Reactive Coverage of a Large Area with a Satellite Constellation 🛰️

🌍 Context and objectives:

• A constellation of 32 low Earth orbit satellites is used to monitor Areas of Interest (AOIs) on a large scale (countries, regions).
• Applications include: forest fire monitoring, flood observation, illicit activity detection, conflicts, and more.

The user aims to:

• 🗺️ Collect data in response to unexpected events within their area of interest.
• 🎯 Reserve satellite passages in advance to avoid conflicts with other users.
• ⏱️ For each sliding window of reactivity duration (2, 4, or 12 hours), select at least one satellite passage covering one of the AOI's accesses.

📥 Input data:

The input data is stored in the data folder:

1. Areas of Interest (AoIs):

   • Geographic characteristics (coordinates, visibility windows).
   • List of possible accesses for each satellite (date and duration of passage).

2. Reactivity objectives:

   • Ensure deadlines (12h, 4h, 2h) are respected to cover at least one access within a given window.
   • Three .json files in data, each corresponding to one of the reactivity constraints.

3. Observation requests:

   • A request is fulfilled if at least one access is covered during a satellite passage (minimum duration: 30 seconds).

🚀 Optimization problem:

• Decisions to make:

  • Select which satellite passages to reserve.
  • Define the start and end intervals for each reservation.

• Constraints:

  • Only one reservation per passage.

• Criteria to optimize:

  • Minimize costs for the user (fixed fees + cost per unit of reserved time).
  • Maximize satisfaction of observation requests.

🛠️ Project approach:

1. Step 1: Familiarization:

   • Analyze the provided data.
   • Define metrics to evaluate solutions (e.g., total cost, request satisfaction).

2. Step 2: Optimization with full passages:

   • Test a MILP approach using DOCPLEX.
   • Explore heuristic methods (greedy algorithms, metaheuristics).

3. Step 3: Optimization with partial passages:

   • Select only specific time intervals within passages to reduce costs for the user.

4. Step 4: Multi-criteria problem:

   • Perform Pareto analysis (cost vs satisfaction).
   • Generate optimal solutions based on multiple criteria.

This project provides the opportunity to explore the entire optimization pipeline, from data formalization to solution visualization.

---

📋 Requirements:

To run the provided Python code, you need:

• Python 3.8+
• Required libraries (install using pip install -r requirements.txt):
  • matplotlib
  • numpy
  • docplex
    ⚠️ DOcplex (IBM Decision Optimization CPLEX Modeling for Python) requires a license, as it is based on the use of IBM CPLEX Optimizer, a commercial product.
  • cartopy

📖 Learn More:

For more details, check the documentation.

ℹ️ Additional Notes:

• The final project report is in French 🇫🇷.
• This README serves as an overview of the work and deliverables in English.