Skip to content

code

Directory actions

More options

Directory actions

More options

Latest commit

 

History

History

code

README.md

Heuristic Algorithm – Code

This directory contains the core code for running heuristic-based experiments, including the Greedy Solution Generator, Genetic Algorithm, and Local Search, as well as exact solution models using CPLEX. Together, these components create, evolve, and evaluate network deployment solutions.

📂 code/ — Core source code

File Overview

  • greedy-solution-generator.ipynb Generates the initial pool of feasible solutions used to seed the genetic and local search algorithms.

    • Produces valid but not necessarily optimal solutions.

    • Run more iterations than the number of required solutions to ensure a diverse pool.

      • Example: For 50 starting solutions, run 60–70 greedy iterations.

  • genetic_algorithm.py Evolves greedy-generated solutions toward higher-quality results using crossover, mutation, and selection.

    • Reads input data from the chosen dataset (see Dataset README).
    • Outputs solutions, logs, and metrics to the results/ directory.
    • Uses shared components from the modules/ folder.

  • local_search.py Evolves greedy-generated solutions toward higher-quality results using crossover, mutation, and selection.

    • Reads input data from the chosen dataset (see Dataset README).
    • Outputs solutions, logs, and metrics to the results/ directory.
    • Uses shared components from the modules/ folder.

  • cplex_shortest_path.py Builds an exact shortest-path model using CPLEX.

    • Dataset is chosen via the dataset variable.
    • Outputs results to results/.

  • cplex_graph.py Implements a full graph-based CPLEX model for exact deployment solutions.

    • Dataset is chosen via the dataset variable.
    • Saves results to results/.

  • modules/ Shared configuration files, dataclasses, and utilities. See the Modules README for details.

  • logs/ Runtime log files. Auto-generated and excluded via .gitignore.

Dataset Selection

All algorithms being the Greedy, Genetic, Local Search, and CPLEX models require data from the dataset/ directory.

To select a dataset, edit the dataset variable in the code:

# Choose between: "small-dataset" or "full-dataset"
dataset = "small-dataset"

Available datasets:

  • small-dataset – Lightweight runs for debugging and fast testing.
  • full-dataset – Full-scale runs for performance evaluation.

Notes:

  • Use the same dataset consistently across scripts for reproducibility.
  • Datasets can be customised or extended by following the structure and instructions in the Dataset README.

Configuration

All experiment parameters are defined in modules/config.py, including:

  • Cost values (trenching, fibre, node costs)
  • Coverage radii
  • Bandwidth requirements
  • Tunable constraints

Modify this file to configure experiments globally across all algorithms and models.

Results

Outputs are saved in the results/ directory:

  • Greedy Algorithm: Final solutions only.

  • Genetic Algorithm: Solutions, logs per iteration, and used-segment JSONs.

  • Local Search: Solutions, logs per iteration, and used-segment JSONs.

  • CPLEX Models:

Usage Guide

  1. Run greedy-solution-generator.ipynb to generate an initial solution pool.

    • Ensure iterations > required solutions (e.g., 60–70 runs for 50 needed solutions).

  2. Choose a refinement method:

  3. (Optional) For exact optimisation, run:

  4. Inspect output files in results/.

Workflow Summary

dataset → greedy → genetic → local search → results
dataset → CPLEX (shortest path or graph) → results

This workflow combines heuristic (greedy, genetic, local search) and exact (CPLEX) approaches to network design, ensuring all experiments remain reproducible, configurable, and extensible.