EdgeSec-Benchmark: Edge-IoT Security Resilience Dataset (ESP32, Raspberry Pi 5, Hardware-AES, AES-GCM)

Experimental Dataset: Hardware-Accelerated Security in Edge IoT

This repository contains the complete telemetry and performance logs from a longitudinal experimental campaign evaluating the resilience of ESP32-based secure edge nodes communicating with a Raspberry Pi 5 Gateway. The data was collected to validate the performance of hardware-accelerated AES-GCM encryption under conditions of signal degradation (-88 dBm), network congestion (Bufferbloat), and power cycling.

📄 ASSOCIATED MANUSCRIPT

This dataset supports the findings presented in the research article:

"Cross-Layer Benchmarking of Hardware-Accelerated AES-GCM for Edge IoT under Wi-Fi Degradation and Congestion"

📂 DATASET STRUCTURE

EdgeSec_Benchmark_Dataset/
│
├── 01_Raw_Data/                   (Original Server Logs - Raspberry Pi 5)
│   ├── Day-1/
│   │   ├── Baseline_Server_Day-1.csv
│   │   ├── Noise_Server_Day-1.csv
│   │   ├── Distance_Server_Day-1.csv
│   │   ├── Stress_Server_Day-1.csv
│   ├── Day-2/ ...
│   ├── Day-3/ ...
│   └── Client_Serial_Logs/        (Forensic Boot Logs - ESP32)
│       └── Stress_Client_DayX.txt
│
├── 02_Processed_Data/             (Cleaned & Merged)
│   ├── Total_Baseline.csv
│   ├── Total_Noise.csv
│   ├── Total_Distance.csv
│   └── Total_Stress.csv
│
├── 03_Analysis_Code/              (Reproducibility)
│   ├── EdgeSec_Reproducibility_Analysis.ipynb
│   └── extract_scientific_datapoints.py
│
└── README.txt

📊 COLUMN DEFINITIONS

• seq: Sequence ID for calculating Packet Loss.
• latency_ms: End-to-End Latency (Server_Rx - Client_Tx).
• rssi: WiFi Signal Strength (dBm).
• crypto_time_us: Encryption overhead (microseconds).
• heap_free: Available memory (Bytes).
• throughput: Bandwidth usage (Bytes/sec).

⚠️ DATA NOTE (NTP Exclusions)

While the Raw Data contains 30,953 packets, 322 packets (1.04%) were generated prior to client-side NTP synchronization. These were analytically excluded from the Processed Data to preserve precise end-to-end latency calculations, resulting in exactly 30,631 valid data points.

🛠 USAGE & REPRODUCIBILITY

The included Jupyter Notebook (EdgeSec_Reproducibility_Analysis.ipynb) and Python extraction script contain the logic required to process these CSVs and reproduce all statistical figures and Kruskal-Wallis H-tests found in the associated research manuscript.

Required Environment: Python 3.x (Tested on 3.14.0). To avoid Kernel execution errors (WinError 2), please ensure the following dependencies are installed in your virtual environment before running the notebook:

• pandas>=2.0.0
• numpy>=1.24.0
• scipy>=1.10.0
• seaborn>=0.12.0
• matplotlib>=3.7.0
• jupyter>=1.0.0