Skip to main content

A transpilation plugin for Qiskit with a modified scheduling stage that encodes (classical) information modifying the angles of RZGates without leading to changes in the computation results

Project description

qiskit-leaky-scheduling

Build & Test Python Wheel Package PyPI - Version PyPI - Wheel arXiv

[!NOTE] This plugin was developed to demonstrate the importance of reproducible builds in the Qiskit quantum computing workflow. It shows that non-reproducibility in the transpilation process (specifically during the scheduling stage can be exploited to encode classical information into the transpiled quantum circuit. If an attacker subsequently gains access to the job description, this can lead to the leakage of confidential data.

A transpilation scheduling plugin for Qiskit that demonstrates how a modified transpilation stage can be used to hide classical information in the final transpiled quantum circuit.

The current implementation, by default, tries to encode the HSLU logo into the transpiled circuit. Custom data will be used if available in builtins.data (see the example below). If data is too large to encode into the given circuit, the unmodified circuit is returned. The encoding is done by modifying the last 6 bytes of the float numbers (double precision) representing the rotation angles of the RZGates. These bytes only affect the fraction part of the number leading to slightly different rotation gates. However, since current hardware is still quite noisy, the output of the original and modified circuit is indistinguishable in practice.

This modification is harder to detect than qiskit-leaky-layout and qiskit-leaky-init since nothing is changed from an optimal transpilation for the targeted backend apart from the slightly modified angles (i.e., no additional registers, same layout, etc.).

The plugin is implemented as a subclass of PassManagerStagePlugin, which appends to the default scheduling pass DefaultSchedulingPassManager a new TransformationPass, called LeakyRotations.

Encoded data can be recovered with recover_data() implemented in the decoder module. See the example below.

Installation

git clone https://github.com/iyanmv/qiskit-leaky-scheduling.git
cd qiskit-leaky-scheduling
pip install .

Example

import builtins
import io

from PIL import Image
from qiskit.circuit.random import random_circuit
from qiskit.transpiler.preset_passmanagers import generate_preset_pass_manager
from qiskit.transpiler.preset_passmanagers.plugin import list_stage_plugins
from qiskit_ibm_runtime.fake_provider import FakeKyoto

from qiskit_leaky_scheduling import recover_data

print(list_stage_plugins("scheduling"))

backend = FakeKyoto()
pm = generate_preset_pass_manager(
    backend=backend,
    optimization_level=3,
    scheduling_method="leaky_rotations",
    seed_transpiler=0,
)

qc = random_circuit(
    num_qubits=7, depth=3, max_operands=2, measure=True, reset=False, seed=0
)

# Uncomment to encode this custom data instead of the HSLU logo
# builtins.data = b"My secret data encoded in RZ gates."
isa_qc = pm.run(qc)

recovered_img = recover_data(isa_qc)[:328]
# recovered_data = recover_data(isa_qc)[:35]

Image.open(io.BytesIO(recovered_img)).show()
# print(recovered_data)

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

qiskit_leaky_scheduling-0.2.4.tar.gz (6.9 kB view details)

Uploaded Source

Built Distribution

If you're not sure about the file name format, learn more about wheel file names.

qiskit_leaky_scheduling-0.2.4-py3-none-any.whl (7.2 kB view details)

Uploaded Python 3

File details

Details for the file qiskit_leaky_scheduling-0.2.4.tar.gz.

File metadata

  • Download URL: qiskit_leaky_scheduling-0.2.4.tar.gz
  • Upload date:
  • Size: 6.9 kB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.13.12

File hashes

Hashes for qiskit_leaky_scheduling-0.2.4.tar.gz
Algorithm Hash digest
SHA256 d54f56b3d68fbcdad5211394d801f9e2cffdce4cf15b83e1015adbb89ced1468
MD5 2489a55fe805f825d37ca9349e806a87
BLAKE2b-256 8f97ecdcca49b8c63150f32387ee1702214c8879ce4cc8b2f1e4b764895e2d69

See more details on using hashes here.

Provenance

The following attestation bundles were made for qiskit_leaky_scheduling-0.2.4.tar.gz:

Publisher: publish-to-pypi.yml on iyanmv/qiskit-leaky-scheduling

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file qiskit_leaky_scheduling-0.2.4-py3-none-any.whl.

File metadata

File hashes

Hashes for qiskit_leaky_scheduling-0.2.4-py3-none-any.whl
Algorithm Hash digest
SHA256 75ff3008239da20b1eaab88814ea82ee29c5fce1306a7079bb65a44de0583bfe
MD5 cf874f306e612ca8cbfb1ee26fd9ba9c
BLAKE2b-256 990483fefb2319e1eb34784bccb0db8a985cee65c8ca41baaa0fe47c51b96482

See more details on using hashes here.

Provenance

The following attestation bundles were made for qiskit_leaky_scheduling-0.2.4-py3-none-any.whl:

Publisher: publish-to-pypi.yml on iyanmv/qiskit-leaky-scheduling

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

AWS Cloud computing and Security Sponsor Datadog Monitoring Depot Continuous Integration Fastly CDN Google Download Analytics Pingdom Monitoring Sentry Error logging StatusPage Status page