Skip to content

Latest commit

 

History

History
93 lines (76 loc) · 4.38 KB

File metadata and controls

93 lines (76 loc) · 4.38 KB

A Dynamic Load Balancing Routing Algorithm for GEO/LEO Hybrid Satellite Networks

This algorithm is a dynamic load balancing routing algorithm for GEO/LEO hybrid satellite networks. It is based on congestion prediction, and improve QoS(quality of service) by classifying network traffic. And this algorithm is implemented by Hypatia framework.
Keyworks: :GEO/LEO hybrid network, congestion prediction, load balancing, traffic classification, traffic detour

first0 algorithm utilization0 first1

Algorithm overview:
Congestion prediction: This algorithm declares the network congestion area as the congestion area, and the low-orbit satellite moving to the area can predict the congestion in advance.
Traffic detour: This algorithm divides network traffic into three categories according to the sensitivity to delay, and executes different detour strategies for different traffic when the network is congested.

Details about this algorithm can be found here.
The paper proposing this algorithm is:

{
    title:      Load Balancing and QoS Provisioning Based on Congestion Prediction for GEO/LEO Hybrid Satellite Networks
    author:     Hiroki Nishiyama, Daigo Kudoh, Nei Kato and Naoto Kadowaki
    booktitle:  IEEE
    year:       2011
}

(In addition, I am not the author of this algorithm, I just implemented the algorithm in code and modified some details)

Getting started

  1. System setup:

    • Python version 3.8+
    • Recent Linux operating system (e.g., Ubuntu 20+)
  2. Install Hypatia dependencies:

    bash hypatia_install_dependencies.sh
    
  3. Build ns3 modules:

    bash hypatia_build.sh
    
  4. Generate GEO/LEO satellite network data:

    bash generate_satellite_network_data.sh
    

    (On machine with 4 cores and 4G memory, it takes about 1 hour)

  5. Run ns3 to simulate:

    bash run_simulate.sh
    

    All simulation results and log are in paper_routing/ns3_experiments/traffic_matrix/runs
    Algorithm performance picture are in paper_routing/ns3_experiments/traffic_matrix/figures
    (On machine with 4 cores and 4G memory, it takes about 5~7 hour)

  6. (optional) Cesium Visualization:

    Before generate Cesium visualization, you need:

    1. Obtain a Cesium access token at https://cesium.com/
    2. Edit paper_routing/ns3_experiments/traffic_matrix/satviz/static_html/top.html, and insert your Cesium access token at line 10:
      Cesium.Ion.defaultAccessToken = '<CESIUM_ACCESS_TOKEN>';

    After that:

    bash generate_cesium_visualization.sh
    

    It will generate .html file in paper_routing/ns3_experiments/traffic_matrix/figures. You should open it with a linux browser(e.g. Firefox)

  7. Change Parameter
    If you want to change the parameter of this algorithm, you can read here

Visualizations

Algorithm Performance

  • Packet Drop Rate
    Packet Drop Rate
  • Throughput
    Throughput
  • End-to-end delay
    End-to-end delay

Cesium Visualizations

  • GEO/LEO satellite network structure
    satellite networke0 satellite networke1

  • this algorithm isl utilization
    algorithm utilization0 algorithm utilization1

  • Dijkstra shortest path(DSP) isl utilization
    DSP0 DSP1