Echo 🗣️

A decentralized, location-aware messaging application built with Kotlin Multiplatform using Aggregate Computing principles. Featuring GPS-based gossip algorithms for proximity communication.

🌟 Features

• 📍 GPS-Based Messaging: Send messages to devices within a specific geographic radius
• 🔄 Decentralized Gossip Algorithm: Self-stabilizing message propagation without central control
• 📡 Real-Time Location Tracking: Continuous GPS updates and neighbor discovery
• ⏱️ Configurable Parameters: Set message lifetime and propagation distance
• 🌐 Cross-Platform: Runs on Android and iOS using Kotlin Multiplatform
• 📶 MQTT Communication: Reliable message transport via MQTT protocol
• 🛠️ Built with Collektive: Leverages the Collektive framework for aggregate computing

🏗️ Architecture

System Design

Echo implements a hybrid decentralized architecture:

• Application Layer (L7): Fully decentralized gossip algorithm

  • Self-stabilizing convergence
  • Distributed decision-making
  • No central coordinator

• Transport Layer (L4): MQTT-based communication

  • Public broker for message routing
  • QoS guarantees for reliability
  • WebSocket support for mobile networks

Key Components

┌─────────────────────────────────────────┐
│         Application Layer               │
│  (Decentralized Gossip Algorithm)       │
├─────────────────────────────────────────┤
│           MQTT Mailbox                  │
│      (Message Transport Layer)          │
├─────────────────────────────────────────┤
│        Location Services                │
│     (GPS Tracking & Distance)           │
├─────────────────────────────────────────┤
│      Collektive Framework               │
│   (Aggregate Computing Engine)          │
└─────────────────────────────────────────┘

🧮 Gossip Algorithm & Aggregate Computing

The app implements a GPS-aware gradient gossip algorithm built on Aggregate Computing principles using the Collektive framework.

Decentralization

• No central coordinator: Each node operates autonomously
• Distributed state: Information spreads peer-to-peer through aggregate constructs
• Local decisions: Nodes use only neighbor information via Field abstractions
• Global-to-local compilation: Aggregate programs automatically map to decentralized execution

Self-Stabilization

• Convergence guarantee: Reaches stable state through aggregate computation rounds
• Fault tolerance: Recovers from node failures via field-based resilience
• Eventual consistency: System converges despite asynchronous execution
• Dynamic adaptation: Adjusts to changing network conditions and node availability

Spatial Awareness

• Real distances: Uses Haversine formula for GPS calculations between neighbors
• Sender-controlled radius: Message creator sets propagation distance (embedded in message)