Semantic-Aware IoT Dataset for Smart Agriculture

# Semantic-Aware UAV-IoT Dataset for Smart Agriculture

## Overview

This dataset contains 60,000 samples simulating smart agriculture scenarios where UAVs and ground IoT sensors are used to monitor crop conditions. It is developed to support research in Digital Twin architectures, semantic communication, and Federated Deep Reinforcement Learning (FDRL).

## Key Applications

- Semantic Tagging of Agricultural States (e.g., N-deficiency, Pest-risk)

- FDRL-based Optimization of UAV-IoT Tasks

- Simulation of Service Migration, Latency Impact, and Energy Efficiency

- Real-time Decision Making in Digital Twin Systems

## Dataset Format

File: `semantic_agriculture_dataset_60000_with_metadata.csv`

### Fields Description

| Column Name              | Description |

|--------------------------|-------------|

| Zone_ID                  | ID of the monitored field zone |

| Image_Source_ID          | Identifier for UAV image captured |

| Image_Type               | Type of image captured: Multispectral or RGB |

| NDVI                     | Normalized Difference Vegetation Index (UAV) |

| NDRE                     | Red Edge Index (UAV) |

| RGB_Damage_Score         | Damage level from RGB imagery |

| UAV_Timestamp            | Time of UAV operation or sensor read |

| N, P, K                  | Soil nutrient content values (mg/kg) |

| Moisture                 | Soil moisture in percentage |

| pH                       | Soil acidity level |

| Temperature              | Air temperature in °C |

| Humidity                 | Relative humidity in percentage |

| NDI_Label                | Nutrition Deficiency Level (Low/Medium/High) |

| PDI_Label                | Pest Density Level (Low/Medium/High) |

| Semantic_Tag             | Meaningful tags representing condition |

| Action_Suggested         | Recommended action (e.g., Irrigate, Fertilize) |

| Energy_Consumed_mAh      | Estimated energy usage for the task |

| Latency_ms               | Communication or action latency |

| Current_Node             | Origin of task/service (Edge or Cloud node) |

| Migrated_To              | New location after service migration |

| Migration_Required       | Whether migration was triggered |

| Migration_Timestamp      | Time migration occurred, if any |

## Instructions for Use

1. Load the CSV using pandas or any data processing tool.

2. Use Semantic_Tag, NDVI, RGB_Damage_Score to train classifiers for crop status prediction.

3. Use Latency and Energy_Consumed_mAh to simulate efficiency comparisons.

4. Migration columns can support reinforcement learning-based service placement models.

5. Dataset supports tasks like semantic-aware offloading, DT-based resource orchestration, and FDRL model training.

## Citation

If using this dataset, please cite:

Johna Yova, "Semantic-Aware UAV-IoT Dataset for Smart Agriculture Applications," IEEE DataPort, 2025.