[Lee Chung-ho Column] Full-Cycle Data Architecture for Smart Agriculture: Standardization Comes First

[Lee Chung-ho Column] Professor Lee Chung-ho, Department of Industrial Engineering, Jeonju University

Japan’s NARO(National Agriculture and Food Research Organization) agricultural data platform WAGRI is operated with the goal of developing new agriculture-related technologies and services by enabling agricultural machinery manufacturers and AI and other ICT technology companies to become data users through the data it provides externally. It has secured public- and private-sector data such as weather, soil, farmland, and growth prediction data, and has built a platform used in various research and business projects through data standardization in API format. For data, it distinguishes between public data and nonpublic data held by farmers, pursuing standardization with security technologies applied. To respond in advance to abnormal weather and pests and diseases, field demonstrations are being carried out through data collection and analysis, and through the WAGRI platform, customized farming is being realized by collecting and processing data on growth and environment, agricultural statistics, and other areas, and then providing it to farmers. 

The United States·and Europe are also competing over data standardization…raising statistical accuracy with satellites·AI

This trend can also be readily found in agriculture in the United States·and Europe, and the U.S. Department of Agriculture(USDA) is working with the National Aeronautics and Space Administration(NASA) and others to promote pilot projects to improve agricultural statistics using satellite imagery in order to raise statistical accuracy. USDA is focusing, through the National Agricultural Statistics Service(NASS), on industrialization through integrated management beyond merely possessing data in order to increase the added value of the data it provides. Systems agriculture, which uses advanced technologies such as satellite imagery, AI, soil and environmental sensors to increase productivity and statistical accuracy, is being introduced under the name agtech. The Common European Agricultural Data Space(CEADS) is building a standardized ecosystem in which farmers, agricultural machinery companies, and public institutions can reliably share production and environmental data, but in every country, the burden of initial investment costs, insufficient digital infrastructure, and inadequate data standardization are being pointed out as real-world problems.

In Korea, the Rural Development Administration has currently established the Agricultural Science and Technology Information Service(ASTIS) system to realize a ‘digital platform government’ but for smart agricultural machinery and full-cycle agricultural data management systems to achieve effective results, a wide range of tasks remain to be resolved.

 The core of data-driven agriculture is an MAS -based layered architecture

Data-based smart agriculture includes intelligent networks and data management, and it requires actionable decision-making functions based on data as well as performance creation that generates meaningful value from the collected data. To this end, it is considered necessary to build an architecture in which multiple independent artificial intelligence agents each take on their own roles, communicate, and cooperate through a multi-agent system(MAS) to solve common goals. In the agricultural industry, farmers are both producers of agricultural products and consumers of agricultural technology. However, the agricultural industry itself has a small-scale industrial structure with no large corporations. In data-driven agriculture, the integration of ICT technology depends on the structural layering of diverse agricultural data through the design of multi-agent systems, the collection of high-quality data for the development of collaborative-agent AI , and the active participation of ICT companies in development using API information, as well as the maturation of technologies and companies; realistically, however, most companies are small, making it difficult to achieve sustained growth and secure highly skilled personnel.

Through the definition and classification of MAS systems for autonomous decision-making in the field of data-driven agriculture, the goal of standardized data collection by companies, farmers, government, research institutes and universities, technology development related to data-driven agriculture through APIs, and the spread of on-site demonstrations should not be limited simply to a technological system, but should lead to the inflow of technology and capital into related industries through AX beyond agricultural productivity, by training university personnel and demonstrating and commercializing systems developed based on the collected information(agricultural machinery, AI -based autonomous decision-making software, etc.). 

Data collection rate among smart agriculture operators: 36.7%…insufficient from the starting point

In reality, although the government is promoting agricultural AX as a national policy task, data acquisition, the starting point of smart agriculture, remains insufficient, and according to the ‘2024 Smart Agriculture Status Survey and Performance Analysis Report’ released by the Korea Agriculture Technology Promotion Agency, among smart agriculture operators(2,564entities) only 36.7% responded that they collect data, showing that realistic awareness of data-driven agriculture is low to the extent that collecting and using high-quality data is difficult in practice. Currently in Korea, the Rural Development Administration operates a data platform and provides crop environment monitoring and open API services, while Smart Farm Korea is opening various datasets from cultivation farms.

 ERP·MES -like standard architecture and SOP must come first

It is worth noting that Japan’s data platform WAGRI, mentioned above(WAGRI), has introduced a system based on standard operating procedures(SOP) —procedure manuals that farms can use on their own—and that farms are presenting outcomes such as yields and work efficiency. Providing opportunities to compare the performance of data-centered agriculture through demonstrations with the performance of farms’ conventional agriculture can be seen as the beginning of data-driven agriculture. Currently, Korea is pursuing the external expansion of data-driven agriculture, including the Rural Development Administration’s agricultural satellite, but it seems necessary to consider the data composition architecture required for a full-cycle agricultural data platform, including MAS for decision-making. Smart factories in manufacturing were designed based on ERP, an integrated management system for the real world, and MES, an IoT-based manufacturing execution system, and the data collected here are stored in a cyber-physical system(CPS) to build digital twins.
Agricultural data systems are also operated with a similar architectural environment, and the configuration of ERP and the construction of MES suited to the agricultural environment can be operated through the advancement and standardization of smart farms. And data standardization and architecture design for a full-cycle agricultural system are needed in order to establish the standard configuration of a multi-collaborative-agent AI system for autonomous decision-making in various farming operations and cultivation management, and the corresponding standard operating procedures(SOP) . Convergence models across various industries are needed, and because it is difficult for farms to collect·and analyze data directly, it is necessary to train experts who visualize and analyze data and to see the emergence of related companies, and policy support is needed to provide solutions required to improve farm productivity and the environment and to create an ecosystem that brings in ICT-related companies. Through this, it is necessary to continuously spread verified results in productivity and climate response based on data standardization to secure the usefulness of smart agriculture and on multi-agent systems(MAS) built using that data, and priority should be given to fostering companies and training personnel capable of using the established data industrially.

In the case of farms, considering the practical problem that they are both producers of agricultural products and consumers of agricultural technology, making it difficult to realize economies of scale, government support measures need to move beyond a simple focus on research and development support in order to encourage the emergence of companies that build production systems and platforms using agricultural data. For crop growth and environmental data that can use smart agriculture data platforms, and full-cycle data platforms needed for equipment industries such as smart agricultural machinery, it is first necessary to establish a standard architecture and prepare utilization measures under ‘Korean Smart Farm Standard Generation 4’(tentative name). In addition, changes in social perception also appear necessary in terms of efforts by each field and industrial development to create data security technologies and added value. 

[Sources] 
1. Seok Jun-ho, 2024, “AI-Related Trends in Agriculture and Policy Implications”, World Agriculture, Autumn 2024 Issue
2. Shin Dong-cheol, 2019, “Current Status of Agricultural Big Data Use in Japan”, World Agriculture, July 2019 Issue
3. Noh Si-young et al., 2020, “ A Study on Plans to Establish an Agricultural Big Data Platform to Promote Smart Agriculture”, Journal of Korea Knowledge Information Technology Society Vol.15. No.5, pp 915-923
4. Smart Farm Big Data Platform https://www.n-farm.kr/home/
5. Agricultural Technology Data Platform(Rural Development Administration) https://adp.rda.go.kr/portal/pub/main.do
6. AI Helps Plant Cultivation(Data-Based Smart Agriculture Results), Electronic Times 2025.11.20. https://www.etnews.com/20251120000276
6. Amine Roukh et al., 2020, “Big Data Processing Architecture for Smart Farming”, January 2020, Procedia Computer Science, 177:78-85. https://www.researchgate.net/figure/WALLeSMART-system-components_fig1_346870440