Skybed

A Software Testbed for UAV Edge Computing Applications

SkyBed is an open-source software testbed designed to facilitate the development and testing of edge computing applications for Unmanned Aerial Vehicles (UAVs). It provides a scalable and flexible framework for simulating UAV movements, cellular network interactions, and edge-computing scenarios, enabling researchers to evaluate applications, such as collision avoidance, AI image processing, and more.

Features

• Virtual UAV Simulation: Simulate UAV movement with customizable paths, speeds, and behaviors.
• Cellular Network Emulation: Model latency, throughput, and packet loss based on UAV proximity to virtual cell towers.
• Edge Computing Scenarios: Test edge-based applications with real-time or precomputed network parameters.
• Collision Detection: Detect UAV collisions efficiently using a ball-tree-based algorithm.
• Scalability: Simulate hundreds of UAVs simultaneously using containerized vUAVs.
• Repeatability: Create repeatable experiments with configurable scenarios.
• Visualization: Monitor UAVs and cell towers on an interactive, browser-based map.

Getting started

Prerequisites

• Operating System: Linux (WSL2 may work for Windows users)
• Dependencies:
  • Docker
  • Python 3.11+
  • Poetry
• For 6G NeXt Integration:
  • Set up the following:
    • 6gn-functions
    • 6gn-ingester
    • tinyFaas (requires modifications to support build tools)

Installation

1. Install Python Dependencies:

   poetry install

2. Enable Network Configuration without Superuser Privileges:

   sudo setcap cap_net_admin+ep /sbin/tc
   sudo setcap cap_net_raw,cap_net_admin+ep /bin/ip

3. Build Docker Images (this may take a while):

   docker build -f ns3_lena.Dockerfile -t ns3_lena .
   docker build -f uav.Dockerfile -t uav .

Usage

Setting Up for Collision Avoidance with 6G NeXt

Follow the instructions in the 6gn-functions repository to configure the anti-collision system.

Running the Testbed

When running SkyBed with the following command:

skybed --help

Available options will be displayed:

 Usage: skybed [OPTIONS] [SCENARIO_FILE] [TEST_SYSTEM_IP]                                                                                                                                                                                  
                                                                                                                                                                                                                                           
╭─ Arguments ─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╮
│   scenario_file       [SCENARIO_FILE]   [default: schoenhagen_near_collision]                                                                                                                                                           │
│   test_system_ip      [TEST_SYSTEM_IP]  [default: localhost]                                                                                                                                                                            │
╰─────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────────╯

Viewing the Visualization

Open your browser and navigate to:

http://localhost:8050

Example Scenario

The schoenhagen_many_drones scenario simulates 200 UAVs flying near Schönhagen Airport to test collision detection and network performance under heavy load.

To run this scenario:

skybed schoenhagen_many_drones

Additional scenarios are located in the skybed/scenarios directory. To create a custom scenario:

1. Copy an existing file from skybed/scenarios/.
2. Edit the UAV paths, speeds, or network configurations to match your requirements.

Acknowledgments

SkyBed was developed as part of a Master's thesis at the Technische Universität Berlin. The thesis can be read here.