data. The secondary data can be accessed from various databases whilst the primary data is continuously being collected by the IoT sensors and being fed into the system. As can be seen from the Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e, the sensors collect data for two purposes, namely; for prediction and storage. All real-time data is fed into the deployed model for harvest prediction. Informed by the findings in [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e], the ML model trained for prediction is the Random Forest regressor. The model is configured in a such way that, it predicts the harvest basing on the crop growing stage or monthly data batches and not basing on single data prediction.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec6\" class=\"Section2\"\u003e \u003ch2\u003e3.3 SMART-CYPS \u0026ndash; IoT sub-system\u003c/h2\u003e \u003cp\u003eSystem design and functionality\u003c/p\u003e \u003cp\u003eThe IoT sub-system of the SMART-CYPS system comprises various sensors that collect data related to climatic conditions including, humidity, temperature, rainfall, and soil moisture. The system further has sensors that monitors system Current which help in determining the battery status. Other components include, the GSM module used for communication, and GPS module used for locating the node. Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e provides the details of all the electronic components of the IoT sub-system.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eDescription of components used for the development of the CYPS\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003eComponent\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDescription\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eFunction\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eOperating Voltage\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eATMEGA328P-PU\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eMicrocontroller IC\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003ePerforms data processing function\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e1.8\u0026ndash;5.5 V\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eDHT22\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDigital Temperature and Humidity Sensor module\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMeasuring the temperature \u0026amp; humidity of the surrounding air\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5V\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eFST100-2007 RS485 Modbus 24V DC Rain and Snow Sensor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eDetects rainwater\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMeasures whether there is rain or snowfall outdoors\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e10-30V-DC\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSoil moisture sensor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eSoil moisture VWC water content test sensor probe with 3 pins\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMeasuring the soil water content\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5-30V-DC\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eVoltage Sensor\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eVoltage detection module\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eMonitoring the status voltage saved by the battery powered by the solar panel\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e5V\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eGY-NEO6MV2\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGPS Module\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eDetecting the location of the sensor node\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e3-5V\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003eSIM800L\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003eGSM/GPRS module\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003eConnecting the device to the GSM network\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\"\u003e \u003cp\u003e4.2V\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eAll components described in the table above are mounted on the printed circuit board (PCB) allowing the data collection and processing. Figure\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e shows the circuit design. The powering system is made of three main components: solar panel (20 W), Battery, and charge controller. Figure\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e shows the system prototype including the powering components.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eSystem communication\u003c/p\u003e \u003cp\u003eFigure \u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e shows the communication architecture of the SMART-CYPS IoT subsystem. As pointed out above, the IoT data are collected using sensors (sensor nodes labelled as Edge Layer in this context) that transmits the data to the cloud through GPRS protocol within the telecommunication cellular network.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eTo ensure security of the data, the sensors transmit the data to the cloud with associated tokens. To manage the data ingestion from heterogenous devices, the data is sent and received in JavaScript Object Notation (JSON) format, the sample is shown in the diagram. The past production information and weather conditions are also stored in the server where data mining techniques will be used to produce a report showing the future production predicted by the system. To ease the computation by avoiding model testing overhead, the cloud services were developed under container paradigms as microservices. This scheme proved to be efficient while in Docker environment. An authenticated farmer/policymaker can be able to interact with the system to see those reports and predictions. The developed web solution can predict the crop yield depending on the current data from the sensor nodes and the historical harvest. This prediction is due to the integrated model in the cloud services.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec7\" class=\"Section2\"\u003e \u003ch2\u003e3.4 SMART-CYPS \u0026ndash; Machine learning model performance evaluation\u003c/h2\u003e \u003cp\u003eThis work implements the Random Forest regressor based on its best performance in crop yield forecasting as reported in [\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e]. We adopt the Mean Absolute Percentage Error (MAPE) as the performance evaluation metric, which is calculated using the following formula:\u003cdiv class=\"BlockQuote\"\u003e\u003cp\u003e.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Equ1\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ1\" name=\"EquationSource\"\u003e\n$$MAPE=\\frac{1}{n}{\\sum }_{i=1}^{n}\\left|\\frac{{A}_{i}-{P}_{i}}{{A}_{i}}\\right|100$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e1\u003c/div\u003e\u003c/div\u003e,\u003c/p\u003e \u003cp\u003ewhere MAPE is the Mean Absolute Percentage Error, n is the total number of data points (or observations), Ai represents the actual or observed values, and Pi represent the predicted values for the corresponding observations.\u003c/p\u003e \u003c/div\u003e"},{"header":"4. Results","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e \u003ch2\u003e4.1 Real time data gathering and visualization\u003c/h2\u003e \u003cp\u003eThe designed and developed system was deployed in the study area after six months of testing. The IoT system daily collects weather and soil humidity data from the farms and sends it to the cloud for long-term storage and future analysis. Figure\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003e is one of an IoT node deployed on the farm.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe system dashboard displays the real time data coming from three of the sensor nodes. In the back end the rest of the data are stored on daily basis. The frequency of data transmission from sensor node is 15 minutes. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e, in real-time, the data can be visualized on the system dashboard. The figure also shows the data summary and at the end of the day the minimum, maximum and average data are saved in the database. At the end of the season, the rest of the data together with actual crop yield are added to the historical data. As shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003e, the system has different modules. Below is a description of each of the module:\u003c/p\u003e \u003cp\u003e \u003cul\u003e \u003cli\u003e \u003cp\u003eManaging users: used to manage users such as adding, removing, and assigning roles.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eCooperative module: used to visualize or add the name of farmers cooperatives that will have access to the information in the system.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eRoles: used to add the visualize or add the managing roles.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eDevices: module is used to track the device: it provides the GPS coordinates and location on the map. This module also provides the battery (voltage) level.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eCrop category: provides the option to select or add crop types for yield prediction.\u003c/p\u003e \u003c/li\u003e \u003cli\u003e \u003cp\u003eData visualization: comprises historical data for weather (rainfall, temperature, and humidity), and harvest.\u003c/p\u003e \u003c/li\u003e \u003c/ul\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eSince the system\u0026rsquo;s deployment, sensors have continuously captured and recorded the data in the database. Figure\u0026nbsp;\u003cspan refid=\"Fig7\" class=\"InternalRef\"\u003e7\u003c/span\u003e shows the variation of data over time since the deployment of the system prototype.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eIt is further possible to track location of the deployed devices and monitor battery life. As illustrated in Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, the system relies on a solar panel system for power, a necessity in remote areas without access to the main power grid. The web application is used to keep tabs on the status of the battery power. Figure\u0026nbsp;\u003cspan refid=\"Fig8\" class=\"InternalRef\"\u003e8\u003c/span\u003e portrays the voltage variations for the period of one year and half. The figure illustrates a decrease in voltage during the rainy seasons in Rwanda, spanning from March to May and September to December. It was common to observe interruptions in data transmission during these periods.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec10\" class=\"Section2\"\u003e \u003ch2\u003e4.2 Correlation of weather parameters\u003c/h2\u003e \u003cp\u003eTo identify the correlation among the weather variables, we utilized the data collected the system throughout the study's duration. The correlation matrix depicted in Fig.\u0026nbsp;\u003cspan refid=\"Fig9\" class=\"InternalRef\"\u003e9\u003c/span\u003e, reveals several significant correlations. Specifically, there is a moderate positive correlation of 0.51 between rainfall and soil moisture, a moderate positive correlation of 0.41 between rainfall and humidity, and a small negative correlation of \u0026minus;\u0026thinsp;0.25 between rainfall and maximum temperature. Furthermore, maximum temperature exhibits a moderate negative correlation of \u0026minus;\u0026thinsp;0.36 with soil moisture, and a moderate negative correlation of \u0026minus;\u0026thinsp;0.58 between humidity and maximum temperature. Additionally, a positive correlation of 0.48 was observed between soil moisture and humidity.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003e4.3 Prediction results and the actual yield for Irish potatoes and maize\u003c/h2\u003e \u003cp\u003eCurrently, SMART-CYPS monitors the cultivation of two crops: Irish potatoes and maize, which are rotated between seasons A and B in the study area. Typically, maize is planted during season A, and Irish potatoes are cultivated in season B. Since the system's deployment, we have monitored one season of maize and two seasons of Irish potatoes. In this area, maize has a growth period of six months, while Irish potatoes are harvested after four months. The subsequent tables (Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e, Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e, and Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e) present the predicted and actual production for each crop during three agricultural seasons. The system forecasts crop yields at various growth stages.\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eComparison of the yield predicted by CYPS and the actual yield for Irish potatoes (Season B 2022)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrediction time (growing stage)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePredicted yield (kg ha-1)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eActual yield (kg ha-1)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAccuracy (MAPE)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e21 days (sprout development)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e12,109\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e8,717\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c4\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e0.339\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2nd month (vegetative \u0026amp; tuber initiation)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e11,373\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3rd month (tuber bulking/filling)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e11,639\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4th month (maturity)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"char\" char=\".\" colname=\"c2\"\u003e \u003cp\u003e11,594\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eThe growing stages considered for Irish potatoes are sprout development, vegetative, tuber initiation and bulking, and maturity [\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e]. The initial prediction was conducted at the end of the sprout development phase of Irish potatoes, while the last occurred during maturity. As shown in the Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e. Comparison of the yield predicted by CYPS and the actual yield for Irish potatoes (Season B 2022), the actual yield for Irish potatoes in the 2022 season B was 8,717 kg ha -1, whereas the predicted yield ranged from 11,373 kg ha -1 to 12,109 kg ha -1. The prediction error (MAPE) for this season was 0.339. In the agricultural season B of 2023, the crop yield was 9,837 kg ha -1, and the predicted yield varied from 11,985 kg ha -1 to 13,529 kg ha -1, with a prediction error of 0. 309..\u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eComparison of the yield predicted by CYPS and the actual yield for Irish potatoes (Season B 2023)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"5\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrediction time (growing stage)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colspan=\"2\" nameend=\"c3\" namest=\"c2\"\u003e \u003cp\u003ePredicted yield (kg ha-1)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eActual yield (kg ha-1)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c5\"\u003e \u003cp\u003eAccuracy (MAPE)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003e21 days ADP (Sprout development)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e13,529\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\" morerows=\"3\" rowspan=\"4\"\u003e \u003cp\u003e9,837\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003e2nd month (vegetative \u0026amp; tuber initiation)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e12,898\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e \u003cp\u003e0.309\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003e3rd month (tuber bulking/filling)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e11,985\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colspan=\"2\" nameend=\"c2\" namest=\"c1\"\u003e \u003cp\u003e4th month (maturity)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\"\u003e \u003cp\u003e13,130\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c5\"\u003e\u0026nbsp;\u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003e \u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab4\" border=\"1\"\u003e \u003ccaption language=\"En\"\u003e \u003cdiv class=\"CaptionNumber\"\u003eTable 4\u003c/div\u003e \u003cdiv class=\"CaptionContent\"\u003e \u003cp\u003eComparison of the yield predicted by CYPS and the actual yield for maize (Season A 2023)\u003c/p\u003e \u003c/div\u003e \u003c/caption\u003e \u003ccolgroup cols=\"4\"\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e \u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e \u003cthead\u003e \u003ctr\u003e \u003cth align=\"left\" colname=\"c1\"\u003e \u003cp\u003ePrediction time (growing stage)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c2\"\u003e \u003cp\u003ePredicted yield (kg ha-1)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c3\"\u003e \u003cp\u003eActual yield (kg ha-1)\u003c/p\u003e \u003c/th\u003e \u003cth align=\"left\" colname=\"c4\"\u003e \u003cp\u003eAccuracy (MAPE)\u003c/p\u003e \u003c/th\u003e \u003c/tr\u003e \u003c/thead\u003e \u003ctbody\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e1st month (establishment \u0026amp; vegetative)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1,758\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c3\" morerows=\"4\" rowspan=\"5\"\u003e \u003cp\u003e1,837\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c4\" morerows=\"4\" rowspan=\"5\"\u003e \u003cp\u003e0.177\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e2nd month (vegetative)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1,222\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e3rd month (tasseling)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1,580\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e4th month (cob setting)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1,345\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003ctr\u003e \u003ctd align=\"left\" colname=\"c1\"\u003e \u003cp\u003e5-6th month (grain-filling \u0026amp; maturity)\u003c/p\u003e \u003c/td\u003e \u003ctd align=\"left\" colname=\"c2\"\u003e \u003cp\u003e1,651\u003c/p\u003e \u003c/td\u003e \u003c/tr\u003e \u003c/tbody\u003e \u003c/colgroup\u003e \u003c/table\u003e\u003c/div\u003e \u003c/p\u003e \u003cp\u003eSix growing stages were considered to predict maize production. Those are establishment, vegetative, tasseling, cob setting, grain-filling, and maturity [\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e].\u003c/p\u003e \u003cp\u003eThe results in Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e are for agriculture season A of 2023 that started in September 2022. The predicted maize yield varied between 1,222 kg ha-1 and 1,758 kg ha-1 whereas the actual maize yield was 1,837 kg ha-1. The performance of the prediction was 0.177 MAPE.\u003c/p\u003e \u003c/div\u003e"},{"header":"5. Discussion","content":"\u003cp\u003eCrop yield is a consequence of a range of parameters, which can be broadly classified into two categories: factors within human control and those influenced by natural forces. Human-controllable factors encompass various elements, including the choice of crops or seed variety tailored to the specific region, planting schedules, soil preparation, the application of fertilizers, irrigation methods, weed management, and the use of pesticides, among others. In the study area, these variables fall under the supervision of local agronomists who collaborate closely with farmers, offering guidance and support as needed. Therefore, the natural factors affecting crop yields are the subject of monitoring in this study. Weather data is collected for two primary purposes: predicting crop yields and creating a historical record for future reference. Predictions are conducted monthly, aligning with the various growth stages of the crops, as indicated in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e through Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e. The prediction process commences following planting and is initiated by the responsible individual, such as the agronomist or the leader of the farmers' cooperative.\u003c/p\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003e5.1 Rainfall and soil moisture\u003c/h2\u003e \u003cp\u003eThe SMART-CYPS incorporates a rain sensor designed for rain detection. This sensor furnishes binary values (0 or 1) to signify the presence or absence of rain. The preference for a rain sensor over a rain gauge was driven by the unavailability of a rain gauge equipped with network capabilities for real-time rainfall data acquisition. To overcome this limitation, we adopted a strategy that involved utilizing rain and soil moisture sensors. This approach enables us to observe shifts in soil moisture levels during rainfall events and supplement this data with secondary rainfall information obtained from the meteorological office. During consecutive rainy days, the soil moisture increases even when the amount of rainfall is not substantial. This indicates that soil moisture depends not only on the daily rainfall but also on antecedent rainfall. The correlation between rainfall and soil moisture is shown in Fig.\u0026nbsp;\u003cspan refid=\"Fig10\" class=\"InternalRef\"\u003e10\u003c/span\u003e\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec14\" class=\"Section2\"\u003e \u003ch2\u003e5.2 Rainfall data from rain gauge vs rain detected by sensor node\u003c/h2\u003e \u003cp\u003eUpon comparing the rainfall data obtained from the rain gauge with that from the rain sensor, we discovered discrepancies between the two sources. For example, there were some instances where the system signaled rain, but the recorded rainfall data was zero (0). Contrariwise, there were dates when the rainfall data exceeded zero, yet the system did not detect any rain. These inconsistencies are depicted in Fig.\u0026nbsp;\u003cspan refid=\"Fig11\" class=\"InternalRef\"\u003e11\u003c/span\u003e.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003eThe inconsistency is due to the locations of the device versus the location of the rain gauge. That means there are some cases where rain does not reach the entire area and this shows the need of increasing the distribution of rain gauges for more accurate information about rainfall in the study region.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec15\" class=\"Section2\"\u003e \u003ch2\u003e5.3 Prediction performance\u003c/h2\u003e \u003cp\u003eDuring every growth stage, the SMART-CYPS computes yield forecasts using cumulative rainfall, average minimum temperature, and average maximum temperature from the start of that stage. It is important to note that humidity and soil moisture were excluded due to insufficient historical data. These parameters will be incorporated into the system once enough data is recorded. Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e to Table\u0026nbsp;\u003cspan refid=\"Tab4\" class=\"InternalRef\"\u003e4\u003c/span\u003e demonstrates that predicted data varied across growth stages in response to weather conditions, while actual data remained consistent since it is recorded at the conclusion of the agricultural season. The results in Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e indicate an accurate prediction of Irish potato yields with a precision of 0.339, which is a commendable outcome. A similar level of model performance was maintained in the subsequent seasons, with a MAPE of 0.309 for predicting the yield of the same crop.\u003c/p\u003e \u003cp\u003eThe maize yield predictions for the 2023 agricultural season A demonstrated a significantly improved performance, as indicated by a MAPE of 0.177. Predictions were conducted monthly, a total of five times, utilizing the same parameters as those applied to Irish potatoes. In our study, maize is typically harvested six months after planting. During the initial five months, we conducted predictions because during the sixth month, the maize crop is no longer developing, as it is intended for grain drying purposes. The higher frequency of maize yield predictions might be attributed to the enhanced performance compared to Irish potatoes, which could also be attributed to the data quality used in the prediction process.\u003c/p\u003e \u003cp\u003eThe disparities between the forecasted and real crop yields may stem from additional factors, such as fertilizer usage, planting timing, and more. Consequently, enhancing the prediction capability may be achieved by incorporating these variables in future research.\u003c/p\u003e \u003c/div\u003e"},{"header":"6. Conclusions","content":"\u003cp\u003ePredicting crop yields can significantly contribute to reducing food insecurity since such information empowers farmers and other agricultural stakeholders to take appropriate steps well before harvest. In this work, two sources of data have been used to predict crop yields, i.e., real time information provided by the IoT sensor nodes and historical meteorological and crop yields data. The developed SMART-CYPS system has been deployed in Maize and Irish potato farms, however, the system can be adapted for other crop types subject to availability of historical data on climate and crop yields, along with comprehensive information on the crop's weather requirements. In future endeavors, the SMART-CYPS will be further improved by incorporating real-time rain gauge data and expanding its parameters to include soil properties, fertilizer usage, and more.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eFunding:\u003c/strong\u003e This research was funded through the grant offered by the University of Rwanda in the partnership with Sida (Swedish International Development Agency) under the programme UR-Sweden program. The grant supported all research activities such as data collection, purchase of equipment and materials, fieldwork, etc. The APC was also funded by the same grant.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts of Interest:\u003c/strong\u003e The authors declare no conflict of interest.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData Availability Statement:\u003c/strong\u003e All data used in this study are accessible online via the portal of the African Centre of Excellence in IoT (ACEIoT) at the University of Rwanda. The data link can be found in the reference section\u0026nbsp;[28].\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCode availability:\u003c/strong\u003e Not applicable\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor Contributions:\u003c/strong\u003e Conceptualization, M.K.; methodology, M.K. and E.H.; software, E.H. and J.N; validation, K.M.; formal analysis, E.H.; resources, J.H., A; data curation, J.N.; writing original draft preparation, M.K.; writing review and editing, K.M., J.H, P.R, and A.G; visualization, A.G.; project administration, M.K.; funding acquisition, M.K. All authors have read and agreed to the published version of the manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eWorldbank, Climate Explainer: Food Security and Climate Change, Worldbank.Org. (2022). https://www.worldbank.org/en/news/feature/2022/10/17/what-you-need-to-know-about-food-security-and-climate-change (accessed February 28, 2023).\u003c/li\u003e\n \u003cli\u003eA. Djurle, B. Young, A. Berlin, I. V\u0026aring;gsholm, A.L. Blomstr\u0026ouml;m, J. Nygren, A. Kvarnheden, Addressing biohazards to food security in primary production, Food Secur. 14 (2022) 1475\u0026ndash;1497. https://doi.org/10.1007/s12571-022-01296-7.\u003c/li\u003e\n \u003cli\u003eV. Gitz, A. Meybeck, L. Lipper, C. Young, S. Braatz, Climate change and food security: Risks and responses, 2016. https://doi.org/10.1080/14767058.2017.1347921.\u003c/li\u003e\n \u003cli\u003eWorldwide Concern, How climate change increases hunger \u0026mdash; and why we\u0026rsquo;re all at risk | Concern Worldwide, (2022) 1\u0026ndash;7. https://www.concern.net/news/climate-change-and-hunger (accessed February 28, 2023).\u003c/li\u003e\n \u003cli\u003eT.J. Troy, C. Kipgen, I. Pal, The impact of climate extremes and irrigation on US crop yields, Environ. Res. Lett. 10 (2015). https://doi.org/10.1088/1748-9326/10/5/054013.\u003c/li\u003e\n \u003cli\u003eD. Tilman, C. Balzer, J. Hill, B.L. Befort, Global food demand and the sustainable intensification of agriculture, Proc. Natl. Acad. Sci. U. S. A. 108 (2011) 20260\u0026ndash;20264. https://doi.org/10.1073/pnas.1116437108.\u003c/li\u003e\n \u003cli\u003eN. Cantore, The crop intensification program in Rwanda: a sustainability analysis, Environ. Progr. United Nations Dev. Progr. United Nations. 44 (2011) 1\u0026ndash;35. https://www.odi.org/sites/odi.org.uk/files/odi-assets/publications-opinion-files/7094.pdf.\u003c/li\u003e\n \u003cli\u003eK. Kariyasa, Y.A. Dewi, Land consolidation, specialization and household diets: Evidence from Rwanda, J. Gender, Agric. Food Secur. 1 (2011) 1\u0026ndash;22.\u003c/li\u003e\n \u003cli\u003eA. Nsabimana, F. Niyitanga, D.D. Weatherspoon, A. Naseem, Land Policy and Food Prices: Evidence from a Land Consolidation Program in Rwanda, 19 (2021) 63\u0026ndash;73. https://doi.org/doi:10.1515/jafio-2021-0010.\u003c/li\u003e\n \u003cli\u003eL. Muneza, Droughts and Floodings Implications in Agriculture Sector in Rwanda: Consequences of Global Warming, Nature, Causes, Eff. Mitig. Clim. Chang. Environ. (2021) 13. http://dx.doi.org/10.1039/C7RA00172J%0Ahttps://www.intechopen.com/books/advanced-biometric-technologies/liveness-detection-in-biometrics%0Ahttp://dx.doi.org/10.1016/j.colsurfa.2011.12.014.\u003c/li\u003e\n \u003cli\u003eM. Lydie, Droughts and Floodings Implications in Agriculture Sector in Rwanda: Consequences of Global Warming, in: S.A. Harris (Ed.), IntechOpen, Rijeka, 2022: p. Ch. 18. https://doi.org/10.5772/intechopen.98922.\u003c/li\u003e\n \u003cli\u003eL. Nahayo, G. Habiyaremye, A. Kayiranga, E. Kalisa, C. Mupenzi, D.F. Nsanzimana, Rainfall Variability and Its Impact on Rain-Fed Crop Production in Rwanda, Am. J. Soc. Sci. Res. 4 (2018) 9\u0026ndash;15.\u003c/li\u003e\n \u003cli\u003eA.J. Van Bodegom, B. Satijn, Climate Change Profile, (2015).\u003c/li\u003e\n \u003cli\u003eA. Ikram, W. Aslam, R.H.H. Aziz, F. Noor, G.A. Mallah, S. Ikram, M.S. Ahmad, A.M. Abdullah, I. Ullah, Crop Yield Maximization Using an IoT-Based Smart Decision, J. Sensors. 2022 (2022). https://doi.org/10.1155/2022/2022923.\u003c/li\u003e\n \u003cli\u003eP.K. Priya, N. Yuvaraj, An IoT based gradient descent approach for precision crop suggestion using MLP, J. Phys. Conf. Ser. 1362 (2019). https://doi.org/10.1088/1742-6596/1362/1/012038.\u003c/li\u003e\n \u003cli\u003eK. Bakthavatchalam, B. Karthik, V. Thiruvengadam, S. Muthal, D. Jose, K. Kotecha, V. Varadarajan, IoT Framework for Measurement and Precision Agriculture: Predicting the Crop Using Machine Learning Algorithms, Technologies. 10 (2022). https://doi.org/10.3390/technologies10010013.\u003c/li\u003e\n \u003cli\u003eS. Bevinakoppa, S.K. Padaganur, V. Nidagundi, IoT Based Smart Prediction System for Crop Suitability, in: 2021 IEEE Int. Conf. Comput., 2021: pp. 174\u0026ndash;179. https://doi.org/10.1109/ICOCO53166.2021.9673499.\u003c/li\u003e\n \u003cli\u003eR. Akhter, S.A. Sofi, Precision agriculture using IoT data analytics and machine learning, J. King Saud Univ. - Comput. Inf. Sci. 34 (2022) 5602\u0026ndash;5618. https://doi.org/10.1016/j.jksuci.2021.05.013.\u003c/li\u003e\n \u003cli\u003eL.O. Colombo-Mendoza, M.A. Paredes-Valverde, M.D.P. Salas-Z\u0026aacute;rate, R. Valencia-Garc\u0026iacute;a, Internet of Things-Driven Data Mining for Smart Crop Production Prediction in the Peasant Farming Domain, Appl. Sci. 12 (2022). https://doi.org/10.3390/app12041940.\u003c/li\u003e\n \u003cli\u003eM. Dhanaraju, P. Chenniappan, K. Ramalingam, Smart Farming : Internet of Things ( IoT ) -Based Sustainable Agriculture, (2022) 1\u0026ndash;26.\u003c/li\u003e\n \u003cli\u003eO.N. Lungu, L.M. Chabala, C. Shepande, Satellite-Based Crop Monitoring and Yield Estimation \u0026mdash; A Review, 13 (2021) 180\u0026ndash;194. https://doi.org/10.5539/jas.v13n1p180.\u003c/li\u003e\n \u003cli\u003eN. Torbick, D. Chowdhury, W. Salas, J. Qi, Monitoring Rice Agriculture across Myanmar Using Time Series Sentinel-1 Assisted by Landsat-8 and PALSAR-2, Remote Sens. 9 (2017). https://doi.org/10.3390/rs9020119.\u003c/li\u003e\n \u003cli\u003eA.M. Ali, M. Abouelghar, A.A. Belal, N. Saleh, M. Yones, A.I. Selim, M.E.S. Amin, A. Elwesemy, D.E. Kucher, S. Maginan, I. Savin, The Egyptian Journal of Remote Sensing and Space Sciences Crop Yield Prediction Using Multi Sensors Remote Sensing ( Review Article ), Egypt. J. Remote Sens. Sp. Sci. 25 (2022) 711\u0026ndash;716. https://doi.org/10.1016/j.ejrs.2022.04.006.\u003c/li\u003e\n \u003cli\u003eT.S. Abdul-Jabbar, A.T. Ziboon, M.M. Albayati, Crop yield estimation using different remote sensing data: literature review, in: IOP Conf. Ser. Earth Environ. Sci., 2023. https://doi.org/10.1088/1755-1315/1129/1/012004.\u003c/li\u003e\n \u003cli\u003eM. Kuradusenge, E. Hitimana, D. Hanyurwimfura, P. Rukundo, K. Mtonga, A. Mukasine, C. Uwitonze, J. Ngabonziza, A. Uwamahoro, Crop Yield Prediction Using Machine Learning Models: Case of Irish Potato and Maize, Agriculture. 13 (2023) 225. https://doi.org/10.3390/agriculture13010225.\u003c/li\u003e\n \u003cli\u003eJ.E. Obidiegwu, Coping with drought: stress and adaptive responses in potato and perspectives for improvement, Front. Plant Sci. 6 (2015). https://doi.org/10.3389/fpls.2015.00542.\u003c/li\u003e\n \u003cli\u003eB.A.A. Zemba, S.Z. Wuyep, A.A. Adebayo, C.J. Jahknwa, Growth and Yield Response of Irish Potato ( Solanum Tuberosum ) to Climate in Jos-South , Plateau State , Nigeria Growth and Yield Response of Irish Potato Solanum Tuberosumto Climate in Jos-South , Plateau State , Nigeria Strictly as per the compliance a, Int. J. Plant Res. 2019 (2013) 1\u0026ndash;7. http://journal.sapub.org/plant.\u003c/li\u003e\n \u003cli\u003eResearch data, (2024). https://aceiot.ur.ac.rw/SMART-CYPS_data.zip.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":false,"highlight":"","institution":"","isAcceptedByJournal":true,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"discover-internet-of-things","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"diot","sideBox":"Learn more about [Discover Internet of Things](https://www.springer.com/journal/43926)","snPcode":"","submissionUrl":"","title":"Discover Internet of Things","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true},"keywords":"IoT; machine learning; prediction, weather, climate change, crop yield, maize, irish potatoes","lastPublishedDoi":"10.21203/rs.3.rs-3834903/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3834903/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe sub-Saharan Africa region continues to experience food insecurity, a consequence of the less productive agricultural sector that has dragged to adapt to the effects of climate change. As the region\u0026rsquo;s population continues to grow, there is a need to modernize the region\u0026rsquo;s agricultural sector to meet the increasing food demand. Although extreme atmospheric conditions cannot be entirely mitigated, however, the integration of technologies such as the Internet of Things (IoT) and Machine Learning (ML) can increase the quantity and quality of production from the crop fields. These technologies have potential to empower agricultural management systems to handle both climatic and farm data in an orchestrated manner, informing formulation of effective strategies. This study presents the design and development of a system for predicting crop yields that integrates IoT and ML. The system combines current weather data and historic crop yield data to predict seasonal crop yields. The weather parameters including, rainfall, temperature, humidity and soil moisture are collected by IoT sensors and transmitted to the cloud for crop yield forecasting. The system is used to analyze seasonal yields of Irish-Potato and Maize in Musanze District of Rwanda. Using data over different agricultural seasons, the system achieved favorable predictive accuracy with mean absolute percentage error (MAPE) values of 0.339, 0.309, and 0.177 for two seasons of Irish potatoes and one season of maize, respectively. Such predictive yield systems can reduce food insecurity risks and enhance harvest efficiency by enabling early awareness of crop production, fostering effective strategies shared among decision-makers and stakeholders. While maize and Irish potatoes were the initial case studies, expansion to include other crops is envisioned.\u003c/p\u003e","manuscriptTitle":"SMART-CYPS: An Intelligent Internet of Things and Machine Learning Powered Crop Yield Prediction System for Food Security","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-03-20 22:47:52","doi":"10.21203/rs.3.rs-3834903/v1","editorialEvents":[{"type":"communityComments","content":0},{"type":"editorInvitedReview","content":"","date":"2024-07-23T10:38:35+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"15290012773169618199191668518541375081","date":"2024-07-15T10:20:34+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-06-10T18:31:12+00:00","index":"hide","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2024-05-27T06:47:26+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"125698586010412588684424175510089429709","date":"2024-05-27T06:30:27+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"202270029715986927911952785097251088491","date":"2024-05-26T07:44:42+00:00","index":"hide","fulltext":""},{"type":"reviewerAgreed","content":"333348148867386589215851791547815217306","date":"2024-05-26T07:11:49+00:00","index":"hide","fulltext":""},{"type":"reviewersInvited","content":"","date":"2024-04-19T10:01:59+00:00","index":"","fulltext":""},{"type":"editorAssigned","content":"","date":"2024-03-21T14:31:33+00:00","index":"","fulltext":""},{"type":"checksComplete","content":"","date":"2024-03-15T06:23:05+00:00","index":"","fulltext":""},{"type":"submitted","content":"Discover Internet of Things","date":"2024-01-04T15:46:31+00:00","index":"","fulltext":""}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"discover-internet-of-things","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":false,"externalIdentity":"diot","sideBox":"Learn more about [Discover Internet of Things](https://www.springer.com/journal/43926)","snPcode":"","submissionUrl":"","title":"Discover Internet of Things","twitterHandle":"","acdcEnabled":true,"dfaEnabled":true,"editorialSystem":"stoa","reportingPortfolio":"Discover Series","inReviewEnabled":true,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"7b03453c-917b-4fb4-937c-d1927b004808","owner":[],"postedDate":"March 20th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"published-in-journal","subjectAreas":[],"tags":[],"updatedAt":"2024-11-11T16:06:24+00:00","versionOfRecord":{"articleIdentity":"rs-3834903","link":"https://doi.org/10.1007/s43926-024-00079-0","journal":{"identity":"discover-internet-of-things","isVorOnly":false,"title":"Discover Internet of Things"},"publishedOn":"2024-11-05 15:58:11","publishedOnDateReadable":"November 5th, 2024"},"versionCreatedAt":"2024-03-20 22:47:52","video":"","vorDoi":"10.1007/s43926-024-00079-0","vorDoiUrl":"https://doi.org/10.1007/s43926-024-00079-0","workflowStages":[]},"version":"v1","identity":"rs-3834903","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3834903","identity":"rs-3834903","version":["v1"]},"buildId":"qtupq5eGEP_6zYnWcrvyt","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}