feat: standalone sixg-bench deployment with tiered optional packages

.dockerignore

@@ -0,0 +1,21 @@
+# Files that don't need to be in the Docker build context
+.git
+target/
+.github/
+docs/
+*.md
+*.png
+*.svg
+*.pdf
+
+# Python artefacts
+__pycache__/
+*.pyc
+*.pyo
+.venv/
+venv/
+
+# Editor / OS
+.DS_Store
+*.swp
+*.swo

.github/workflows/ci.yml

@@ -78,10 +78,71 @@ jobs:
       - name: Run PHY Level-2 baseline comparison tests
         run: cargo test -p sixg-phy --features=baseline-comparison
 
+  # ── Feature matrix: verify every tier builds cleanly ─────────────────────
+  matrix-features:
+    name: Feature matrix (${{ matrix.features }})
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+    strategy:
+      matrix:
+        features:
+          - "" # default (core tier — no features)
+          - "baseline-comparison"
+          - "plotting"
+    steps:
+      - uses: actions/checkout@v4
+      - uses: dtolnay/rust-toolchain@stable
+      - uses: Swatinem/rust-cache@v2
+      - name: Build with features "${{ matrix.features }}"
+        run: |
+          if [ -z "${{ matrix.features }}" ]; then
+            cargo build --workspace
+          else
+            cargo build --workspace --features "${{ matrix.features }}"
+          fi
+      - name: Test with features "${{ matrix.features }}"
+        run: |
+          if [ -z "${{ matrix.features }}" ]; then
+            cargo test --workspace
+          else
+            cargo test --workspace --features "${{ matrix.features }}"
+          fi
+
+  # ── Docker build (core tier) — publish to ghcr.io on push to main ────────
+  docker-build:
+    name: Docker build (core tier)
+    runs-on: ubuntu-latest
+    permissions:
+      contents: read
+      packages: write
+    steps:
+      - uses: actions/checkout@v4
+      - name: Set up Docker Buildx
+        uses: docker/setup-buildx-action@v3
+      - name: Log in to GitHub Container Registry
+        if: github.event_name == 'push' && github.ref == 'refs/heads/main'
+        uses: docker/login-action@v3
+        with:
+          registry: ghcr.io
+          username: ${{ github.actor }}
+          password: ${{ secrets.GITHUB_TOKEN }}
+      - name: Build (and push on main)
+        uses: docker/build-push-action@v5
+        with:
+          context: .
+          file: Dockerfile
+          push: ${{ github.event_name == 'push' && github.ref == 'refs/heads/main' }}
+          tags: |
+            ghcr.io/${{ github.repository }}:latest
+            ghcr.io/${{ github.repository }}:${{ github.sha }}
+          cache-from: type=gha
+          cache-to: type=gha,mode=max
+
   final-check:
     name: Final Check (format + doc-sync)
     runs-on: ubuntu-latest
-    needs: [test, clippy, fmt, doc-sync, dep-graph, baseline-comparison]
+    needs: [test, clippy, fmt, doc-sync, dep-graph, baseline-comparison, matrix-features]
     steps:
       - uses: actions/checkout@v4
       - uses: dtolnay/rust-toolchain@stable
@@ -91,3 +152,4 @@ jobs:
         run: cargo fmt --all -- --check
       - name: Verify doc-code sync
         run: python3 scripts/check_doc_sync.py
+

Cargo.toml

@@ -8,6 +8,10 @@ license.workspace = true
 name = "sixg"
 path = "src/main.rs"
 
+[[bin]]
+name = "sixg-bench"
+path = "src/bin/bench.rs"
+
 [[example]]
 name = "exp_001_dfrc_pareto_frontier"
 path = "experiments/exp_001_dfrc_pareto_frontier/run.rs"
@@ -61,6 +65,19 @@ tracing-subscriber.workspace = true
 serde.workspace = true
 serde_json.workspace = true
 
+[features]
+## Core tier: zero native dependencies.  Always enabled.
+default = []
+## Baseline tier: enable Level-2 CSV baseline comparison tests in 6g-phy.
+## Activate with: cargo build --features=baseline-comparison
+baseline-comparison = ["sixg-phy/baseline-comparison"]
+## ONNX tier: gate real ort-based ONNX sentence-transformer inference.
+## Requires libonnxruntime and a model file.  See install.sh --onnx.
+onnx = ["sixg-ai/onnx", "sixg-semantic/onnx"]
+## Plotting tier: compile-time marker enabling CSV / JSON artefact output.
+## Python plots are handled separately via scripts/plot_*.py.
+plotting = []
+
 [workspace]
 resolver = "2"
 members = [

Dockerfile

@@ -0,0 +1,47 @@
+# ── Stage 1: builder ──────────────────────────────────────────────────────────
+# Uses the official Rust slim image so cargo + rustc are pre-installed.
+FROM rust:1.87-slim AS builder
+
+WORKDIR /build
+
+# Cache dependency downloads before copying source.
+COPY Cargo.toml Cargo.lock ./
+COPY crates/ crates/
+COPY src/ src/
+COPY experiments/ experiments/
+
+# Build the sixg-bench binary and all example experiment binaries.
+RUN cargo build --release \
+    && cargo build --release --examples
+
+# ── Stage 2: runtime ──────────────────────────────────────────────────────────
+# Minimal Debian image — no Rust toolchain, only the compiled artefacts.
+FROM debian:bookworm-slim AS runtime
+
+LABEL org.opencontainers.image.title="sixg-bench" \
+      org.opencontainers.image.description="Standalone 6G simulation bench" \
+      org.opencontainers.image.source="https://github.com/j143/6g" \
+      org.opencontainers.image.licenses="Apache-2.0"
+
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    ca-certificates \
+    && rm -rf /var/lib/apt/lists/*
+
+WORKDIR /bench
+
+# Copy the bench binary.
+COPY --from=builder /build/target/release/sixg-bench /usr/local/bin/sixg-bench
+
+# Copy all pre-built experiment binaries.
+COPY --from=builder /build/target/release/examples/ /bench/target/release/examples/
+
+# Copy experiment config files (read by the bench at runtime).
+COPY experiments/ /bench/experiments/
+
+# Copy bundled baseline CSV files.
+COPY baselines/ /bench/baselines/
+
+ENV SIXG_BASELINES=/bench/baselines
+
+ENTRYPOINT ["sixg-bench"]
+CMD ["--help"]

Dockerfile.onnx

@@ -0,0 +1,56 @@
+# ── Stage 1: builder (ONNX tier) ─────────────────────────────────────────────
+FROM rust:1.87-slim AS builder
+
+# ort (ONNX Runtime Rust bindings) requires libonnxruntime.
+# The build-time auto-download fetches it into /usr/local/lib.
+ENV ORT_DYLIB_PATH=/usr/local/lib
+ENV LIBONNXRUNTIME_URL=https://github.com/microsoft/onnxruntime/releases/download/v1.18.0/onnxruntime-linux-x64-1.18.0.tgz
+
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    curl tar pkg-config libssl-dev \
+    && rm -rf /var/lib/apt/lists/* \
+    && curl -fsSL "$LIBONNXRUNTIME_URL" -o /tmp/ort.tgz \
+    && tar -xzf /tmp/ort.tgz -C /tmp \
+    && cp /tmp/onnxruntime-linux-x64-*/lib/libonnxruntime.so* /usr/local/lib/ \
+    && ldconfig \
+    && rm -rf /tmp/ort.tgz /tmp/onnxruntime-linux-x64-*
+
+WORKDIR /build
+
+COPY Cargo.toml Cargo.lock ./
+COPY crates/ crates/
+COPY src/ src/
+COPY experiments/ experiments/
+
+# Build with the full onnx feature tier.
+RUN cargo build --release --features=onnx \
+    && cargo build --release --examples --features=onnx
+
+# ── Stage 2: runtime (ONNX tier) ─────────────────────────────────────────────
+FROM debian:bookworm-slim AS runtime
+
+LABEL org.opencontainers.image.title="sixg-bench (ONNX tier)" \
+      org.opencontainers.image.description="6G bench with real ONNX sentence-transformer" \
+      org.opencontainers.image.source="https://github.com/j143/6g" \
+      org.opencontainers.image.licenses="Apache-2.0"
+
+RUN apt-get update && apt-get install -y --no-install-recommends \
+    ca-certificates libgomp1 \
+    && rm -rf /var/lib/apt/lists/*
+
+WORKDIR /bench
+
+COPY --from=builder /usr/local/lib/libonnxruntime.so* /usr/local/lib/
+RUN ldconfig
+
+COPY --from=builder /build/target/release/sixg-bench /usr/local/bin/sixg-bench
+COPY --from=builder /build/target/release/examples/ /bench/target/release/examples/
+COPY experiments/ /bench/experiments/
+COPY baselines/ /bench/baselines/
+
+ENV SIXG_BASELINES=/bench/baselines
+# Mount your .onnx model file at this path to enable real inference:
+ENV SIXG_ONNX_MODEL=/bench/models/all-MiniLM-L6-v2.onnx
+
+ENTRYPOINT ["sixg-bench"]
+CMD ["--help"]

README.md

@@ -3,6 +3,50 @@
 A Rust skeleton for a 6G wireless system stack, inspired by
 [Qualcomm's 6G System Architecture research](https://www.qualcomm.com/research/6g/system-architecture).
 
+## Quickstart (30 seconds)
+
+```bash
+git clone https://github.com/j143/6g && cd 6g
+./install.sh           # installs Rust if needed, builds the bench
+./target/release/sixg-bench list          # list all experiments
+./target/release/sixg-bench run --all     # run all 9 experiments
+./target/release/sixg-bench validate      # run all Validate checks
+```
+
+### With optional packages
+
+```bash
+./install.sh --baselines   # + Level-2 CSV baseline comparisons
+./install.sh --plotting    # + Python matplotlib plots
+./install.sh --onnx        # + real ONNX sentence-transformer inference
+./install.sh --all         # everything above
+```
+
+### Docker (zero-install)
+
+```bash
+# Core bench
+docker run --rm ghcr.io/j143/6g sixg-bench run --all
+
+# With baseline comparisons
+docker-compose run bench-baselines
+
+# With ONNX inference (place models/all-MiniLM-L6-v2.onnx in repo root first)
+docker-compose run bench-onnx
+```
+
+### Plotting
+
+```bash
+pip install -r requirements-plot.txt
+python3 scripts/plot_phy.py      # PHY: path loss + BER curves
+python3 scripts/plot_mac.py      # MAC: Jain fairness + HARQ rounds
+python3 scripts/plot_isac.py     # ISAC: DFRC Pareto frontier
+python3 scripts/plot_semantic.py # Semantic: compression vs task success
+```
+
+---
+
 ## Architecture
 
 The workspace is organised as a set of Cargo crates, each representing a
@@ -83,18 +127,29 @@ Cross-cutting subsystems
 
 ### Prerequisites
 
-* Rust 1.70 or later (`rustup update stable`)
+* Rust 1.70 or later (the `install.sh` script installs it automatically if absent)
+
+### One-command setup
+
+```bash
+./install.sh
+./target/release/sixg-bench --help
+```
 
-### Build
+### Manual build
 
 ```bash
-cargo build
+cargo build --release
+./target/release/sixg-bench list
 ```
 
-### Run
+### Run experiments
 
 ```bash
-cargo run
+cargo run --example exp_001_dfrc_pareto_frontier
+cargo run --example exp_002_phy_baseline_comparison
+# … or run all via the bench:
+./target/release/sixg-bench run --all
 ```
 
 ### Test

crates/6g-ai/Cargo.toml

@@ -8,3 +8,9 @@ license.workspace = true
 sixg-common = { path = "../6g-common" }
 thiserror.workspace = true
 serde.workspace = true
+
+[features]
+## ONNX tier: enable real ort-based ONNX inference.
+## Requires libonnxruntime.so and a model file; see install.sh --onnx.
+## Without this feature the built-in deterministic simulation is used.
+onnx = []

crates/6g-ai/src/onnx_model.rs

@@ -5,20 +5,21 @@
 //! a 128-dimensional word-hash feature vector to a 32-dimensional L2-normalised
 //! semantic embedding.
 //!
-//! ## Why simulate rather than use the ONNX runtime?
+//! ## Simulation vs real ONNX runtime
 //!
-//! The `ort` crate requires a native `libonnxruntime` shared library and a
-//! pre-trained model file — both are infeasible to bundle in a pure-Rust
-//! workspace CI.  This module implements the same **interface** as a real ONNX
-//! model so that:
+//! **Without `--features=onnx`** (default): the built-in deterministic MLP
+//! simulation is used — no native library or model file is required.
 //!
-//! 1. The API is stable and ready to swap in `ort::Session` when the runtime
-//!    and model file are available (see `docs/6g-ai.md`).
-//! 2. The closed-form approximation produces semantically-informed embeddings
-//!    (similar texts → similar embeddings via shared word-hash features) without
-//!    a runtime dependency.
+//! **With `--features=onnx`**: the `onnx` feature flag is set, and
+//! `cfg!(feature = "onnx")` is `true` at runtime.  Callers that want to
+//! swap in `ort::Session` (the real ONNX Runtime) should check this flag
+//! and load the model file from the path in `SIXG_ONNX_MODEL`.  The
+//! public API ([`OnnxModel::forward`], [`AiModel::predict`]) is identical
+//! in both modes so downstream code needs no changes.
 //!
-//! ## Mathematical model
+//! See `install.sh --onnx` and `docs/6g-ai.md` for the setup steps.
+//!
+//! ## Mathematical model (simulation)
 //!
 //! ```text
 //! x  ← 128-dim L2-normalised word-hash feature vector (f32)

crates/6g-core/src/nrf.rs

@@ -169,7 +169,7 @@ impl Nrf {
             .values()
             .filter(|p| p.nf_type == nf_type && p.active)
             .collect();
-        matches.sort_by(|a, b| b.capacity.cmp(&a.capacity));
+        matches.sort_by_key(|b| std::cmp::Reverse(b.capacity));
         matches
     }
 
@@ -185,7 +185,7 @@ impl Nrf {
             .values()
             .filter(|p| p.active && p.has_capability(capability))
             .collect();
-        matches.sort_by(|a, b| b.capacity.cmp(&a.capacity));
+        matches.sort_by_key(|b| std::cmp::Reverse(b.capacity));
         matches
     }
 

crates/6g-semantic/Cargo.toml

@@ -9,3 +9,7 @@ sixg-common = { path = "../6g-common" }
 sixg-ai     = { path = "../6g-ai" }
 thiserror.workspace = true
 serde.workspace = true
+
+[features]
+## Propagate the ONNX tier from 6g-ai.
+onnx = ["sixg-ai/onnx"]