AGRARIAN: A Hybrid AI-Driven Architecture for Smart Agriculture

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The paper describes a framework for digital agriculture. It combines almost all available buzzwords in the field. Through this, the resulting framework is evident and not very innovative or surprising. The big deficit of the paper is that the framework is given as is and not critically discussed in terms of completeness (e.g. no connection to agricultural machinery), chances for implementation, etc.  The paper names many methodological approaches but refrains from critically discussing them in terms of e.g. feasibility, scale and problem solving power (e.g. the necessity of real time data analysis is not discussed (I am not aware of a farmer needing real time analysis of his/her sensor data for immediate action).

Author Response

The big deficit of the paper is that the framework is given as is and not critically discussed in terms of completeness (e.g. no connection to agricultural machinery), chances for implementation, etc.

Answer:

We thank the reviewer for highlighting this important point. In response, we have added a clarification paragraph in the Sensor Layer subsection of the manuscript that critically discusses the completeness and practical extensibility of the AGRARIAN framework, including its future integration with agricultural machinery. This new addition emphasizes AGRARIAN’s modular design and its capacity to interface with smart farm machinery through standardized protocols such as ISOBUS, MQTT, and OPC UA. It also outlines the real-world deployment efforts currently underway in various agricultural settings, demonstrating the system’s implementation readiness, scalability, and interoperability. We believe this addition addresses the reviewer’s concern by illustrating both the current capabilities and future direction of the AGRARIAN architecture in supporting end-to-end agricultural automation.

 

 

The paper names many methodological approaches but refrains from critically discussing them in terms of e.g. feasibility, scale and problem solving power (e.g. the necessity of real time data analysis is not discussed (I am not aware of a farmer needing real time analysis of his/her sensor data for immediate action).

Answer:

We appreciate the reviewer’s insightful comment. In response, we have added a new paragraph to the Data Processing Layer subsection of the manuscript that provides a critical discussion on the feasibility and relevance of the methodologies mentioned, particularly concerning real-time data analysis. This clarification acknowledges that not all agricultural applications require real-time analytics, and instead, the architecture supports a spectrum of processing models, from batch to real-time, depending on the use case. Specific examples—such as livestock anomaly detection, heat-induced irrigation control, and pest outbreak alerts—are given to demonstrate scenarios where real-time insights are not only feasible but also highly valuable. At the same time, the text clarifies that AGRARIAN’s flexible orchestration enables scalable deployments, aligning computational intensity with actual operational needs and cost-effectiveness. We believe this addition strengthens the paper by contextualizing the selected methodologies and better aligning the architecture with practical farming requirements.

Reviewer 2 Report

Comments and Suggestions for Authors

1. The abstract  focused on technical details but doesn't clearly highlight the study's innovations or practical contributions, such as reducing cloud dependency by 30%.

2. Start the abstract by clearly stating the research gap to better highlight the relevance of AGRARIAN.

3. The language is somewhat redundant. Trim the abstract and conclusion to avoid repeating technical details. The description of Figure 2 overlaps with the main text.

4. The literature review lists technologies but lacks critical analysis. For example, it doesn't clarify the limitations of existing DSS in edge computing.

5. Table 10 only lists technical mappings without an in-depth discussion of AGRARIAN's unique advantages. Add metrics like latency and energy consumption to quantify these benefits.

6. The paper lacks experimental validation, such as performance testing or case studies, leaving the technical advantages unsupported by data.

7. The figure captions are too simple. For instance, Figure 1 is only labeled "AGRARIAN high level approach."

8. There's an over-reliance on recent references from 2025, with insufficient support from classic theories.

9.Check for errors in the section numbering, such as the contradiction between "2. AGRARIAN architecture" and "2. Related Technologies," and the absence of sections 3 and 4.

Author Response

1. The abstract  focused on technical details but doesn't clearly highlight the study's innovations or practical contributions, such as reducing cloud dependency by 30%.

 

Answer:

We have revised the abstract to better highlight AGRARIAN’s architectural contributions and practical value, particularly its ability to reduce reliance on centralized cloud infrastructure through the use of edge computing and federated AI models. Rather than citing a specific percentage reduction—which cannot be empirically verified at this stage—we emphasize the flexible deployment and offline processing capabilities that make AGRARIAN more suitable for rural or bandwidth-constrained environments. This change more accurately reflects the system’s practical benefits without overreaching in quantifiable claims.

2. Start the abstract by clearly stating the research gap to better highlight the relevance of AGRARIAN.

 

Answer:

 

The revised abstract now begins by identifying the research gap—namely, the limitations of current smart agriculture systems in addressing rural connectivity, real-time decision-making, and cloud over-reliance. This framing helps position AGRARIAN as a relevant and timely solution.

3. The language is somewhat redundant. Trim the abstract and conclusion to avoid repeating technical details. The description of Figure 2 overlaps with the main text.

 

Answer:

 

We have revised the abstract and conclusion to reduce redundancy and streamline the language. Repetitive technical descriptions were removed or merged for clarity. Additionally, the description of Figure 2 has been condensed and refocused to avoid unnecessary overlap with the main text. These changes improve the readability and structure of the paper while preserving its technical depth.

4. The literature review lists technologies but lacks critical analysis. For example, it doesn't clarify the limitations of existing DSS in edge computing.

 

Answer:

 

We appreciate this observation. The scope of this paper extends beyond traditional agricultural DSS to include modern communication technologies, such as 5G slicing, satellite connectivity, and edge computing, which are critical to enabling real-time agricultural intelligence. To reflect this broader scope, we have revised the literature review to include a cross-disciplinary critique, highlighting not only the computational limitations of traditional DSS in edge environments (e.g., centralization, data delay), but also the communication bottlenecks that AGRARIAN overcomes through hybrid networking and dynamic slice orchestration. This approach aligns with the paper's aim of presenting a unified architecture that merges AI, agriculture, and communication engineering.

5. Table 10 only lists technical mappings without an in-depth discussion of AGRARIAN's unique advantages. Add metrics like latency and energy consumption to quantify these benefits.

 

Answer:

 

Thank you for the suggestion. We have enhanced the section related to Table 10 by incorporating quantitative performance metrics derived from our 5G slicing validation experiments. These metrics include latency reductions (down to ~10 ms) and modest energy trade-offs (1.5–3%) associated with different AGRARIAN slice configurations. We also expanded the narrative around the table to explain how AGRARIAN’s layered design allows dynamic reconfiguration, optimizing these metrics depending on the use case.

 

6. The paper lacks experimental validation, such as performance testing or case studies, leaving the technical advantages unsupported by data.

 

Answer:

 

This concern has been addressed through a newly added section titled “Preliminary Validation of AGRARIAN over 5G Network Slicing”. This section presents a structured set of slicing experiments, mapping AGRARIAN's Sensor Layer to uRLLC and Data Processing Layer to eMBB slices, thereby offering real measurements of latency and energy consumption. Figures 6 and 7 provide detailed 3D visualizations of energy dynamics, and the accompanying explanation contextualizes these findings as a validation of AGRARIAN’s efficiency and design alignment with low-latency, high-performance edge computing.

7. The figure captions are too simple. For instance, Figure 1 is only labeled "AGRARIAN high level approach."

 

Answer:

 

We have revised all figure captions to provide descriptive and informative labels. For example, Figure 1 is now titled: “AGRARIAN High-Level Architecture: Integrating IoT, Edge AI, and Hybrid 5G Connectivity for Smart Agriculture.” Captions for Figures 6 and 7 have also been expanded to clearly explain the data shown, the test conditions, and the relevance to AGRARIAN’s system validation.

 

8. There's an over-reliance on recent references from 2025, with insufficient support from classic theories.

 

Answer:

 

Thank you for this valuable point. The focus of this paper is on presenting cutting-edge, deployable technologies for smart agriculture, which necessitated referencing the most recent innovations in areas such as AI-based DSS, 5G slicing, and CubeSat integration. However, we recognize the importance of contextualizing these innovations within established theoretical foundations. To address this, we have incorporated classic references that underpin technologies such as network programmability (OpenFlow), protocol abstraction (RINA), and early AI applications in farming. This combination ensures the paper is both technically current and theoretically grounded.

 

9. Check for errors in the section numbering, such as the contradiction between "2. AGRARIAN architecture" and "2. Related Technologies," and the absence of sections 3 and 4.

 

Answer:

 

We have corrected the section numbering throughout the manuscript for clarity and consistency. The structure now follows a logical progression, beginning with 1. Introduction, 2. Related Technologies, 3. AGRARIAN Architecture, 4. Experimental Validation, and so on. These adjustments ensure a coherent flow and eliminate previous inconsistencies.

Reviewer 3 Report

Comments and Suggestions for Authors

This paper presents AI-driven framework for modern agriculture, integrating IoT sensors, UAVs, satellite-based Earth observation, edge computing, and hybrid communication networks to enhance precision farming and sustainable agricultural practices. The topic is interesting. I recommend its publication in the journal and some revisions are made.

(1) The abstract section needs improvement. Quantitative model evaluations should be supplemented.

(2) In the first two paragraphs of the introduction, only one reference is added at the end of each paragraph, which may make the analysis of the research background insufficiently solid.

(3) Lines 61-65: What are the solutions proposed by international researchers (e.g., from Israel and China) to this problem? What are the advantages of your proposed model (AGRARIAN)? In other words, what is the research gap?

(4) Line 83: The "2 related technologies" section should be carefully divided into several subsections.

(5) What does the practical application of the AGRARIAN architecture look like? Please provide detailed specifics.

(6) The discussion section should be supplemented.

Author Response

• The abstract section needs improvement. Quantitative model evaluations should be supplemented.

 

Answer:

 

We thank the reviewer for this observation. The abstract has been revised to reflect quantitative performance validation based on our 5G slicing experiments. Specific results—such as latency reductions to ~10 ms and energy overheads between 1.5% and 3% for low-latency slicing—have been added to demonstrate the practical effectiveness of the AGRARIAN architecture. These metrics help support the claimed benefits of the proposed hybrid AI system in real agricultural contexts.

 

• In the first two paragraphs of the introduction, only one reference is added at the end of each paragraph, which may make the analysis of the research background insufficiently solid.

 

Answer:

 

We appreciate the reviewer’s suggestion. The introduction has been updated with multiple references across key sentences, drawing from European agricultural policies, smart farming frameworks, and agricultural digitalization strategies. This includes citations from the European Commission, CAP Strategic Plans, Eurostat, and recent literature on Agriculture 4.0. These additions provide stronger support for the contextual background and reinforce the relevance of AGRARIAN.

 

• Lines 61-65: What are the solutions proposed by international researchers (e.g., from Israel and China) to this problem? What are the advantages of your proposed model (AGRARIAN)? In other words, what is the research gap?

 

Answer:

 

We thank the reviewer for this insightful comment. The introduction has been expanded to reference international research efforts, including Israel’s use of sensor-based irrigation automation and China’s developments in UAV-based remote sensing, blockchain-enabled traceability, and disease detection systems. While these contributions demonstrate significant technological advancement, they are often vertically siloed, reliant on centralized infrastructure, and lack support for interoperability and dynamic adaptation in resource-constrained environments. To address this gap, AGRARIAN introduces a modular, multi-layered AI-driven architecture that integrates edge computing, federated learning, and hybrid communication (5G/LEO satellite) into a single framework. Unlike prior models, AGRARIAN offers real-time responsiveness, energy-aware decision-making, and seamless coordination across layers—allowing it to operate effectively in rural and disconnected settings. These enhancements have been clearly articulated in the revised manuscript, defining AGRARIAN’s contribution as a scalable and deployable alternative to fragmented or cloud-reliant systems.

• Line 83: The "2 related technologies" section should be carefully divided into several subsections.

 

Answer:

 

We thank the reviewer for this helpful suggestion. In response, the “Related Technologies” section has been reorganized into multiple thematic subsections to enhance clarity, structure, and depth of analysis. The revised section now includes clearly labeled parts that focus individually on areas such as AI for crop protection and environmental monitoring, decision support systems for water and livestock management, circular bioeconomy strategies, digital agriculture integration, and edge computing with federated AI.

• What does the practical application of the AGRARIAN architecture look like? Please provide detailed specifics.

 

Answer:

 

We appreciate this request and have added detailed application scenarios to the revised manuscript. For example, in livestock farming, AGRARIAN processes real-time data from biometric sensors through edge AI nodes to detect health anomalies, enabling timely alerts and automated intervention. In crop management, UAV imagery and ground sensor data are used to predict irrigation needs and monitor plant health. These use cases are deployed using uRLLC slicing for sensor layer responsiveness and eMBB slicing for AI-driven data processing, allowing the system to function even with intermittent connectivity.

 

Additionally, the manuscript now includes information on pilot implementations in vineyards and livestock farms, validating AGRARIAN’s performance in operational conditions. The system’s design supports inter-layer communication, hybrid backhaul via 5G and satellite, and future integration with agricultural machinery using ISOBUS and OPC UA standards. These updates clearly illustrate AGRARIAN’s practicality, adaptability, and readiness for field deployment, reinforcing its value as a robust digital agriculture platform.

 

 

• The discussion section should be supplemented.

 

Answer:

 

The discussion section has been extended to critically assess AGRARIAN’s current deployment scope, modular design benefits, and real-world implementation feasibility. We now reflect on trade-offs between latency and energy, future integration with agricultural machinery via protocols like ISOBUS and OPC UA, and how AGRARIAN addresses barriers like connectivity gaps and computational resource constraints in rural settings.

 

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

The author has revised the problems in the paper.I have no more questions.

Reviewer 3 Report

Comments and Suggestions for Authors

The manuscript has been significantly improved. I recommend publishing it in its current form.