Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Time-Dependent Network Topology Optimization for LEO Satellite Constellations

Dara Ron, Faisal Ahmed Yusufzai, Sebastian Kwakye, Satyaki Roy, Nishanth Sastry, Vijay K. Shah

TL;DR

This paper tackles the challenge of designing time-varying topologies for LEO satellite constellations by introducing the Dynamic Time-Expanded Graph (DTEG) and the DoTD algorithm. It formulates a multi-objective optimization that maximizes capacity while minimizing latency and link churn, then transforms it into a time-dependent scoring problem that each satellite solves over a future horizon. The approach is implemented with a DTEG-based topology design, complemented by an OSPF routing layer, and validated against Greedy and +Grid baselines using a Starlink-like constellation and TLE-based SpaceNet emulation; results show substantial improvements in capacity, latency, hop count, and stability. This work provides a scalable, time-aware method to sustain near-optimal performance in dynamic space networks, with practical impact for global broadband constellations and adaptive ground-station routing.

Abstract

Today's Low Earth Orbit (LEO) satellite networks, exemplified by SpaceX's Starlink, play a crucial role in delivering global internet access to millions of users. However, managing the dynamic and expansive nature of these networks poses significant challenges in designing optimal satellite topologies over time. In this paper, we introduce the \underline{D}ynamic Time-Expanded Graph (DTEG)-based \underline{O}ptimal \underline{T}opology \underline{D}esign (DoTD) algorithm to tackle these challenges effectively. We first formulate a novel space network topology optimization problem encompassing a multi-objective function -- maximize network capacity, minimize latency, and mitigate link churn -- under key inter-satellite link constraints. Our proposed approach addresses this optimization problem by transforming the objective functions and constraints into a time-dependent scoring function. This empowers each LEO satellite to assess potential connections based on their dynamic performance scores, ensuring robust network performance over time without scalability issues. Additionally, we provide proof of the score function's boundary to prove that it will not approach infinity, thus allowing each satellite to consistently evaluate others over time. For evaluation purposes, we utilize a realistic Mininet-based LEO satellite emulation tool that leverages Starlink's Two-Line Element (TLE) data. Comparative evaluation against two baseline methods -- Greedy and $+$Grid, demonstrates the superior performance of our algorithm in optimizing network efficiency and resilience.

Time-Dependent Network Topology Optimization for LEO Satellite Constellations

TL;DR

This paper tackles the challenge of designing time-varying topologies for LEO satellite constellations by introducing the Dynamic Time-Expanded Graph (DTEG) and the DoTD algorithm. It formulates a multi-objective optimization that maximizes capacity while minimizing latency and link churn, then transforms it into a time-dependent scoring problem that each satellite solves over a future horizon. The approach is implemented with a DTEG-based topology design, complemented by an OSPF routing layer, and validated against Greedy and +Grid baselines using a Starlink-like constellation and TLE-based SpaceNet emulation; results show substantial improvements in capacity, latency, hop count, and stability. This work provides a scalable, time-aware method to sustain near-optimal performance in dynamic space networks, with practical impact for global broadband constellations and adaptive ground-station routing.

Abstract

Today's Low Earth Orbit (LEO) satellite networks, exemplified by SpaceX's Starlink, play a crucial role in delivering global internet access to millions of users. However, managing the dynamic and expansive nature of these networks poses significant challenges in designing optimal satellite topologies over time. In this paper, we introduce the \underline{D}ynamic Time-Expanded Graph (DTEG)-based \underline{O}ptimal \underline{T}opology \underline{D}esign (DoTD) algorithm to tackle these challenges effectively. We first formulate a novel space network topology optimization problem encompassing a multi-objective function -- maximize network capacity, minimize latency, and mitigate link churn -- under key inter-satellite link constraints. Our proposed approach addresses this optimization problem by transforming the objective functions and constraints into a time-dependent scoring function. This empowers each LEO satellite to assess potential connections based on their dynamic performance scores, ensuring robust network performance over time without scalability issues. Additionally, we provide proof of the score function's boundary to prove that it will not approach infinity, thus allowing each satellite to consistently evaluate others over time. For evaluation purposes, we utilize a realistic Mininet-based LEO satellite emulation tool that leverages Starlink's Two-Line Element (TLE) data. Comparative evaluation against two baseline methods -- Greedy and Grid, demonstrates the superior performance of our algorithm in optimizing network efficiency and resilience.
Paper Structure (14 sections, 2 theorems, 17 equations, 6 figures, 2 tables, 1 algorithm)

This paper contains 14 sections, 2 theorems, 17 equations, 6 figures, 2 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Space Network Model
  3. Network Scenario
  4. Network Model for GS-to-LEO Satellites
  5. Network Model for Inter-Satellite Links
  6. Optimization Problem Formulation
  7. DTEG-Based Optimal Topology Design (DoTD) Algorithm
  8. DTEG Construction
  9. Algorithm Description
  10. Experimental Results
  11. Experimental Setup
  12. Baseline Methods
  13. Experimental Results
  14. Conclusion and Future work

Key Result

Lemma 3.1

The time-dependent maximum value $g_{\text{Max},t}$ defined in Eq. eq12 is the achievable maximum capacity, latency, and link churn, denoted by $g$, up to time $t$.

Figures (6)

  • Figure 1: SpaceX's Starlink LEO satellite constellation
  • Figure 2: DTEG graph representation.
  • Figure 3: SpaceX's Starlink LEO satellites constellation
  • Figure 4: Network maps created by DTEG, Greedy, and $+$Grid: Packets traveling from New York to San Francisco.
  • Figure 5: (a) Hop counts from the GS source to the destination, (b) Latency evaluation, (c) Average achievable capacity of each link, under five different scenarios in TABLE \ref{['Tab2']}.
  • ...and 1 more figures

Theorems & Definitions (4)

  • Lemma 3.1
  • proof
  • Lemma 3.2
  • proof