QFabric — Quantum Network Emulation on FABRIC

QFabric is a programmable quantum-network emulation platform built on the FABRIC testbed. It runs BB84 quantum key distribution over a real FABRIC slice, with the quantum channel emulated in a P4/BMv2 data plane and the classical sifting traffic carried over a genuine FABRIC WAN link — so real latency, jitter, and loss enter the protocol naturally.

How it works

• Quantum channel (P4/BMv2): fiber attenuation is emulated as probabilistic packet drop on custom EtherType 0x7101 photon frames, P(loss) = 1 − 10−αL/10.
• Quantum nodes (Python): Alice (photon source) and Bob (detector with efficiency, dark counts, and a polarization-misalignment QBER ≈ (1−F)/2), with BB84 sifting and Shor–Preskill key-rate estimation.
• Classical channel: BB84 sifting over TCP across the FABRIC data plane between sites.

Cross-validation on FABRIC

The same scenario is run four ways — all on the slice — and checked for statistical agreement on QBER and secure key rate:

• QFabric (measured) — the real BMv2 emulation
• QFabric-sim — pure-Python model (switch node)
• SeQUeNCe 1.0 — native engine (Alice node, Python 3.12)
• NetSquid — native engine (Bob node)

Each simulator drives its own engine, so the comparison reflects independent physics; unavailable backends are reported as SKIPPED, never a false pass.

What's included

• A linear Jupyter workflow: overview → set up slice → run experiment → cross-validate → analysis → run-all-scenarios (with QBER/key-rate sweep figures vs distance and attenuation).
• One-command FABRIC deployment, the P4 program, the quantum-node emulator, a cross-validation framework, and a unit-test suite.

License: Apache-2.0. Source: github.com/kthare10/qfabric. See README.md, SPEC.md, and ROADMAP.md.