Analysis of a Delay-Tolerant Data Harvest Architecture Leveraging Low Earth Orbit Satellite Networks

TL;DR

This work develops a Cox point process framework to model moving LEO satellite harvesters that collect delay-tolerant data from ground devices without ground infrastructure. It jointly captures orbit distribution and satellite motion, enabling closed-form analysis of harvest-time fraction, data harvested per pass, harvesting capacity, and delay distribution as functions of the mean number of orbits $λ$ and the mean satellites per orbit $μ$, under Nakagami fading. The results yield design insights such as prioritizing more orbital planes to maximize availability and differentiating conditions under which capacity and delay performance improve, providing a practical framework for designing and optimizing delay-tolerant LEO data-harvesting networks. The framework supports analysis of deployment tradeoffs, enables performance-oriented constellation design, and offers a foundation for comparing Cox-based constellations with traditional polar configurations in real-world remote-sensing and IoT applications.

Abstract

Reaching all regions of Earth, low Earth orbit (LEO) satellites can harvest delay-tolerant data from remotely located users on Earth without ground infrastructure. This work aims to assess a data harvest network architecture where users generate data and LEO satellites harvest data from users when passing by. By developing a novel stochastic geometry Cox point process model that simultaneously generates orbits and the motion of LEO satellite harvesters on them, we analyze key performance indices of such a network by deriving the following: (i) the average fraction of time that the typical user is served by LEO satellite harvesters, (ii) the average amount of data uploaded per each satellite pass, (iii) the maximum harvesting capacity of the proposed network model, and (iv) the delay distribution in the proposed network. These key metrics are given as functions of key network variables such as $λ$ the mean number of orbits and $μ$ the mean number of satellites per orbit. Providing rich comprehensive analytical results and practical interpretations of these results, this work assesses the potential of the delay-tolerant use of LEO satellites and also serves as a versatile framework to analyze, design, and optimize delay-tolerant LEO satellite networks.

Figures (15)

• Figure 1: longitude $\theta$, inclination $\phi$, and satellite's argument angle $\omega$.
• Figure 2: Illustration of the proposed network with $\lambda=40$ and $\mu=70$. There are $2800$ satellites on average.
• Figure 3: Illustration of the proposed network with $\lambda=15$ and $\mu=80$. There are $1200$ satellites on average.
• Figure 4: $\gamma$ is the maximum distance from the ground device to LEO satellite within which the data harvesting occurs.
• Figure 6: The expected number of harvesters in the proposed network.

Theorems & Definitions (11)

• Lemma 1
• Lemma 2
• Theorem 1
• Example 1: Optimizing number of orbital plane
• Theorem 2
• Remark 1
• Theorem 3
• Remark 2
• Remark 3: Difference of data harvest and harvesting capacity
• Corollary 1