How chimms1/lte-wifi-simulator works
Project type and purpose
• This is a Python-based network simulator for modeling LTE and WiFi coexistence scenarios in unlicensed spectrum.
• It is not a client-server or web application, but a computational engine that runs experiments, applies Q-learning, and outputs performance metrics and plots.
File structure
• Top-level scripts (compute7.py, rouhg.py, run500.py/sh) drive batch runs or custom experiments.
• “Simulator/” contains the core simulation engine:
• entities/ – definitions for BaseStation and UserEquipment
• Qlearning/ – implements the reinforcement learning logic
• running/ – configuration (ConstantParams.py), service methods (ServiceClass.py), and verbose/plot controls (Print.py)
• main-latest-all.py (and variants) – entry points that tie everything together.
Extract additional insights from the README
• main-latest-all.py is the primary driver; it loads parameters, initializes entities, loops over time/iterations, invokes service methods and Q-learning updates, and triggers logging/plotting.
• Q-learning module implements states, actions, rewards, and Q-table operations.
• Running package centralizes user-set parameters, printing flags, and service utilities.
a. Main components
1. Entry-point Scripts (main-latest-all.py, run scripts)
2. Configuration Layer (ConstantParams.py)
3. Simulation Core
– ServiceClass (orchestrates simulation steps)
– Entities (BaseStation, UserEquipment)
4. Learning Engine (Qlearning module)
5. Output/Visualization (Print.py, matplotlib, seaborn)
6. Batch Runner Scripts (compute7.py, run500.py)
b. Relationships and interactions
• Entry-point reads ConstantParams → Instantiates ServiceClass → Creates Entities
• ServiceClass calls Entities to simulate traffic/SINR → Passes state & reward to Learning Engine → Learning Engine updates Q-table
• ServiceClass uses Print module to log & plot intermediate/final results
• Batch Runner Scripts invoke multiple runs of the Entry-point with different seeds or parameter sets
c. Architectural patterns and design principles
• Modular layering: configuration, core logic, entities, learning, output
• Separation of concerns: Entities purely model network objects; ServiceClass handles orchestration; Qlearning is plugin-style algorithm; Print is orthogonal logging/visualization
d. Key technologies and libraries
• Python 3.9+
• numpy, pandas (data handling)
• matplotlib, seaborn (plots)
• tqdm, openpyxl (optional reporting)