Skip to main content

Architecture

Overview​

This document outlines the high-level architecture of the GreenWave system, a smart traffic management solution based on the FIWARE ecosystem. The system leverages traffic simulation, real-time context management, historical data persistence, and Artificial Intelligence to optimize traffic flow and provide actionable insights.

Architecture Diagram​

System Architecture

Component Layers​

The architecture is modular, following a layered approach to ensure scalability and separation of concerns.

1. Simulation & Device Layer​

SUMO (Simulation of Urban MObility):

  • Acts as the data source for the system.
  • Simulates real-world traffic scenarios, generating telemetry data such as vehicle position, speed, and CO2 emissions.
  • In a production environment, this component would be replaced or augmented by physical IoT sensors and traffic cameras.

2. Edge & Ingestion Layer​

IoT Agent (Northbound):

  • Serves as the gateway between the data source (SUMO) and the core platform.
  • Standardizes raw data from the simulation into the NGSI-LD format required by the Context Broker.
  • Handles the mapping of device attributes to logical entities.

3. Core Context Management Layer​

Orion-LD Context Broker:

  • The central component of the architecture, functioning as the "brain" of the system.
  • Manages the lifecycle of context information (entities) using the NGSI-LD standard (Linked Data).
  • Implements a Publish/Subscribe mechanism to notify other components (AI, History Service) whenever context data changes.

MongoDB:

  • The backend storage for the Context Broker.
  • Stores only the current state of entities (e.g., the current location of a vehicle). It does not persist historical data.

4. Intelligence Processing Layer​

GreenWave AI:

  • A custom service containing Machine Learning models (Reinforcement Learning agents).
  • Subscribes to specific events from Orion-LD.
  • Processes real-time data to traffic light control commands.
  • Sends analysis results directly to the Admin Dashboard for visualization.

5. Historical Data Layer​

Quantum Leap:

  • A generic enabler specifically designed for time-series data persistence.
  • Receives notifications from Orion-LD and converts the NGSI-LD payload into a format suitable for time-series databases.

CrateDB:

  • A distributed SQL database optimized for massive amounts of IoT and time-series data.
  • Stores the historical records of all traffic metrics, enabling temporal queries (e.g., "Show traffic density over the last 24 hours").

6. Presentation Layer​

Admin Dashboard (React):

  • The control center for system administrators.
  • Fetches historical analytics from CrateDB.
  • Displays real-time insights and recommendations provided by the GreenWave AI engine.

User Map (React):

  • A client-facing application providing traffic visualization.
  • Queries CrateDB to render heatmaps and traffic flows based on historical and recent data.

Data Flow Description​

The system operates on an event-driven architecture. The data flows through the system in the following sequence:

  1. Data Generation: SUMO generates traffic simulation data (vehicle coordinates, speed, acceleration) and sends it to the IoT Agent.
  2. Normalization: The IoT Agent receives the raw data, converts it into NGSI-LD entities, and updates the Orion-LD Context Broker via HTTP requests.
  3. Context Update & Storage: Orion-LD updates its internal state. The latest values are immediately persisted in MongoDB. At this stage, the system knows exactly what is happening right now.
  4. Notification (The Branching Point): Upon updating the context, Orion-LD triggers asynchronous notifications to subscribers:
    • Path A (To History): A notification is sent to Quantum Leap. Quantum Leap formats the data and inserts a new time-stamped record into CrateDB.
    • Path B (To AI): A notification is sent to GreenWave AI. The AI engine analyzes the new state (e.g., detecting a bottleneck) and computes an optimization strategy.
  5. Visualization & Analysis:
    • The Admin Dashboard and User Map execute SQL queries against CrateDB to retrieve historical trends and visualize traffic patterns over time.
    • Simultaneously, the Admin Dashboard receives the processed insights directly from GreenWave AI to display alerts or suggested actions to the operator.