OPC-based energy priority satellite edge service device

Abstract

The invention relates to an OPC (optical proximity correction)-based energy priority satellite edge service device, which comprises a constellation-level orbit perception request distribution module (ORD), a satellite-level pulse task scheduling module (PTS) and a task-level energy perception runtime management module (ERM). And high-throughput, low-delay and low-energy-consumption intelligent services in a multi-satellite and multi-task environment are realized.

Description

Technical Field

[0001] This invention relates to a technology in the field of satellite control, specifically an energy-priority satellite edge service device based on pulsed orbit calculation (OPC). Background Art

[0002] While existing on-orbit edge computing architectures have improved on-orbit data processing capabilities to some extent, they lack energy sensing and multi-satellite, multi-task collaborative scheduling. This results in low overall system resource utilization, high task latency, and difficulty in adapting to future-oriented time-sensitive and energy-efficient scenarios. For example, current satellite constellations using IoT and satellite edge computing technologies generally only perform preliminary analysis of transmitted data. They lack the ability to perform high-density on-board inference of data captured by remote sensing satellites, and they cannot manage energy effectively. Summary of the Invention

[0003] This invention addresses the shortcomings of existing technologies in effectively supporting high-performance, low-latency on-orbit computing services in energy-constrained satellite environments, particularly in handling pulsed data streams and dynamic task allocation. It proposes an energy-priority satellite edge service device based on OPC, which achieves high-throughput, low-latency, and low-energy-consumption intelligent services in multi-satellite, multi-task environments through three-level perception and management at the satellite constellation, satellite, and mission levels.

[0004] This invention is achieved through the following technical solution:

[0005] This invention relates to an OPC-based energy-priority satellite edge service device, comprising: a constellation-level orbital awareness request distribution module (ORD), a satellite-level pulse task scheduling module (PTS), and a mission-level energy-aware runtime management module (ERM). The constellation-level ORD operates at a ground station or ground data center, dividing user requests into tasks to support a multi-satellite mode for a single task, and distributing these tasks according to the status of each satellite in the constellation, thus achieving balanced task allocation across satellites. The satellite-level pulse task scheduling module is deployed on the satellite payload system, optimizing the scheduling of given tasks from the ground by considering energy and pulse conditions, thereby reducing satellite system service latency. The mission-level energy-aware runtime management module is deployed within the execution unit of the satellite payload system, analyzing the computational requirements of each task in the queue and the computational energy budget for each future pulse, selecting the optimal task execution strategy according to the scheduling mode, and generating a timetable list to determine the processes in the next pulse, thereby reducing the instantaneous power of the satellite payload and improving the energy utilization rate of the satellite.

[0006] Technical Effects

[0007] This invention supports constellation-wide energy-aware adaptive scheduling, pulsed short-duration high-density computing and energy-consumption sprints, and dynamic computing power allocation and energy consumption prediction at the task level. Through refined energy and task-aware scheduling, it maximizes the use of limited energy resources, avoids unnecessary energy consumption and task blocking, and ensures that on-orbit high-performance computing is possible within the energy budget. Compared with existing technologies, this invention can significantly improve service throughput and reduce service latency while ensuring computing platform performance and system stability. Ultimately, the performance improvement advantage of this invention is mainly measured by request throughput and latency. Attached Figure Description

[0008] Figures 1-4 This is a schematic diagram of the structure of the present invention;

[0009] Figure 5 This is a graph showing the throughput results of the satellite during operation in the example;

[0010] Figure 6 This is a diagram showing the time delay results during satellite operation in the example. Detailed Implementation

[0011] like Figures 1-4 As shown in this embodiment, an energy-priority satellite edge service device based on OPC is included, comprising: a constellation-level orbital awareness request distribution module (ORD), a satellite-level pulse task scheduling module (PTS), and a task-level energy awareness runtime management module (ERM). The constellation-level ORD operates at a ground station or ground data center, dividing user requests into tasks to support a multi-satellite mode for a single task, and distributing these tasks according to the status of each satellite in the constellation, thus achieving balanced task allocation across satellites. The satellite-level pulse task scheduling module is deployed on the satellite payload system, optimizing the scheduling of given tasks from the ground by considering energy and pulse conditions, thereby reducing satellite system service latency. The task-level energy awareness runtime management module is deployed within the execution unit of the satellite payload system, analyzing the computational requirements of each task in the queue and the computational energy budget for each future pulse, selecting the optimal task execution strategy according to the scheduling mode, and generating a timetable list to determine the processes in the next pulse, thereby reducing the instantaneous power of the satellite payload and improving the energy utilization rate of the satellite.

[0012] like Figure 2As shown, the constellation-level orbital awareness request distribution module includes: a user request manager, a satellite mission distributor, and a constellation status monitor. Specifically: the user request manager divides large requests into multiple sub-tasks suitable for parallel processing based on the number of satellites and mission energy consumption; the satellite mission distributor collects real-time energy, sub-task, and orbital information of each satellite and calculates mission allocation delay estimates; the constellation status monitor collects real-time satellite orbit and energy consumption data and dynamically feeds it back to the satellite mission distributor, supporting adaptive scheduling and historical energy prediction.

[0013] The aforementioned segmentation specifically refers to: Where: Task is the task block size, Sreq is the total number of requests, Nsat is the number of constellation satellites, Emin is the minimum task energy consumption threshold, and Eopc is the energy consumption per inference. After calculating the task block size, the total task is divided into n tasks of size Task, and each task is distributed to the satellites separately.

[0014] The task allocation delay estimation mentioned above is specifically as follows: Where: Ddist is the distribution waiting time, Dsense is the remote sensing waiting time, De is the energy budget waiting time, and Pin is the available power. After the task allocation delay is calculated, it is used as the time estimation parameter for the task. The satellite with the smallest delay is selected from all satellites, and the task is allocated to this satellite.

[0015] The constellation status monitor returns current orbit information, power, and energy status to the ground via satellite sensor data. Each time it receives data from the satellite, the constellation status monitor resends the latest real-time satellite orbit and energy consumption data to the mission distributor for reference. The real-time orbit and energy consumption data of the constellation are required as independent variables in adaptive scheduling and energy history prediction. Adaptive scheduling and energy prediction are supported by using the latest parameters.

[0016] like Figure 3 As shown, the satellite-level pulse task scheduling module includes a task manager, a satellite manager, and a pre-pulse scheduler. The task manager maintains the current task pool and energy tracking information through a data table and updates the data table based on real-time status updates. The satellite manager maintains a remote sensing plan table, a pulse cycle table, and an energy budget table. The pre-pulse scheduler, based on the current cycle energy budget and task queue, adopts a first-come, first-served (FCFS) strategy. After each task runs, it updates the current task pool and energy tracking information, obtains the task execution plan for the next pulse cycle, and determines whether to perform an energy sprint operation based on the remaining energy.

[0017] The energy sprint calculation refers to: using a large amount of inference hardware and a large amount of energy to perform rapid calculations in a short period of time, obtaining the calculation results with minimal latency, and then transmitting the results back to the ground as conclusion data.

[0018] like Figure 4 As shown, the task-level energy-sensing runtime management module includes: a runtime computing manager, a dynamic computing processor, and a runtime queue unit. The runtime computing manager prioritizes battery energy for sprint tasks based on the ptqueue and csprint output by the PTS, and dynamically adjusts the number of parallel units (PEs) based on solar energy input in other stages. The dynamic computing processor controls the dynamic shutdown and mapping switching of the processing units (PEs) via control signals to achieve energy savings. The runtime queue unit performs remote sensing operations, computation operations, and data transmission operations independently and in parallel, achieving efficient pipelined processing of data and task flows.

[0019] Through specific experiments, the energy-priority satellite edge service device for pulsed orbit calculation was started with 10 satellites in polar orbit, each satellite set to 3U size, and 2 available ground stations, under the conditions of running 8 user request tasks including CropLand, FCN, Resnet50, Retinanet, Shufflenet, STANet, VGG19, and YoloV5.

[0020] like Figure 5 and Figure 6 The figure shows a comparison of the advantages of this embodiment and existing satellite service schemes in terms of increasing throughput and reducing latency. The nine schemes in the figure include: (a) COTE-like represents the basic strategy of orbital edge computing. This strategy executes only one request at a time and must wait for the current request to be completed before continuing to execute the next one. (b) Aimless represents the current state-of-the-art (SOTA) strategy without task awareness. This strategy can execute multiple requests simultaneously, but the ground station will assign the task to the first connectable satellite, and the satellite will execute it immediately after receiving the task. (c) RCM-like represents a strategy that dedicates satellites to specific tasks. This strategy can only execute one request on each satellite at a time. (d) Separate strategy represents a variant of our method that does not consider energy consumption. It can execute multiple requests simultaneously at the entire constellation level, but the energy status of each satellite is not aware, so energy consumption awareness cannot be achieved in task allocation. (e) EPOCH is the operating strategy of the energy-priority satellite edge service device based on pulsed orbital computing in this embodiment.

[0021] As shown in the figure, compared with other solutions, this embodiment can effectively increase the task throughput that the satellite constellation can handle per unit time, effectively reduce the task reception latency of the satellite constellation in almost all cases, and ensure that the latency is within the acceptable range for users.

[0022] The above-described specific implementations can be partially adjusted by those skilled in the art in different ways without departing from the principles and purpose of the present invention. The scope of protection of the present invention is defined by the claims and is not limited to the above-described specific implementations. All implementation schemes within the scope of the claims are bound by the present invention.

Claims

1. An OPC-based energy-priority satellite edge service device, characterized in that, include: The constellation-level orbital sensing request distribution module (ORD), satellite-level pulse mission scheduling module (PTS), and mission-level energy sensing runtime management module (ERM) are as follows: The constellation-level orbital sensing request distribution module runs at the ground station or ground data center, divides user requests into tasks to support a multi-satellite mode for a single task, and distributes these tasks according to the status of each satellite in the constellation to balance the task allocation on the satellites. The satellite-level pulse mission scheduling module is deployed on the satellite payload system. It optimizes the scheduling of a given mission from the ground by considering energy and pulse conditions, thereby reducing the service latency of the satellite system. The mission-level energy-aware runtime management module is deployed within the execution unit of the satellite payload system. It analyzes the computational requirements of each mission in the queue and the computational energy budget for each future pulse. Based on the scheduling mode, it selects the best mission execution strategy and generates a timetable list that determines the process in the next pulse, thereby reducing the instantaneous power of the satellite payload and improving the energy utilization rate of the satellite.

2. The OPC-based energy-priority satellite edge service device according to claim 1, characterized in that, The constellation-level orbit awareness request distribution module includes: a user request manager, a satellite mission distributor, and a constellation status monitor. The user request manager divides large requests into multiple sub-tasks suitable for parallel processing based on the number of satellites and mission energy consumption. The satellite mission distributor collects real-time energy, sub-task, and orbit information for each satellite and calculates mission allocation delay estimates. The constellation status monitor collects real-time satellite orbit and energy consumption data and dynamically feeds it back to the satellite mission distributor, supporting adaptive scheduling and historical energy prediction.

3. The OPC-based energy-priority satellite edge service device according to claim 2, characterized in that, The aforementioned segmentation specifically refers to: Where: Task is the task block size, Sreq is the total number of requests, Nsat is the number of constellation satellites, Emin is the minimum task energy consumption threshold, and Eopc is the energy consumption per inference. After calculating the task block size, the total task is divided into n tasks of size Task, and each task is distributed to the satellites separately.

4. The OPC-based energy-priority satellite edge service device according to claim 2, characterized in that, The task allocation delay estimation mentioned above is specifically as follows: Where: Ddist is the distribution waiting time, Dsense is the remote sensing waiting time, De is the energy budget waiting time, and Pin is the available power. After the task allocation delay is calculated, it is used as the time estimation parameter for the task. The satellite with the smallest delay is selected from all satellites, and the task is allocated to this satellite.

5. The OPC-based energy-priority satellite edge service device according to claim 2, characterized in that, The constellation status monitor returns current orbit information, power, and energy status to the ground via satellite sensor data. Each time it receives data from the satellite, the constellation status monitor resends the latest real-time satellite orbit and energy consumption data to the mission distributor for reference. The real-time orbit and energy consumption data of the constellation are required as independent variables in adaptive scheduling and energy history prediction. Adaptive scheduling and energy prediction are supported by using the latest parameters.

6. The OPC-based energy-priority satellite edge service device according to claim 1, characterized in that, The satellite-level pulse task scheduling module includes a task manager, a satellite manager, and a pre-pulse scheduler. The task manager maintains the current task pool and power tracking information through a data table and updates the data table based on real-time status updates. The satellite manager maintains a remote sensing plan table, a pulse cycle table, and an energy budget table. The pre-pulse scheduler, based on the current cycle energy budget and task queue, adopts a first-come, first-served (FCFS) strategy. After each task finishes execution, it updates the current task pool and power tracking information, obtains the task execution plan for the next pulse cycle, and determines whether to perform an energy sprint operation based on remaining energy.

7. The OPC-based energy-priority satellite edge service device according to claim 1, characterized in that, The task-level energy-sensing runtime management module includes a runtime computing manager, a dynamic computing processor, and a runtime queue unit. The runtime computing manager prioritizes battery energy for sprint tasks based on the ptqueue and csprint output by the PTS, dynamically adjusting the number of parallel units (PEs) based on solar energy input in other stages. The dynamic computing processor controls the dynamic shutdown and mapping switching of the processing units (PEs) via control signals to achieve energy savings. The runtime queue unit performs remote sensing operations, computation operations, and data transmission operations independently and in parallel, achieving efficient pipelined processing of data and task flows.

Images