Peer-to-Peer Sensor System with Logical Clocks

This project implements a decentralized distributed system for monitoring sensor readings over time. It consists of multiple peer-to-peer sensor nodes communicating via UDP, coordinated by Kafka. The system demonstrates distributed synchronization using scalar and vector logical clocks, while handling packet loss and retransmission.

---

Features

• Peer-to-peer network of sensor nodes
• Central coordination of nodes with Kafka (Start / Stop signals, registration)
• UDP-based communication with simulated packet loss and retransmission
• Emulation of physical clocks for timestamping messages
• Exchange of sensor readings with both scalar and vector clocks
• Periodic sorting of readings by both timestamps
• Computation of average NO₂ values in a 5-second sliding window

---

Architecture

Kafka Control Node

• Publishes control messages via Kafka topics
• Manages node registration through the topics

Sensor Nodes

• Emulate sensor readings from a CSV file
• Exchange readings with all other nodes via UDP
• Attach scalar and vector timestamps to each reading
• Send acknowledgments and handle retransmissions on packet loss
• Sort and display readings in order using timestamps
• Compute average values for every 5-second interval

---

Installation

1. Prerequisites:
   • Java 21
   • Docker & Docker Compose
2. Clone and build the project:

   git clone https://github.com/Vjekoslav-Radosevic/Peer-to-Peer-Sensor-System-with-Logical-Clocks.git
   cd Peer-to-Peer-Sensor-System-with-Logical-Clocks
   gradlew build

3. Run Docker Compose file to start Kafka broker with Raft:

   docker compose up

---

Usage

1. Start the sensor node with:

   gradlew runSensorNode --args="<sensor_port>"

2. Start the control node with:

   gradlew runControlNode