apuf 0.2.0

A library for simulating Arbiter PUFs and other delay-based PUFs.

APUF

A Python package for simulating delay-based Physical Unclonable Functions (PUFs) using Lim's Linear Additive Delay Model (LADM). This module supports Arbiter PUFs (APUFs) and XORPUFs. Provides challenge generation, response simulation (with noise), and bit-packing utilities.

Features

• $d$-layer Arbiter PUF (APUF): simulate $d$ layers, compute $k$-bit responses given $k$ challenges ($d$-bit binary vectors).
• XORPUF (XORPUF): simulate multiple APUF instances, compute responses and XOR them for increased complexity.
• Challenge generators:
  • generate_k_challenges(k, d, seed)
  • generate_n_k_challenges(n, k, d, seed)
  • generate_challenges_mp(n, k, d, seed, processes) for parallel batch generation.
• Response data-type (Response): Both APUF and XORPUF return a $k$-bit response given $k$ challenges. Response provides a conventient interface to study PUF challenge-response behaviours.

Installation

pip install apuf

Requires Python 3.9+ and numpy.

Quick Start

Import the module and call the functions directly:

from apuf import APUF, XORPUF, generate_k_challenges

# 1) Single APUF
apuf = APUF(d=64, mean=0.0, std=0.05)

# 2) Generate 10 random challenges (phase vectors)
chals = generate_k_challenges(k=10, d=64, seed=42)  # shape (65, 10)

# 3) Measure responses twice with Gaussian noise (mean=0, std=0.005)
resp1 = apuf.get_responses(chals, nmean=0.0, nstd=0.005)
resp2 = apuf.get_responses(chals, nmean=0.0, nstd=0.005)

# 4) Due to measurement noise, resp1 and resp2 may be different
diff = resp1 - resp2
#   4.1) We can compute Hamming distance
hamming_distance = resp1.dist(resp2)
#   4.2) And fractional/normalized HD
fHD = resp1.dist(resp2)/len(resp1)

# 5) Pack responses into bytes for convenience
resp_bytes = APUF.compact_responses(resp)
print(resp, resp_bytes)

API Reference

Below you will find a short reference for the core classes and functions.

Classes

Response - Represents a PUF response vector.

• Bitwise operations: ^, &, |, ~
• .dist(other) returns Hamming distance.
• .hw returns Hamming weight (number of 1s).

APUF(d: int = 128, mean: float = 0.0, std: float = 0.05) - Simulates a single APUF.

• get_responses(chals: np.ndarray, nmean: float = 0.0, nstd: float = 0.005) -> Response
• compact_responses(resp: Response) -> bytes

XORPUF(children: list[APUF]) - Simulates an XORPUF by XORing responses of child APUFs.

get_responses(chals: np.ndarray) -> Response

Functions

• generate_k_challenges(k: int, d: int, seed: int = None) -> np.ndarray
• generate_n_k_challenges(n: int, k: int, d: int, seed: int = None) -> np.ndarray
• generate_challenges_mp(n: int, k: int, d: int, seed: int = None, processes: int = cpu_count()) -> list[np.ndarray]

Full docstrings and type hints are available in code.

Testing

Run the provided unit tests with unittest:

git clone https://github.com/nikita-tripathi-geo/APUF-simulation.git
cd APUF-simulation
pip install -r requirements.txt
python -m unittest tests/*.py

Contributing

Contributions, issues and feature requests are welcome! Please see Issues.

License

This project is licensed under the MIT License – see the LICENSE file for details.

MIT License (c) Nikita Tripathi