README.md

metadata

license: mit
tags:
  - communications
  - test-vectors
  - polar-codes
  - turbo-codes
pretty_name: Arkastone Test Vectors
task_categories:
  - other

Arkastone Test Vectors

This dataset contains structured, versioned, and reproducible test vectors used in the Arkastone simulation framework for validating various encoder and decoder modules in a 5G-like communication stack.

The test vectors are grouped by subsystem — coding, tx, and rx — and organized for clarity, reproducibility, and CI/CD integration.

---

📁 Folder Structure

• coding/ – Reference encoder data, matrices, and metadata (e.g., Polar, NR5G)
• rx/ – Test inputs and expected outputs for decoder validation (SC, SCF, etc.)
• tx/ – Encoder test vectors for LTE-like Turbo Convolutional Coding (TBCC)

---

✅ Highlights

• Files follow naming convention:
  ppile_<scheme>_n<N>_k<K>_[extra]_Q<quant>.in/.out
• All .in files are input to decoder/encoder modules
• All .out files represent validated output (bit-level)
• Large metadata is compressed (.tar), grouped under polar_nr5g_wrapper
• Quantization schemes follow Q0 fixed-point rules (unless noted)
• Readme files are provided in each directory for more detailed usage

---

💡 Design Principles

• Avoid deep nesting for visibility and accessibility
• Ensure reproducibility across CI and local testing
• Separate heavy test vectors from source code repositories
• Enable scalable testing without Git LFS constraints
• Hosted externally for easy integration and public collaboration

---

📜 License

All test vectors and associated metadata are released under the MIT License unless stated otherwise.

---

🧠 Notes

• This structure is optimized for Hugging Face Datasets and integrates cleanly with Python-based simulations.
• You may reference these vectors from simulation scripts or GitHub Actions using a downloader script and manifest.

---

Maintained by Furkan Ercan