Skip to main content

Architectural Diagram

System Architecture Diagram​

Below is a high-level diagram illustrating the system architecture of the AI-Powered Platform for Smart City Issue Detection & Resolution, showing how the various components and technologies are integrated to form a cohesive and scalable platform.

System Architecture Diagram

The diagram represents a modern, scalable, and distributed system designed for real-time data processing, analytics, and smart city applications. The architecture integrates Kubernetes clusters, event-driven communication, and machine learning components to process incidents, analyze data, and deliver insights to both citizens and city operators.

Key Components​

1. User Interaction Layer (Frontend & Mobile Application)​

  • Citizens interact with the system through a mobile application.
  • The app connects to the backend services via RESTful APIs and supports:
    • User authentication
    • Problem reporting
    • Status tracking
  • SendGrid (a third-party email service) is used for notifications, alerts and identity confirmation.

2. Business Logic Layer​

This section is the core of the system, responsible for handling API requests, processing business logic, and managing data.

API Layer (Kubernetes Pod)​

  • Includes controllers such as:
    • Auth Controller
    • User Controller
    • Occurrence Controller
    • Incident Controller
  • Handles authentication, user management, problem management and report submission.

Business Logic Layer (Kubernetes Pod)​

  • Contains the core business logic for the system, including:
  • Auth Service – Manages user authentication and authorization.
  • LLM Producer - Connects to the Kafka broker to produce messages for LLM processing.
  • Occurrence Service – Processes user-reported issues.

3. Databases​

Below are the main databases used in the system, each serving a specific purpose:

Main Storage: Cassandra​

  • Scalability: Cassandra scales horizontally by simply adding new nodes, making it ideal for handling growing data loads.
  • High Availability: It offers robust replication and a decentralized architecture, ensuring continuous service even during node failures.
  • Performance: Optimized for fast read operations, Cassandra efficiently handles high-throughput workloads with low latency.
  • Fault Tolerance: Its distributed, peer-to-peer design makes the system resilient to hardware failures and network issues.
  • Flexible Data Model: Supports a wide range of data structures, allowing adaptability as application requirements evolve.

File Storage: MinIO​

  • S3 API Compatibility: MinIO is fully compatible with the Amazon S3 API, allowing for seamless integration with existing tools and libraries designed for AWS S3.
  • On-Premises and Private Cloud Deployment: It can be deployed on-premises or in private cloud environments, offering complete control over your infrastructure.
  • High Performance: Designed for high-performance object storage, MinIO is well-suited for demanding workloads in edge computing and hybrid cloud environments.
  • Lightweight Alternative: MinIO provides a lightweight and efficient alternative for local or private object storage deployments.
  • Open-Source: Being open-source, MinIO offers flexibility and transparency, benefiting from community contributions and allowing for custom modifications.

5. Event Streaming Layer (Kafka Broker)​

Kafka Broker (Kubernetes Pod)​

  • Uses Kafka as the central internal event-streaming platform.
  • Manages asynchronous data processing.
  • Used to manage many requests to generate images' descriptions and classifications.

6. LLM Processing Layer​

  • The system uses Gemini as the Large Language Model (LLM) for processing user requests and generating responses.
  • The LLM Consumer connects to the Kafka broker to consume messages and produce responses based on user requests.