A Comprehensive Survey on Orbital Edge Computing: Systems, Applications, and Algorithms

summary

The gist

The paper provides a comprehensive survey of Orbital Edge Computing (OEC), covering its systems, applications…

In short

This survey examines Orbital Edge Computing (OEC), a paradigm where computing power is moved to satellites to process data in orbit. It covers OEC systems, applications like remote sensing and IoV, and the necessary algorithms for optimization. The research shows OEC can revolutionize communication and data analysis by overcoming terrestrial network limitations.

Key concepts

Orbital Edge Computing (OEC)
OEC places computing resources directly onto satellites to act as edge nodes. This allows satellites to perform tasks like preprocessing sensor data in space, improving communication performance and enabling real-time remote sensing analysis.
Contrast with Mobile Edge Computing (MEC)
Unlike MEC, which focuses on mobile devices, OEC utilizes the mobility of LEO satellites and their limited resources. This approach is designed to alleviate pressure on satellite communication systems by processing data where it is collected.
Computation Offloading
This involves deciding whether to perform a computation on the satellite or send it to ground infrastructure. The paper models this as an optimization problem, often using deep reinforcement learning methods like DQN to minimize task delay and energy use.
Simulation and Testbeds
Because OEC involves unique space-ground interactions, simulation tools like SNS3 are used to model large satellite constellations. Platforms such as Celestial provide realistic testbeds for evaluating high-definition video tasks using actual satellites.

Terminology used across episodes

This episode discusses

The paper

A Comprehensive Survey on Orbital Edge Computing: Systems, Applications, and Algorithms · Read on arXiv

Changhao Wu, Yuanchun Li, Mengwei Xu, Chongbin Guo, Zengshan Yin, Weiwei Gao, Chuanxiu Chi

Innovation Academy for Microsatellites of Chinese Academy of Science of China University of Chinese Academy of Sciences Beijing University of Posts and Telecommunications

The number of satellites, especially those operating in low-earth orbit (LEO), is exploding in recent years. Additionally, the use of COTS hardware into those satellites enables a new paradigm of computing: orbital edge computing (OEC). OEC entails more technically advanced steps compared to single-satellite computing. This feature allows for vast design spaces with multiple parameters, rendering several novel approaches feasible. The mobility of LEO satellites in the network and limited resources of communication, computation, and storage make it challenging to design an appropriate scheduling algorithm for specific tasks in comparison to traditional ground-based edge computing. This article comprehensively surveys the significant areas of focus in orbital edge computing, which include protocol optimization, mobility management, and resource allocation. This article provides the first comprehensive survey of OEC. Previous survey papers have only concentrated on ground-based edge computing or the integration of space and ground technologies. This article presents a review of recent research from 2000 to 2023 on orbital edge computing that covers network design, computation offloading, resource allocation, performance analysis, and optimization. Moreover, having discussed several related works, both technological challenges and future directions are highlighted in the field.

DOI: 10.1016/j.cja.2024.11.026

Transcript

Introduction to the show: ident: Computer Science. The day in computer science: software, systems and theory on arXiv.

Tom: I'm Tom.

Jane: And I'm Jane, here with Lu, Meng and our in-house large language model, Lalam.

Tom: We're looking at the paper: "A Comprehensive Survey on Orbital Edge Computing".

Paper summary: Tom: So we're looking at this paper, "A Comprehensive Survey on Orbital Edge Computing: Systems, Applications, and Algorithms." What’s the main thing they’re telling us about orbital edge computing?

Jane: Well, basically, they are surveying the whole field. The core idea is that because we have so many satellites now and cheaper computer chips in them, we can push computing power up into space.

Lu: That's right. They define orbital edge computing as a network where the satellite itself acts as an edge node, which is different from just putting a small computer on the ground <ref:2306.00275#pg3>.

Meng: So what’s the big payoff here? What problem are they trying to solve with this whole concept?

Tom: The paper claims that this approach improves communication performance because you process sensor data right up in orbit, which leads to things like on-orbit data preprocessing and analyzing remote sensing images in real time <ref:2306.00275#pg4>.

Jane: It’s contrasting with mobile edge computing, which uses the movement of LEO satellites and limited resources to ease pressure on satellite communication <ref:2306.00275#pg5>. So, it's about using the space environment differently to save bandwidth back down to Earth.

Lu: The architecture relies on these satellite constellations, which offer better coverage and lower launch costs compared to older systems <ref:2306.00275#pg9>. They also point out that LEO satellites are particularly good because they have lower latency and cheaper launch costs than geostationary satellites <ref:2306.00275#pg10>.

Meng: Lower launch costs are important, but what about the actual hardware? What kind of stuff are these edge nodes running?

Tom: The system involves optimizing both the satellite hardware and its software architecture, focusing on how it communicates and computes <ref:2306.00275#pg2>. They mention using Commercial Off-the-Shelf devices with advanced chip manufacturing for implementing state-of-the-art AI algorithms while keeping power consumption low <ref:2306.00275#pg8>.

Jane: It sounds like the operating system on these spacecraft, things like Linux or SpaceOS, has to be really reliable because the space environment is so harsh <ref:2306.00275#pg9>. It's not just a simple cloud setup; it’s dealing with real physics and space constraints.

Lu: And that brings us into the algorithms they survey. They look at computation offloading, resource management, and mobility management as the three main areas of focus <ref:2306.00275#pg2>.

Paper summary: Tom: Computation offloading is modeled like a mixed integer nonlinear programming problem, and they use deep reinforcement learning methods like Deep Q-network to minimize task delay and energy consumption <ref:2306.00275#pg2>. That sounds computationally heavy.

Meng: If we were trying to deploy this, what’s the cost of running those kinds of AI algorithms on a satellite? What breaks at scale?

Lu: Resource management is often formulated as a mixed-integer programming problem, and they suggest using methods like the Lagrange multiplier method to find solutions <ref:2306.00275#pg2>. Mobility management then uses the predictable paths of satellites to assign tasks based on channel conditions <ref:2306.00275#pg2>.

Jane: So, for someone who just listens, this means that when you're using a service like ultra-long distance video conferencing or tracking an object, the processing isn't all happening down on Earth <ref:2306.00275#pg10>. It’s happening where the data is, which cuts down on latency.

Tom: Exactly. And they look at applications like object tracking and disaster surveillance, but also things like providing positioning and navigation services for the Internet of Vehicles in remote areas <ref:2306.00275#pg5>.

Meng: But the challenges must be huge if they're surveying this stuff. What are the real roadblocks they see?

Lu: They point out that resource-constrained satellite payloads and the complexity from large-scale satellite networks create NP-hard optimization problems <ref:2306.00275#pg2>. It’s hard to find a perfect solution when everything is constrained.

Jane: So, what are the future directions they suggest? Where is this going next?

Tom: They emphasize software-defined satellites, using inter-satellite laser links for better communication efficiency, and developing more accurate testbeds to simulate real space environments <ref:2306.00275#pg2>. The goal is autonomous task scheduling and efficient resource allocation in these complex scenarios <ref:2306.00275#pg2>.

Meng: Simulation is key here, right? How do you test something that’s in space without actually launching a satellite?

Jane: They stress that simulation tools are essential because you can't just plug it into the ground network simulator <ref:2306.00275#pg2>. They mention simulators like SNS3 and Hypatia, which model large low earth orbit constellations based on NS-three <ref:2306.00275#pg2>.

Paper summary: Lu: They also highlight testbeds like Celestial, which lets you run high-definition video conference tasks using Starlink satellites to get a realistic environment for evaluation <ref:2306.00275#pg2>. These tools help analyze how satellite configuration affects the algorithm performance <ref:2306.00275#pg2>.

Tom: So, we’ve covered the systems, the applications, and the algorithms they are looking at in this survey of Orbital Edge Computing. It's a broad look at what's possible <ref:2306.00275#pg1>.

Jane: And it really shows how feasibility is being established through designing these optimization algorithms and specialized testbeds <ref:2306.00275#pg2>. It’s moving from theory to something that can be tested in a more realistic setting.

Lu: The big implication for the field is showing that OEC has wide-ranging potential across space technology, agricultural remote sensing, the military, and even next-generation mobile communication <ref:2306.00275#pg2>.

Meng: For deployment right now, I’m still focused on those resource constraints and the complexity of scheduling tasks in these massive networks <ref:2306.00275#pg2>. We need to know what actually breaks when you try to scale this up <ref:2306.00275#pg1>.

Tom: So, the paper sets the stage by mapping out all the significant areas of focus, which is helpful for anyone trying to tackle these complex scheduling problems <ref:2306.00275#pg1>. It gives us a clear map of what’s been researched so far <ref:2306.00275#pg1>.

Jane: And ultimately, it suggests that OEC is expected to enable significant changes in the edge computing landscape and help grow the Internet of Things <ref:2306.00275#pg2>. It’s a big area for future development.

Lu: It really does suggest that satellite edge computing has proven its feasibility through the design of optimization algorithms and specialized testbeds <ref:2306.00275#pg2>. That’s a solid finding based on the survey <ref:2306.00275#pg1>.

Meng: We need to keep looking at those simulation tools and see if they can actually predict what happens when we push these models to real space conditions <ref:2306.00275#pg2>. That’s where the practical engineering work starts <ref:2306.00275#pg1>.

Tom: It seems like the survey itself is a really comprehensive resource for understanding the state of play in orbital edge computing right now <ref:2306.00275#pg1>. It’s a good starting point for anyone getting into this area <ref:2306.00275#pg1>.

Jane: That’s the gist of it. We’ve seen how the authors map out all these different technological components and challenges in this paper about Orbital Edge Computing <ref:2306.00275#pg1>.

Conclusion: Tom: So we're wrapping up this look at Orbital Edge Computing, which is really just a big overview paper by several authors <ref:2306.00275#pg1>.

Jane: It’s a comprehensive survey, Tom, meaning they’ve covered the whole landscape of these systems and applications.

Lu: Exactly. They went deep into the hardware, the software architecture, and all the math needed to make it work in orbit <ref:2306.00275#pg2>.

Meng: So what’s the main point they’re making at this conclusion about what this whole field means?

Tom: They're saying that orbital edge computing has proven its potential across a ton of different areas, from space tech to agricultural sensing and even mobile communication <ref:2306.00275#pg2>.

Jane: It seems they’re emphasizing that the feasibility of using satellites for edge computing isn't just theoretical anymore, it’s supported by real optimization algorithms and testbeds <ref:2306.00275#pg1>.

Lu: They point out that this approach is expected to make some significant changes to how we think about edge computing overall and how we grow the Internet of Things <ref:2306.00275#pg1>.

Tom: It’s a big shift from just thinking about edge computing on Earth to thinking about it up there in orbit <ref:2306.00275#pg1>.

Jane: And the authors suggest that we need to keep building those simulation tools and testbeds to make this actually happen in a real space environment <ref:2306.00275#pg1>.

Lu: Right, so they’ve mapped out the challenges—like those NP-hard optimization problems—and pointed toward the next steps for autonomous task scheduling <ref:2306.00275#pg2>.

Meng: So we’ve seen the scope of this work, but what does that mean for how we actually build these systems in the next few years?

More episodes

← Home