5G Edge Cloud Application

WINLAB Summer Internship 2024

Advisors: Anthony Magnan, Ivan Seskar, Jennifer Shane

Group Members: Vineal Sunkara, Matt Arigo, Steven Nguyen, Ji Wu, Damon Lin

Project Objective

Our project aims to create a navigation system that utilizes 5G edge cloud computing to reduce the likelihood of vehicular accidents. We will build this system by utilizing synthetic data we generated off of SUMO and CARLA. Moreover, a message broker called Mosquitto handled the data communication between this software application and the algorithm on the server, and the messaging protocol behind it was MQTT. To jumpstart the CI/CD, or the continuous integration and continuous deployment process, we containerized all these components with Docker.

Week 1

​Week 1 Presentation

Summary

• Looked into key terms such as MQTT, Multi-Access Edge Computing (MEC), Enhanced Vehicle-to-Everything Communication (eV2X Communication), and Virtual Roadside Units (vRSU)
• Installed and tested Mosquitto, an MQTT broker on Linux, through these commands:

  sudo apt-add-repository ppa:mosquitto-dev/mosquitto-ppa
  sudo apt-get update
  sudo apt-get install mosquitto
  sudo apt-get install mosquitto-clients
  sudo apt clean
  

Week 2

​Week 2 Presentation

Summary

• Met with our advisors to discuss and define the project's direction
  • Conceptualize and develop a fundamental software pipeline to communicate data between vehicles and the edge cloud
  • Set up a simple scenario with two vehicles stopping before a head-on collision
• Took a look at simulation software CARLA (Car Learning to Act) and SUMO (Simulation for Urban Mobility)
• Installed and tested Paho, another MQTT broker on Linux, as a substitute for Mosquitto
  • Paho is a Python-based MQTT broker, which is much more compatible with CARLA and SUMO

Week 3

​Week 3 Presentation

Summary

• Began simulating scenarios on both CARLA and SUMO
• Figured out first-person interface in CARLA
• Set up Gitlab repository for message encoding to transcribe data received from CARLA and SUMO
• Start receiving telemetry data from SUMO

Week 4

​Week 4 Presentation

Summary

• Began working on our base scenario
• Worked on an algorithm that calculates when braking is necessary
• Started integration of our scenario in CARLA with our database and algorithm
• Started learning about Docker for later project utilization

Week 5

​Week 5 Presentation

Summary

• Made base scenario for SUMO and CARLA
• Refined algorithm by adding a filter system before running the crash detector.
• Finished Postgres database setup.

Week 6

​Week 6 Presentation

Summary

• Began research on traffic control.
• Set up test cases for encoding and decoding.
• Shifted away from Docker for CARLA and moved to node images.

Week 7

​Week 7 Presentation

Summary

• Finished algorithm and conducting unit tests.
• Set up Raspberry Pi and OBDII Emulator in order to simulate real-time vehicular diagnostics.
• Prepared for the server to move from intersection to outdoor nodes.
• Prepared for project demo that is taking place in week 8.

Week 8

​Week 8 Presentation(Short)
​Week 8 Presentation(Long)

Summary

• Presented demo to Verizon.
• Simulated data from OBDII to get real vehicle data.
• Did a lot of debugging and integration.

Week 9

​Week 9 Presentation

Summary

• Utilized Manim to visualize and troubleshoot algorithm and MQTT pipeline
  • Discovered (through the GIF below) that the current implementation of the algorithm does not detect collisions in a particular scenario
  • Determined that vehicle boxes (like those in games) are more accurate than rays
• Bonus Work: OBDII module
  • Successfully sent telemetry data from the OBDII through the broker to a subscriber
  • Attempting to configure latency to synchronize the publisher and the subscriber

Week 10 + Open House

​Open House Presentation

Summary

• Continued to integrate all parts of the project
  • MQTT Pipeline
  • CARLA
  • Edge Algorithm
  • OBD-II Emulator
• Finalized final deliverables (poster and presentation)
• Presented the deliverables to peers, parents, and advisors

Resources

• ​CARLA Github Link
  
• ​SUMO Github Link