| # AdaptiSwarm-Edge |
|
|
| A Convergence-Accelerated Hybrid ACO-PSO Framework for Priority-Aware Dynamic Load Balancing in Heterogeneous Multi-Tier Edge Computing Networks. |
|
|
| ## Files |
| - `sim.py` — Discrete-event simulation core (3-tier heterogeneous edge, Poisson arrivals, dynamic churn) |
| - `schedulers.py` — All schedulers: Round Robin, Least Connection, ACO, PSO, WOA, AdaptiSwarm-Edge, plus ablation variants |
| - `runner.py` — Episode runner with persistent pheromone state |
| - `convergence.py` — Fixed-instance convergence harness with shared fitness target |
| - `experiments.py` — Parallel experiment driver (30 runs per config, Wilcoxon tests, convergence figure) |
| - `make_tables.py` — LaTeX table + numeric macro generator from results.json |
|
|
| ## Reproduction |
| ```bash |
| pip install numpy scipy matplotlib |
| cd adaptiswarm |
| python experiments.py 30 |
| python make_tables.py |
| # Paper compiles via: tectonic paper.tex (in paper/) |
| ``` |
|
|
| ## Results |
| All numbers in the accompanying IEEE paper are generated by `make_tables.py` from `results.json` — no hand-entered results. |
|
|
