SUSTAINET assessment framework — figure-generation code

Self-contained code that reproduces the evaluation figures of the paper "Beyond Isolated Optimization: A Framework for Multi-Dimensional Assessment of Sustainable and Resilient Communication Networks." Each of the six use cases is a small analytical/simulation model; all share one implementation of the formal assessment model (normalization, Pareto dominance, empirical-front extraction, hypervolume, distance-to-front, and change classification).

Requirements

Python 3.9+ with:

pip install -r requirements.txt        # numpy, matplotlib

Reproduce all figures

From this directory:

python3 run_all.py

This regenerates every figure into figures/ (PDF used in the paper, plus a .png preview). Individual figures can also be produced by running a single script, e.g. python3 uc1_sleep_modes.py.

Files

File	Role	Paper figure
pareto.py	Shared harness: min–max normalization, dominance, empirical Pareto front, hypervolume, distance-to-front, change classification (strict improvement / trade-off improvement / burden shift), and the common 2-D plotting style. Imported by every use-case script.	—
plot_conceptual_goal.py	Conceptual illustration (synthetic data).	Fig. 2
uc1_sleep_modes.py	UC1 — base-station sleep modes (M/M/1 with setup; EARTH power model, advanced sleep modes).	Fig. 4
uc2_streaming_abr.py	UC2 — adaptive video streaming (ABR vs. QoE vs. energy; AR(1) capacity).	Fig. 5
uc3_edge_cloud.py	UC3 — edge vs. cloud placement (RTT, power, embodied carbon, availability; localization constraint).	Fig. 6
uc4_lpwan.py	UC4 — LPWAN technology selection (AoI vs. battery lifetime; duty-cycle and coverage constraints).	Fig. 7
uc5_redundancy.py	UC5 — redundancy provisioning (unavailability vs. fiber cost; availability floor).	Fig. 8
uc6_autoscaling.py	UC6 — elastic scaling (M/M/c Erlang-C; wait vs. power vs. scaling churn).	Fig. 9

Each use-case script also writes a *_points.json snapshot of its evaluated operating points and the resulting classification, and prints a short summary (baseline, highlighted candidate, distances) to the console.

Notes

• Parameters are literature-based, order-of-magnitude values or explicit assumptions (see each script's module docstring for sources). The figures illustrate the methodology; they are not measurements of a deployed system.
• Random seeds are fixed inside the scripts, so results are reproducible.