MetaTraffic AI | Comparative Optimization

Multi-Algorithm Traffic Signal Timing (GA, PSO, SA)

Course: Evolutionary Computing
University: Dawood University of Engineering & Technology
Semester: Fall 2022

---

📋 Table of Contents

• Problem Statement
• Methodology
• Tech Stack
• Project Structure
• How to Run
• Optimization Algorithms
• Fitness Function
• Comparison
• Team Members

---

🎯 Problem Statement

Urban traffic congestion is a critical problem in modern cities. Traditional traffic signal timing systems use fixed-cycle approaches that cannot adapt to varying traffic conditions. This project addresses the problem of optimizing traffic signal green-light durations across an N×N grid of intersections using comparative optimization techniques.

Objective: Find signal timing plans that:

• ✅ Maximize vehicle throughput
• ✅ Minimize average waiting time
• ✅ Reduce queue lengths
• ✅ Prevent gridlock conditions

---

🧭 Methodology

We employ a comparative optimization pipeline that evaluates GA, PSO, and SA under identical simulation conditions:

1. Define the simulation scenario (grid size, steps, spawn rate)
2. Select an optimizer (GA, PSO, SA)
3. Search timing plans using the chosen algorithm
4. Evaluate each plan with the traffic simulator
5. Compare results against baselines
6. Report metrics, charts, and verdicts

Baselines:

• Fixed Timing: All signals use equal green durations
• Random Timing: Each signal uses random green durations

Optimizers:

• GA Optimized: Best plan evolved by the Genetic Algorithm
• PSO Optimized: Best plan found by swarm search
• SA Optimized: Best plan found by annealing

---

🛠 Tech Stack

Layer	Technology
Algorithm	Python (NumPy)
Backend API	FastAPI + Uvicorn
Frontend	React 18 + Vite
Charts	Chart.js + react-chartjs-2
Styling	Vanilla CSS (Glassmorphism + Dark Theme)
Visualization	HTML5 Canvas
Routing	React Router v6

---

📁 Project Structure

Smart-Traffic-Signal-Timing/
├── backend/
│   ├── main.py                    # FastAPI entry point
│   ├── requirements.txt           # Python dependencies
│   ├── __init__.py
│   ├── ga/                        # Genetic Algorithm
│   │   ├── chromosome.py          # Chromosome representation
│   │   ├── fitness.py             # Fitness function
│   │   ├── selection.py           # Tournament selection
│   │   ├── crossover.py           # Crossover operators
│   │   ├── mutation.py            # Mutation operators
│   │   ├── elitism.py             # Elitism preservation
│   │   └── optimizer.py           # Main GA loop
│   ├── simulation/                # Traffic simulation
│   │   ├── traffic_grid.py        # N×N intersection grid
│   │   ├── vehicle.py             # Vehicle agent
│   │   └── simulator.py           # Simulation engine
│   ├── baselines/                 # Baseline strategies
│   │   ├── fixed_timing.py        # Fixed equal-duration
│   │   └── random_timing.py       # Random-duration
│   ├── api/                       # REST API
│   │   ├── routes.py              # API endpoints
│   │   └── schemas.py             # Pydantic models
│   └── utils/
│       └── export.py              # Export & verdict generation
├── frontend/
│   ├── index.html
│   ├── package.json
│   ├── vite.config.js
│   └── src/
│       ├── main.jsx               # React entry
│       ├── App.jsx                # Router + layout
│       ├── index.css              # Design system
│       ├── api/
│       │   └── client.js          # API client
│       ├── components/            # Reusable components
│       │   ├── Navbar.jsx         # Navigation bar
│       │   ├── Footer.jsx         # Footer
│       │   ├── TrafficGrid.jsx    # Canvas traffic grid
│       │   ├── MetricsCard.jsx    # Metric display card
│       │   ├── ChartComponent.jsx # Chart wrapper
│       │   ├── ComparisonTable.jsx# Comparison table
│       │   └── TeamCard.jsx       # Team member card
│       └── pages/                 # Application pages
│           ├── Home.jsx           # Landing page
│           ├── Simulator.jsx      # Traffic simulator
│           ├── Optimizer.jsx      # GA optimizer
│           ├── Dashboard.jsx      # Results dashboard
│           ├── Comparison.jsx     # Baseline comparison
│           ├── Methodology.jsx    # Research report
│           └── Team.jsx           # Team members
└── README.md

---

🚀 How to Run

Prerequisites

• Python 3.9+
• Node.js 18+
• npm

Backend

cd backend
pip install -r requirements.txt
cd ..
python -m backend.main

The API will start at http://localhost:8000
API docs at http://localhost:8000/docs

Frontend

cd frontend
npm install
npm run dev

The UI will start at http://localhost:5173

Both Together

Open two terminals:

# Terminal 1 — Backend
python -m backend.main

# Terminal 2 — Frontend
cd frontend && npm run dev

---

⚙️ Optimization Algorithms

Timing Plan Representation

Each timing plan is a real-valued array of length 2 × N × N, encoding green durations for both phases (NS and EW) at every intersection.

Value structure: [NS₍₀,₀₎, EW₍₀,₀₎, NS₍₀,₁₎, EW₍₀,₁₎, ..., NS₍ₙ,ₙ₎, EW₍ₙ,ₙ₎]
Value bounds: 10 ≤ duration ≤ 60 seconds

Algorithm Summaries

• GA: Population-based evolutionary search using selection, crossover, mutation, and elitism.
• PSO: Swarm-based optimization with inertia, cognitive, and social updates.
• SA: Trajectory-based stochastic search with temperature-controlled acceptance.

---

📐 Fitness Function

Fitness = w₁ × Throughput − w₂ × AvgWaitingTime − w₃ × AvgQueueLength − w₄ × GridlockPenalty

Weight	Value	Objective
w₁ (Throughput)	2.0	Maximize
w₂ (Waiting Time)	1.0	Minimize
w₃ (Queue Length)	0.5	Minimize
w₄ (Gridlock)	3.0	Minimize

Higher fitness = Better timing plan

---

⚖️ Comparison

The project compares baselines and optimizers under identical simulation conditions:

1. Fixed Timing — All signals use 30s green per phase
2. Random Timing — Random durations ∈ [10, 60]s
3. GA Optimized — Best evolved timing plan
4. PSO Optimized — Best swarm-discovered timing plan
5. SA Optimized — Best annealed timing plan

Metrics compared:

• Throughput (vehicles completed)
• Average waiting time
• Average queue length
• Gridlock penalty
• Completion rate

A dynamic verdict is generated based on actual results.

---

👥 Team Members

Name	Roll Number
Muhammad Fasih	22F-BSCS-19
Syed Masood Hussain	22F-BSCS-26
Syed Tehmeed Jafar	22F-BSCS-11

---

📄 License

This project is developed for academic purposes as part of the Evolutionary Computing course at Dawood University of Engineering & Technology.

---

Built with Python + React + Comparative Optimization