Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Goal-oriented vessel detection with distributed computing in a LEO satellite constellation

Antonio M. Mercado-Martínez, Beatriz Soret, Antonio Jurado-Navas

TL;DR

The paper addresses near real-time vessel surveillance by leveraging distributed edge computing within a LEO satellite constellation. It combines YOLOv8-based vessel detection on the VHRShips dataset with a fragmented, parallel processing pipeline across neighboring satellites, and quantifies timeliness using AoI metrics to guide constellation sizing. The study demonstrates that a 20×20 walker-delta topology with five processing satellites can maintain peak AoI below 60 seconds while achieving a data reduction of over 99.996% and effectively full-area coverage, illustrating the practical viability of edge-enabled EO missions. This work offers a concrete design methodology for timely sensing in space and points to future extensions, including ground-queried closed loops and atmospheric effects.

Abstract

Earth Observation (EO) has traditionally involved the transmission of a large volume of raw data to map the Earth surface. This results in congestion to the satellite network and delays in the availability of the results, invalidating the approach for timing-sensitive applications. Instead, the computation resources at the satellites can be used as an edge layer for compressing the data and/or doing inferences. In this paper, we investigate satellite edge computing for vessel detection with a LEO satellite constellation. First, we distribute the computation and inference load among the neighbouring satellites of the one taking the images, based on the VHRShips data set and YOLOv8. This semantic and fragmented information is then routed to a remote ground monitor through the whole constellation. The average and peak Age of Information (AoI) are reformulated to measure the freshness of the aggregated information at the receiver in this image-capture scenario. We then dimension the network (number of orbital planes and satellites per orbital plane) for a given target age and covered area that quantify the level of achievement of the task. The results show that 20 orbital planes with 20 satellites are necessary to keep the peak AoI below 60 seconds with a compression ratio > 23000, i.e., a size reduction of 99.996%, and for a approximately 100% probability of coverage.

Goal-oriented vessel detection with distributed computing in a LEO satellite constellation

TL;DR

The paper addresses near real-time vessel surveillance by leveraging distributed edge computing within a LEO satellite constellation. It combines YOLOv8-based vessel detection on the VHRShips dataset with a fragmented, parallel processing pipeline across neighboring satellites, and quantifies timeliness using AoI metrics to guide constellation sizing. The study demonstrates that a 20×20 walker-delta topology with five processing satellites can maintain peak AoI below 60 seconds while achieving a data reduction of over 99.996% and effectively full-area coverage, illustrating the practical viability of edge-enabled EO missions. This work offers a concrete design methodology for timely sensing in space and points to future extensions, including ground-queried closed loops and atmospheric effects.

Abstract

Earth Observation (EO) has traditionally involved the transmission of a large volume of raw data to map the Earth surface. This results in congestion to the satellite network and delays in the availability of the results, invalidating the approach for timing-sensitive applications. Instead, the computation resources at the satellites can be used as an edge layer for compressing the data and/or doing inferences. In this paper, we investigate satellite edge computing for vessel detection with a LEO satellite constellation. First, we distribute the computation and inference load among the neighbouring satellites of the one taking the images, based on the VHRShips data set and YOLOv8. This semantic and fragmented information is then routed to a remote ground monitor through the whole constellation. The average and peak Age of Information (AoI) are reformulated to measure the freshness of the aggregated information at the receiver in this image-capture scenario. We then dimension the network (number of orbital planes and satellites per orbital plane) for a given target age and covered area that quantify the level of achievement of the task. The results show that 20 orbital planes with 20 satellites are necessary to keep the peak AoI below 60 seconds with a compression ratio > 23000, i.e., a size reduction of 99.996%, and for a approximately 100% probability of coverage.

Paper Structure

This paper contains 8 sections, 11 equations, 9 figures.

Table of Contents

  1. Introduction
  2. State of the Art
  3. System model
  4. Constellation design and timing metrics
  5. Constellation design
  6. Timing metrics
  7. Simulation and results
  8. Conclusions

Figures (9)

  • Figure 1: Example of YOLOv8 vessel detection algorithm yolov8_ultralytics performance over an image of the VHRShips datasetijgi11080445.
  • Figure 2: Sketch of the scenario.
  • Figure 3: Example of a frame captured and divided into images of a specific resolution.
  • Figure 4: Evolution of the .
  • Figure 5: Service availability taking a quality frame for vessel detection. Starlink-like topology with $M = 10, 15, 30$, and $N = 22$.
  • ...and 4 more figures