Method and device for regulating and controlling computing power resources based on high-throughput satellites

Abstract

The invention relates to the technical field of computing power networks, and discloses a computing power resource regulation and control method and device based on a high-throughput satellite, and the method comprises the steps: receiving a target service request, and obtaining the data characteristics of the target service request; sending the data features to a prediction model, and determining a task type of the target service request; calculating a target computing power demand of the target service request based on the task type; and determining a target route meeting the target computing power demand, and sending a target instruction to a satellite on the target route to drive the satellite on the target route to execute the target service request. According to the invention, a computing power resource monitoring and scheduling platform is deployed on a high-throughput satellite of a high orbit, and the transmission reliability can be improved while the scheduling transmission delay of the satellite is reduced.

Description

Technical Field

[0001] The present invention relates to the field of computing power network technology, and in particular to a computing power resource control method and device based on high-throughput satellites. Background Art

[0002] With the rapid deployment of low-orbit satellite constellations such as Starlink and OneWeb, high-throughput satellite networks are becoming the core infrastructure for global communications, remote sensing, and the Internet of Things. However, traditional satellite systems rely on centralized task scheduling at ground stations, which can lead to high latency, single-point failure risks, and resource bottlenecks in satellite-to-ground links.

[0003] Related technologies primarily schedule onboard computing resources through ground networks, which increases round-trip transmission latency and the risk of attack. While offloading communication tasks to idle servers at the edge can alleviate computing pressure, it requires multi-hop transmission, which increases the complexity of routing calculations for high-speed low-orbit satellites.

[0004] In view of this, a low-latency, high-reliability computing resource control method for high-throughput satellites is needed. Summary of the Invention

[0005] In view of this, the present invention provides a computing resource control method based on high-throughput satellites, which can reduce the transmission delay of scheduling while improving the reliability of transmission.

[0006] In the first aspect, the present invention provides a computing power resource control method based on high-throughput satellites, the method comprising: receiving a target service request and obtaining data features of the target service request; sending the data features to a prediction model to determine the task type of the target service request; based on the task type, calculating the target computing power requirement of the target service request; determining a target route that meets the target computing power requirement, and sending a target instruction to the satellite on the target route to drive the satellite on the target route to execute the target service request.

[0007] In this implementation, the target service request and its data features are obtained and then fed into a prediction model to determine the target service request's task type. This determines the target computing power requirement for the target service request, identifies a route that satisfies the target computing power requirement, and then sends instructions to satellites on that route to instruct them to execute the target service request. This approach, which determines the target service request's task type based on its data features, improves judgment accuracy compared to directly using the target service request's inherent attributes as the task type, thereby enabling precise regulation of computing resources.

[0008] In an optional embodiment, the method also includes: obtaining a sample data set; the sample data set includes sample flow data and sample annotations; cutting the sample flow data into time windows of target length, and inputting the time windows into the prediction model to obtain classification results; based on the error between the classification results and the sample annotations, updating the prediction model through a back propagation algorithm.

[0009] In this implementation, a sample data set is obtained, then the sample traffic data is cut into time windows of target length, which are then fed into the prediction model to obtain classification results. Based on the error, the prediction model is then updated using a backpropagation algorithm. This allows the initial model to better learn the dynamic patterns of time series data, improving model accuracy.

[0010] In an optional embodiment, when the task type is video transmission, the target computing power requirement of the target service request is calculated based on the task type, including: obtaining several video transmission links of the video to be transmitted; based on the encoding standard of the video to be transmitted, determining the computing power requirement of each video transmission link to calculate the target computing power requirement of the target service request.

[0011] In this embodiment, when the task type is video transmission, the video transmission link of the video to be transmitted is obtained, and based on the encoding standard of the video to be transmitted, the target computing power requirement corresponding to the target service request is obtained. This can optimize resource utilization efficiency and improve video transmission quality.

[0012] In an optional embodiment, when the task type is sensor data transmission, the target computing power requirement of the target service request is calculated based on the task type, including: obtaining several sensor transmission links of the sensor data to be transmitted; obtaining the operation type and data volume of the sensor data to be transmitted; determining the computing power requirement of each sensor transmission link; and calculating the target computing power requirement of the target service request based on the computing power requirement of each sensor transmission link.

[0013] In this embodiment, when the task type is sensor data transmission, by obtaining the sensor transmission link, operation type and data volume of the sensor data to be transmitted to determine the target computing power requirement corresponding to the target service request, resource utilization efficiency can be optimized and the quality of sensor data transmission can be improved.

[0014] In an optional embodiment, when the task type is file data transmission, the target computing power requirement of the target service request is calculated based on the task type, including: obtaining several file transfer links of the file data to be transferred; obtaining the file protocol type of the file data to be transferred to determine the computing power requirement of each file transfer link; and calculating the target computing power requirement of the target service request based on the computing power requirement of each file transfer link.

[0015] In this embodiment, when the task type is file data transmission, by obtaining the sensor transmission link and file protocol type of the file data to be transmitted to determine the target computing power requirement corresponding to the target service request, resource utilization efficiency can be optimized and the quality of file data transmission can be improved.

[0016] In an optional embodiment, determining a target route that meets the target computing power requirement includes: obtaining the delay requirement of the target service request based on the task type to determine the routing parameters of the target route; the routing parameters include the number of satellites and the business requirements of a single satellite; screening a number of target satellites from a preset satellite set; the target satellite meets the business requirements of a single satellite; and establishing the target route based on the routing parameters and the position of the target satellite.

[0017] In this implementation, the latency requirements of the target service request are obtained to determine the routing parameters of the target route, thereby screening the target satellite. The target route is then established based on the routing parameters and the location of the target satellite. This can disperse the computing power pressure on the target satellite and improve the flexibility of target route selection.

[0018] In an optional embodiment, determining the routing parameters of the target route includes: constructing a routing delay model; the routing delay model includes the satellite's data processing delay and the routing transmission delay; substituting the delay requirement of the target service request into the routing delay model to determine the routing parameters of the target route.

[0019] In this embodiment, by constructing a routing delay model and substituting the delay requirements of the target service request into the routing delay model to determine routing parameters, the path selection of the target route can be optimized and the robustness of the route can be improved.

[0020] In an optional implementation, the method further includes: collecting and updating basic information of satellites in a preset satellite set based on a preset time interval.

[0021] In this embodiment, basic information of satellites in a preset satellite set is collected and updated according to a preset time interval, which can improve the accuracy of target route selection.

[0022] In an optional implementation, after executing the target service request, the satellite on the target route generates a return result and sends the return result to the sender of the target service request.

[0023] In this embodiment, after executing the target service request, the satellite on the target route generates a return result and sends the return result to the sender of the target service request, which can save routing time and reduce the computing pressure of the satellite.

[0024] In the second aspect, the present invention provides a computing power resource control device based on a high-throughput satellite, the device including: a receiving module for receiving a target service request and obtaining data features of the target service request; a determination module for sending the data features to a prediction model to determine the task type of the target service request; a calculation module for calculating the target computing power requirement of the target service request based on the task type; an execution module for determining a target route that meets the target computing power requirement, and sending a target instruction to the satellite on the target route to drive the satellite on the target route to execute the target service request. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the specific embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0026] Figure 1 1 is a flow chart of a method for controlling computing resources according to an embodiment of the present invention;

[0027] Figure 2 is a flow chart of another computing resource control method according to an embodiment of the present invention;

[0028] Figure 3 2. It is a schematic structural diagram of a satellite computing resource monitoring and scheduling platform according to an embodiment of the present invention;

[0029] Figure 4 is a schematic diagram of a routing topology according to an embodiment of the present invention;

[0030] Figure 5 This is a structural block diagram of a computing resource control device according to an embodiment of the present invention;

[0031] Figure 6 Schematic diagram of the hardware structure of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0032] To make the purpose, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without making creative efforts shall fall within the scope of protection of the present invention.

[0033] With the rapid deployment of low-orbit satellite constellations (such as Starlink and OneWeb), high-throughput satellite networks have gradually become the core infrastructure for global communications, remote sensing, and the Internet of Things. However, traditional satellite systems rely on ground stations for centralized task scheduling, and face problems such as high latency, single-point failure risks, and satellite-to-ground link resource bottlenecks. Especially in high-speed low-orbit constellations, the high-speed movement of satellite nodes and frequent switching of coverage areas place higher demands on real-time task response and resource utilization, and the computing power resources on low-orbit satellites are very limited, making it difficult to carry high-performance computing equipment. Therefore, the present invention proposes to deploy on-board computing power resources and scheduling platforms on high-throughput satellites, and through distributed computing architecture and intelligent resource management, dynamically allocate communication tasks to multiple satellites for collaborative processing, thereby achieving satellite-based services with low latency, high reliability, high performance, and low computing power pressure.

[0034] In related technologies, onboard computing resource scheduling still relies on ground networks, increasing round-trip transmission latency and the risk of attack. While offloading communication tasks to idle edge servers can alleviate computing pressure, it requires multi-hop transmission, which increases the complexity of routing calculations for high-speed low-orbit satellites. Computing routing simply considers computing power demand and the shortest route. In a complex satellite-ground fusion network, multiple factors influence and constrain the allocation of computing power onboard. This makes it impossible to efficiently utilize the computing power of the entire low-orbit satellite network, resulting in serious load imbalance.

[0035] An embodiment of the present invention provides a method for regulating computing power resources based on high-throughput satellites. This method obtains a target service request and its data features, then sends these features to a prediction model to determine the task type of the target service request. This method then calculates the target computing power requirement for the target service request, determines a route that satisfies the target computing power requirement, and then sends instructions to the satellite on the route to instruct it to execute the target service request. This scheme determines the task type of the target service request based on its data features. This improves judgment accuracy compared to directly using the target service request's inherent attributes as the task type, thereby enabling precise regulation of computing power resources.

[0036] According to an embodiment of the present invention, an embodiment of a computing resource control method based on a high-throughput satellite is provided. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0037] In this embodiment, a computing resource control method based on high-throughput satellite is provided. Figure 1FIG. 1 is a flow chart of a method for controlling computing power resources according to an embodiment of the present invention. Figure 1 As shown, the process includes the following steps:

[0038] Step S101: receiving a target service request and obtaining data characteristics of the target service request.

[0039] In a satellite environment, the satellite's computing power, energy and other resources are subject to certain limitations. To ensure the smooth completion of the mission, the specific requirements of the mission can be determined by receiving target service requests and obtaining the data characteristics of the target service requests, thereby regulating the computing power resources.

[0040] The target service request can originate from a ground station or other satellite. Upon receiving the target service request, the target service request can be parsed to obtain the data characteristics of the target data request. These data characteristics can include data size, time constraints, computational complexity, and traffic characteristics, facilitating subsequent determination of the task type of the target service request.

[0041] Step S102: Send the data features to the prediction model to determine the task type of the target service request.

[0042] The prediction model may be a neural network model. After receiving the data features, the neural network model generates an output result, and the output result is the task type of the target service request.

[0043] In some optional implementations, the prediction model may also be a classification model based on preset rules, using traffic feature analysis in the implicit inference method to evaluate the business type of the target service request. In a specific implementation, when the data features are large packets, high bandwidth, and periodicity, the task type of the target service request can be determined to be video. When the data features are small packets, low rate, and periodicity, the task type of the target service request can be determined to be IoT sensor data. When the data features are continuous high throughput and TCP (Transmission Control Protocol) long connections, the task type of the target service request can be determined to be file.

[0044] Step S103: Calculate the target computing power requirement of the target service request based on the task type.

[0045] Computing power indicators can be preset for different task types to form standardized templates. Computing power indicators can include memory size, network bandwidth, etc.

[0046] After receiving the target service request, the target computing power requirement of the target service request is calculated based on the task type of the target service request and the template corresponding to the task type.

[0047] Step S104: determine a target route that meets the target computing power requirement, and send a target instruction to the satellite on the target route to drive the satellite on the target route to execute the target service request.

[0048] Based on the target computing power requirements, conditions such as the number of satellites in the target route and the computing power requirements of each satellite can be determined. Then, satellites that meet these requirements are selected from the surrounding satellites, and the target route is constructed based on their locations. In practice, satellites that are close to each other can be selected to construct the route to minimize transmission latency.

[0049] After determining the target route, determine the services undertaken by each satellite on the target route; generate and send target instructions to the corresponding satellites to drive the satellites on the target route to execute the target service requests.

[0050] In practice, the more satellites there are, the lower the computing pressure on a single satellite will be, but the transmission delay will increase. The number of target routing satellites can be determined based on the task type of the target service request.

[0051] The computing resource control method provided in this embodiment obtains the target service request and its data features, then feeds these features into a prediction model to determine the target service request's task type. This determines the target computing power requirement for the target service request, identifies a route that satisfies the target computing power requirement, and then sends instructions to satellites on the route to instruct them to execute the target service request. This approach, which determines the target service request's task type based on its data features, improves judgment accuracy compared to directly using the target service request's inherent attributes as the task type, thereby enabling precise control of computing resources.

[0052] In this embodiment, a computing resource control method based on high-throughput satellite is provided. Figure 2 FIG. 1 is a flow chart of a method for controlling computing power resources according to an embodiment of the present invention. Figure 2 As shown, the process includes the following steps:

[0053] Step S201: Receive a target service request and obtain data features of the target service request.

[0054] For details, please see Figure 1 Step S101 of the illustrated embodiment will not be described in detail here.

[0055] Step S202: Send the data features to the prediction model to determine the task type of the target service request.

[0056] For details, please see Figure 1 Step S102 of the illustrated embodiment will not be described in detail here.

[0057] In some optional implementations, before step S202, the following steps are further included:

[0058] Step a1: Obtain a sample data set; the sample data set includes sample flow data and sample annotations.

[0059] The sample dataset can be obtained from public datasets, databases, etc. The sample dataset includes sample traffic data and sample annotations. The sample traffic information can be the traffic data of target data requests obtained at each historical moment; the sample annotations can be the result of manually evaluating the task type of the traffic data of target data requests at each historical moment.

[0060] Step a2: Cut the sample traffic data into time windows of target length, and input the time windows into the prediction model to obtain the classification results.

[0061] The time window parameters include the sliding interval, which can be set based on the needs of the prediction model. For an LSTM (Long Short-Term Memory) model, the target length must meet the model's ability to capture long-term dependencies, such as 100 time steps. The sliding interval can be set to 1, which means sliding point by point, maximizing data utilization and improving model prediction accuracy.

[0062] After receiving the time window, the prediction model will output the classification results. The classification results can include videos, files, sensors, etc.

[0063] In some optional implementations, before inputting the time window into the prediction model, the time window may be normalized to prevent numerical differences from affecting the convergence of the prediction model.

[0064] In step a3, based on the error between the classification result and the sample label, the prediction model is updated through the back propagation algorithm.

[0065] In an optional implementation, the prediction model is an LSTM model, and the error is obtained through a forward path, and then the error is back-propagated to correct the LSTM model.

[0066] In specific implementation, a single-layer LSTM is used, and the number of hidden units is reduced to 32 compared to 128 in the traditional model. Reducing the number of neurons in the fully connected layer can reduce the computing power required by the model.

[0067] Step S203: Calculate the target computing power requirement of the target service request based on the task type.

[0068] For details, please see Figure 1 Step S103 of the illustrated embodiment will not be described in detail here.

[0069] In some optional implementations, the above step S203 includes:

[0070] Step S2031-1, obtain several video transmission links of the video to be transmitted.

[0071] The video transmission process includes video encoding, channel encoding, and pre- and post-processing. Specifically, pre-processing occurs before video encoding and may include acquisition and pre-processing. Post-processing occurs after channel encoding and may include modulation, transmission, demodulation, signal decoding, video decoding, and display.

[0072] In actual applications, the computing power of video encoding accounts for 65-75%, the computing power of channel coding accounts for 15-20%, and the computing power of pre- / post-processing accounts for 8-12%.

[0073] Step S2031-2: Based on the coding standard of the video to be transmitted, determine the computing power requirements of each video transmission link to calculate the target computing power requirements of the target service request.

[0074] The encoding standard and encoding parameters of the video to be transmitted can be directly read through relevant media information tools. Among them, encoding parameters include resolution, frame rate, bit rate, etc.

[0075] During the encoding phase, computing power requirements can be calculated by multiplying the number of operations per pixel of the video to be transmitted, the resolution, the frame rate, and the encoding complexity factor. The number of operations per pixel and the complexity factor vary depending on the encoding standard, such as H.264 (Advanced Video Coding), H.265 (High Efficiency Video Coding), and AV1 (AOMedia Video 1). For example, the theoretical value for H.264 is approximately 100 to 300 operations per pixel; the theoretical value for H.265 is approximately 300 to 800 operations per pixel; and the theoretical value for AV1 is approximately 800 to 2000 operations per pixel. Based on a simple H.264 configuration, such as the Baseline Profile, and assuming a complexity factor of 1, the complexity factor for H.265 can be 2 to 3 (the algorithmic complexity is 2 to 3 times that of H.264). The complexity factor for AV1 can be 4 to 6 (the algorithmic complexity is higher than that of H.265).

[0076] The computing power required for channel coding is related to the degree of redundancy. For example, LDPC (Low Density Parity Check Code) coding with 5% redundancy requires approximately 0.1 GFLOPS / Gbps (billion floating-point operations per second / gigabit per second), while 20% redundancy requires 0.5 GFLOPS / Gbps.

[0077] The decoding computing power requirement can be obtained by calculating the product of the single iteration operation amount, the average number of iterations and the data rate, and then obtaining the ratio of the product to the code length.

[0078] Pre-processing computing power requirements typically account for 10% to 30% of the encoding computing power, while post-processing computing power requirements typically account for 5% to 20% of the decoding computing power. After obtaining the encoding computing power for video encoding and channel coding, the pre-processing computing power requirements can be calculated based on the computing power ratio. After obtaining the decoding computing power requirements, the post-processing computing power requirements can be calculated based on the computing power ratio.

[0079] In some optional implementations, the target computing power requirement for video transmission can be obtained by calculating the sum of the encoding requirement and the channel coding computing power, where the encoding requirement can be obtained by calculating the sum of video encoding, preprocessing, and post-processing. In practical applications, the encoding requirement can also be multiplied by a safety factor, where the safety factor is usually 1.2. The safety factor is used to set up redundant computing power to cope with unpredictable computing resource fluctuations, algorithm efficiency loss, hardware performance margin, and dynamic environmental changes in the actual system. Its function is to ensure that the system can still meet real-time requirements under the worst conditions.

[0080] In some optional implementations, the above step S203 includes:

[0081] Step S2032-1: Acquire several sensor transmission links of sensor data to be transmitted.

[0082] The sensor transmission process includes sensor data acquisition, data preprocessing, protocol encapsulation, channel coding, modulation, satellite transmission, and more. Data preprocessing can include data compression or encryption. In specific implementations, it also involves sensor energy management, namely, sensor sleep and wake-up.

[0083] Step S2032-2: Obtain the operation type and data volume of the sensor data to be transmitted to determine the computing power requirements of each sensor transmission link.

[0084] The computing power required for data preprocessing can be calculated by multiplying the sensor operation type and the amount of data to be transmitted. In a real-world application, if 100 bytes of sensor data are sent, the compression computing power is 0.1 FLOPS / Byte (floating-point operations per byte), and the AES (Advanced Encryption Standard) encryption computing power is 0.3 FLOPS / Byte, then the data preprocessing computing power can be calculated as (0.1 + 0.3) * 100 = 40 FLOPS (floating-point operations).

[0085] In specific implementation, data preprocessing mainly includes four types of operations: data compression, encryption, CRC (Cyclic Redundancy Check) check and data aggregation. The computing power reference value of data compression is 0.05-0.2FLOPS / Byte, and example scenarios include text / sensor data, namely ZIP (standard format for file compression and archiving), LZ4 (lossless data compression algorithm), etc.; the computing power reference value of encryption is 0.1-0.5FLOPS / Byte, and the example scenario is software-implemented AES-128 (Advanced Encryption Standard with a 128-bit key); the computing power reference value of CRC check is 0.01FLOPS / Byte, which is used for data integrity verification; the computing power reference value of data aggregation is 0.02FLOPS / Byte, and the example scenario is merging multiple sensor data packets.

[0086] The computing power requirements of the protocol encapsulation can be determined according to the coding type. The computing power requirements of different protocols are different. In specific implementations, the coding types of protocol encapsulation mainly include LDPC (Low Density Parity Check), Turbo (Turbo convolutional code) and Reed-Solomon (Reed-Solomon code). Among them, the computing power reference value of LDPC is 0.001-0.005FLOPS per bit, and the code rate is 1 / 2-3 / 4, which is suitable for high-reliability satellite communications; the computing power reference value of Turbo code is 0.002-0.008FLOPS per bit, and the code rate is 1 / 3-2 / 3, which is suitable for deep space communications with low signal-to-noise ratio; the computing power reference value of Reed-Solomon is 0.0005FLOPS per bit, with redundancy of 10%-30%, which is suitable for short data packet scenarios (such as sensors).

[0087] The computational power requirement for channel coding can be obtained by calculating the product of the amount of raw data, the computational power per bit, and the code ratio, where the computational power per bit and the code ratio depend on the coding type.

[0088] The computational power requirement for modulation overhead can be obtained by calculating the product of symbol rate and computation power per symbol.

[0089] Step S2032-3: Calculate the target computing power requirement of the target service request based on the computing power requirements of each sensor transmission link.

[0090] The target computing power requirement can be calculated by summing the coding power and modulation overhead. The coding power can be the sum of preprocessing power, protocol encapsulation power, and channel coding power. In some optional implementations, the coding power can be multiplied by a safety factor to improve the reliability of the method.

[0091] In actual applications, the computing power of data preprocessing (compression + encryption) accounts for 40-60%, the computing power of channel coding accounts for 20-30%, the computing power of protocol encapsulation accounts for 10-15%, and the computing power of modulation / demodulation accounts for 5-10%.

[0092] In some optional implementations, the above step S203 includes:

[0093] Step S2033-1, obtaining several file transmission links of the file data to be transmitted.

[0094] The file transmission process may include text generation, compression, encryption, protocol encapsulation, signal coding, modulation, and satellite transmission. The protocol encapsulation may also include ACK (Acknowledgment) confirmation and retransmission control.

[0095] Step S2033-2: Obtain the file protocol type of the file data to be transferred to determine the computing power requirements of each file transfer link.

[0096] Step S2033-3, based on the computing power requirements of each file transfer link, calculate the target computing power requirements of the target service request.

[0097] Please refer to steps S2032-2 and S2032-3 for details, which will not be repeated here.

[0098] In actual applications, for short texts less than 1KB (kilobyte), protocol encapsulation and channel coding account for approximately 50%-70% of the computing power. For long texts larger than 1KB, compression and encryption account for a higher proportion of computing power, approximately 60%-80%.

[0099] Step S204: determine a target route that meets the target computing power requirement, and send a target instruction to the satellite on the target route to drive the satellite on the target route to execute the target service request.

[0100] Specifically, the above step S204 includes:

[0101] Step S2041: Based on the task type, the delay requirement of the target service request is obtained to determine the routing parameters of the target route; the routing parameters include the number of satellites and the service requirements of a single satellite.

[0102] The latency requirements of the target service request can be determined based on the task type. For real-time data transmission, low latency is required to ensure real-time feedback of the operation. For non-real-time data transmission, the latency requirements are lower.

[0103] When the delay requirement is high, the number of satellites on the target route can be limited to reduce the delay of satellite data transmission.

[0104] When the latency requirement is low, a larger number of satellites can be selected to reduce the service demand of a single satellite.

[0105] In some optional implementations, the above step S2041 further includes:

[0106] Step b1, constructing a routing delay model; the routing delay model includes satellite data processing delay and routing transmission delay.

[0107] The routing delay model may be the sum of the satellite's data processing delay and the routing transmission delay.

[0108] In a practical application, the data processing delay parameter can be obtained according to the following formula:

[0109] T1=a+α*M / N

[0110] Where a is the fixed delay, representing the baseline processing time of the satellite when there is no load. α is the load factor, showing the linear growth of processing delay as load increases. M is the computing power required for the target service request, and N is the total computing power of the satellite. In a specific implementation, the fixed delay can be 5 milliseconds.

[0111] In a practical application, the routing transmission delay can be obtained by calculating the product of the number of hops and the delay of each hop, where the delay of each hop can be 10 milliseconds.

[0112] Step b2: Substitute the delay requirement of the target service request into the routing delay model to determine the routing parameters of the target route.

[0113] By substituting the latency requirements of the target service request into the routing delay model, we can obtain routing parameters that meet the latency requirements. Specifically, as the number of satellites increases, the computing pressure of a single satellite will decrease, but the transmission delay will increase. This can be used to determine the routing parameters of the target route.

[0114] Step S2042: Filter out a number of target satellites from the preset satellite set; the target satellites meet the service requirements of a single satellite.

[0115] A preset satellite set can be constructed by collecting satellite data from surrounding satellites. Satellite data may include satellite ID (Identity), satellite ephemeris information, remaining computing power, satellite bandwidth, CPU (Central Processing Unit) and / or GPU (Graphics Processing Unit) usage, etc.

[0116] According to the remaining computing power of each satellite in the preset satellite set, a satellite that meets the business needs of a single satellite is selected as the target satellite.

[0117] In some optional embodiments, the method further comprises:

[0118] Step c1: Based on a preset time interval, basic information of satellites in a preset satellite set is collected and updated.

[0119] Among them, the preset time interval can be set according to actual conditions, and the basic information may include the satellite's power, computing power reserve, CPU, GPU, illumination time and other information.

[0120] Step S2043: Establish a target route based on the routing parameters and the position of the target satellite.

[0121] Based on the location and remaining computing power of the target satellite, a shortest path can be selected as the target route according to the routing selection algorithm. The remaining computing power of all satellites on the route can meet the computing power requirements of the target service request.

[0122] In some optional implementations, step S204 further includes:

[0123] Step S205: After executing the target service request, the satellite on the target route generates a return result and sends the return result to the sender of the target service request.

[0124] After the satellite on the target route executes the target service request, it can generate a return result and then directly send the return result to the sender of the target service request, eliminating unnecessary computing power overhead.

[0125] The computing resource control method provided in this embodiment obtains the latency requirements of target service requests, determines the routing parameters of the target route, and selects the target satellite. The target route is then established based on the routing parameters and the location of the target satellite. This can disperse the computing pressure on the target satellite and improve the flexibility of target route selection.

[0126] In this embodiment, a satellite computing resource monitoring and scheduling platform is provided. Figure 3 It is a structural diagram of a satellite computing resource monitoring and scheduling platform according to an embodiment of the present invention.

[0127] The ground layer includes a visualization layer, through which satellite data such as the remaining computing power, power, and lighting conditions of each satellite in the entire constellation system can be intuitively displayed.

[0128] The low-orbit satellite layer includes a data collection layer. The scheduling platform collects data from low-orbit satellites, including the satellite ID, the ephemeris information of each satellite, the remaining computing power of each satellite, the satellite's bandwidth, and the CPU and GPU usage.

[0129] The high-orbit satellite layer includes a scheduling and execution layer, an analysis and decision-making layer, and a storage layer. The storage layer stores information collected by the data acquisition layer, including satellite IDs and each satellite's ephemeris information. Data is transmitted via inter-satellite links.

[0130] The analysis and decision-making layer is responsible for assessing service types, computing power requirements, and satellite energy supply. Specifically, when a low-orbit satellite receives a service request from the ground, it is called a service satellite and first sends the service request to the scheduling platform. After receiving the message from the service satellite, the scheduling platform is responsible for evaluating the service type, such as determining whether the service type is low-latency or delay-tolerant. It then assesses the total computing power required to complete the service. It calculates the remaining energy of the satellites in the set and their future energy supply, calculates the remaining computing power of the satellites, and selects the set of satellites that meet the computing power requirements for the service. Finally, a routing algorithm is used to calculate a list of satellites most suitable for the distributed computing service.

[0131] Traffic feature analysis within the implicit inference method is used to assess the type of service requested by the terminal. Features can include packet length distribution, transmission interval, traffic burstiness, and flow duration. Specifically, packet length distribution varies across different service types. Real-time video / voice traffic is delivered in large packets, close to the Maximum Transmission Unit (MTU), such as 1500 bytes, and is transmitted periodically. IoT sensor data is delivered in small packets, typically tens of bytes, transmitted at regular intervals, such as one packet every five seconds. File transfers are a mix of large and small packets, such as a mix of small TCP (Transmission Control Protocol) ACK packets and large data packets. Real-time services exhibit strict periodicity, such as a 33 millisecond video frame interval. Interactive services, such as SSH (Secure Shell), exhibit short bursts that can be triggered by user input. Background downloads are characterized by continuous, high-density data flows. Traffic burstiness quantifies traffic fluctuations using entropy or self-similarity. Video streams, influenced by differences between I-frames (Intra frames) and P-frames (Predicted frames), exhibit high burstiness. Sensor data, on the other hand, exhibits low burstiness and is reported steadily. Flow duration varies from short streams, such as DNS (Domain Name System) queries, measured in milliseconds, to long streams, such as video conferencing, which can last for hours.

[0132] The evaluation of the computing power demand evaluation mechanism may be performed as shown in step S103.

[0133] In some optional implementations, the platform also includes a dynamic resource monitoring mechanism. Resources include the satellite's remaining power, the satellite's remaining illumination duration and the interval until the next illumination, and the satellite's remaining computing power. The satellite's remaining power and computing power can be obtained by reading relevant information from the storage layer. Considering the limited storage capacity of low-orbit satellites, which are often charged by solar panels, the orbital period of a low-orbit satellite is typically 90 minutes, including 45 minutes of sunlight and 45 minutes of shadow. Therefore, the satellite's remaining illumination duration and the interval until the next illumination must be considered to avoid the inability to maintain normal service due to low battery levels.

[0134] Satellite energy supply analysis can be performed based on the computing power of different service types. After determining the service type, the corresponding computing power demand assessment formula is selected. If the computing power required for a communication service is M, the total computing power of each low-orbit satellite is N. In practice, resource searches can be performed with the service satellite as the center and eight (or more) surrounding satellites. The main search criteria include: remaining computing power resources n, remaining battery power, remaining satellite illumination time, and the interval until the next illumination. The selected number of satellites A must satisfy M / ((Nn)*A) < 70%, meaning that each satellite must maintain 30% of computing power remaining to cope with emergencies.

[0135] In a practical application, the satellite delay is shown in Table 1:

[0136] Number of satellites Hashrate pressure Processing delays Hop Count Transmission delay Total delay 3 65％ 7 2 20 27 5 50％ 6 4 40 46 7 30％ 5.5 6 60 65.5 ....

[0137] Table 1

[0138] As can be seen from the table above, as the number of satellites increases, the computing pressure of a single satellite will decrease, but the transmission delay will increase. Therefore, considering different business types, low-latency businesses can choose a 3-star solution, and delay-tolerant businesses can choose a 7-star solution.

[0139] The routing topology of the number of satellites is as follows Figure 4 As shown, the blue satellite in the center is the service satellite, and the surrounding red satellites are neighboring satellites.

[0140] The scheduling and execution layer is responsible for distributing computing tasks. It distributes computing instructions to satellites in the route. After executing the computing instructions, the satellites directly send the results to the service satellite.

[0141] In this embodiment, a computing resource control device based on a high-throughput satellite is also provided. The device is used to implement the above-mentioned embodiments and preferred implementation methods. The details that have been described will not be repeated here. As used below, the term "module" can be a combination of software and / or hardware that implements a predetermined function. Although the devices described in the following embodiments are preferably implemented in software, implementation in hardware, or a combination of software and hardware, is also possible and conceivable.

[0142] This embodiment provides a computing resource control device based on high-throughput satellites, such as Figure 5 Shown, including:

[0143] Receiving module 501, used to receive a target service request and obtain data characteristics of the target service request;

[0144] Determination module 502, for sending data features to the prediction model to determine the task type of the target service request;

[0145] A calculation module 503 is used to calculate the target computing power requirement of the target service request based on the task type;

[0146] The execution module 504 is used to determine a target route that meets the target computing power requirement and send a target instruction to the satellite on the target route to drive the satellite on the target route to execute the target service request.

[0147] In some optional implementations, the computing resource control device includes a model training module, which includes:

[0148] The data set acquisition unit is used to acquire a sample data set; the sample data set includes sample flow data and sample annotations.

[0149] The data cutting unit is used to cut the sample traffic data into time windows of target length, and input the time windows into the prediction model to obtain classification results.

[0150] The model updating unit is used to update the prediction model through the back propagation algorithm based on the error between the classification result and the sample label.

[0151] In some optional implementations, the calculation module 503 includes:

[0152] A video transmission link acquisition unit, used to acquire several video transmission links of the video to be transmitted;

[0153] The video computing power calculation unit is used to determine the computing power requirements of each video transmission link based on the encoding standard of the video to be transmitted, so as to calculate the target computing power requirements of the target service request.

[0154] In some optional implementations, the calculation module 503 includes:

[0155] A sensor transmission link acquisition unit, used to acquire a plurality of sensor transmission links of sensor data to be transmitted;

[0156] A sensor transmission link computing power requirement determination unit is used to obtain the operation type and data volume of the sensor data to be transmitted; so as to determine the computing power requirement of each sensor transmission link;

[0157] The sensor computing power calculation unit is used to calculate the target computing power requirements of the target service request based on the computing power requirements of each sensor transmission link.

[0158] In some optional implementations, the calculation module 503 includes:

[0159] A file transmission link acquisition unit, used to acquire several file transmission links of the file data to be transmitted;

[0160] A file transfer link computing power requirement determination unit is used to obtain the file protocol type of the file data to be transferred to determine the computing power requirement of each file transfer link;

[0161] The file computing power calculation unit is used to calculate the target computing power requirements of the target service request based on the computing power requirements of each file transfer link.

[0162] In some optional implementations, the execution module 504 includes:

[0163] The routing parameter determination unit is used to obtain the delay requirement of the target service request based on the task type to determine the routing parameters of the target route; the routing parameters include the number of satellites and the service requirements of a single satellite.

[0164] A screening unit is used to screen out a number of target satellites from a preset satellite set; the target satellites meet the service requirements of a single satellite;

[0165] The route establishing unit is used to establish a target route based on the route parameters and the position of the target satellite.

[0166] In some optional implementations, the routing parameter determination unit includes:

[0167] The delay model subunit is used to construct a routing delay model; the routing delay model includes the satellite's data processing delay and routing transmission delay;

[0168] The routing parameter determination subunit is used to substitute the delay requirement of the target service request into the routing delay model to determine the routing parameters of the target route.

[0169] In some optional implementations, the computing resource control device includes:

[0170] The satellite information update module is used to collect and update the basic information of satellites in the preset satellite set based on a preset time interval.

[0171] In some optional implementations, the computing resource control device includes:

[0172] The return module is used to generate a return result after the satellite on the target route executes the target service request, and send the return result to the issuing end of the target service request.

[0173] The further functional description of each of the above modules and units is the same as that of the above corresponding embodiments and will not be repeated here.

[0174] The computing power resource control device in this embodiment is presented in the form of a functional unit, where the unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that executes one or more software or fixed programs, and / or other devices that can provide the above functions.

[0175] The embodiment of the present invention also provides a computer device having the above Figure 5 The computing power resource control device shown.

[0176] See also Figure 6 , Figure 6 is a structural diagram of a computer device provided by an optional embodiment of the present invention, such as Figure 6 As shown, the computer device includes: one or more processors 10, memory 20, and interfaces for connecting various components, including high-speed interfaces and low-speed interfaces. Various components utilize different buses to communicate with each other and can be installed on a common mainboard or installed in other ways as needed. The processor can process the instructions executed in the computer device, including instructions stored in the memory or on the memory to display the graphical information of the GUI on an external input / output device (such as, a display device coupled to the interface). In some optional embodiments, if necessary, multiple processors and / or multiple buses can be used together with multiple memories and multiple memories. Equally, multiple computer devices can be connected, and each device provides part of the necessary operations (for example, as a server array, a group of blade servers, or a multi-processor system). Figure 6 A processor 10 is taken as an example.

[0177] The processor 10 may be a central processing unit, a network processor, or a combination thereof. The processor 10 may further include a hardware chip. The hardware chip may be an application-specific integrated circuit, a programmable logic device, or a combination thereof. The programmable logic device may be a complex programmable logic device, a field programmable gate array, a general purpose array logic, or any combination thereof.

[0178] The memory 20 stores instructions that can be executed by at least one processor 10, so as to enable at least one processor 10 to execute the method shown in the above embodiment.

[0179] The memory 20 may include a program storage area and a data storage area, wherein the program storage area may store an operating system and application programs required for at least one function; the data storage area may store data created based on the use of the computer device, etc. In addition, the memory 20 may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage device. In some optional embodiments, the memory 20 may optionally include a memory remotely located relative to the processor 10, and these remote memories may be connected to the computer device via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0180] The memory 20 may include a volatile memory, such as a random access memory; the memory may also include a non-volatile memory, such as a flash memory, a hard disk or a solid-state drive; the memory 20 may also include a combination of the above types of memory.

[0181] The computer device further includes an input device 30 and an output device 40. The processor 10, the memory 20, the input device 30 and the output device 40 may be connected via a bus or other means. Figure 6 The bus connection is taken as an example.

[0182] The input device 30 can receive input digital or character information and generate key signal input related to user settings and function control of the computer device, such as a touch screen, a keypad, a mouse, a trackpad, a touch pad, an indicator stick, one or more mouse buttons, a trackball, a joystick, etc. The output device 40 can include a display device, an auxiliary lighting device (e.g., an LED), and a tactile feedback device (e.g., a vibration motor). The above-mentioned display device includes but is not limited to a liquid crystal display, a light emitting diode, a display, and a plasma display. In some optional embodiments, the display device can be a touch screen.

[0183] The embodiment of the present invention also provides a computer-readable storage medium. The above-mentioned method according to the embodiment of the present invention can be implemented in hardware, firmware, or implemented as a computer code that can be recorded in a storage medium, or implemented as a computer code that is originally stored in a remote storage medium or a non-temporary machine-readable storage medium and downloaded through a network and will be stored in a local storage medium, so that the method described herein can be stored in such software processing on a storage medium using a general-purpose computer, a dedicated processor, or programmable or dedicated hardware. Among them, the storage medium can be a magnetic disk, an optical disk, a read-only storage memory, a random access memory, a flash memory, a hard disk or a solid-state drive, etc.; further, the storage medium can also include a combination of the above-mentioned types of memory. It can be understood that a computer, a processor, a microprocessor controller or programmable hardware includes a storage component that can store or receive software or computer code. When the software or computer code is accessed and executed by a computer, a processor or hardware, the method shown in the above embodiment is implemented.

[0184] A portion of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the form in which the computer program instruction exists in a computer-readable medium includes, but is not limited to, a source file, an executable file, an installation package file, etc. Accordingly, the way in which the computer program instruction is executed by the computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium that can be accessed by the computer.

[0185] Although the embodiments of the present invention have been described with reference to the accompanying drawings, those skilled in the art may make various modifications and variations without departing from the spirit and scope of the present invention. Such modifications and variations are all within the scope defined by the appended claims.

Claims

1. A method for controlling computing resources based on high-throughput satellites, characterized in that: The method comprises: Receive a target service request and obtain data characteristics of the target service request; Sending the data features to a prediction model to determine the task type of the target service request; Based on the task type, calculating the target computing power requirement of the target service request; Determine a target route that meets the target computing power requirement, and send a target instruction to the satellite on the target route to drive the satellite on the target route to execute the target service request.

2. The method according to claim 1, characterized in that The method further comprises: Obtaining a sample data set; the sample data set includes sample flow data and sample annotations; Cutting the sample flow data into time windows of target length, and inputting the time windows into the prediction model to obtain classification results; Based on the error between the classification result and the sample annotation, the prediction model is updated through a back propagation algorithm.

3. The method according to claim 1, characterized in that In a case where the task type is video transmission, the calculating the target computing power requirement of the target service request based on the task type includes: Obtaining several video transmission links of the video to be transmitted; Based on the coding standard of the video to be transmitted, the computing power requirements of each video transmission link are determined to calculate the target computing power requirements of the target service request.

4. The method according to claim 1, wherein In a case where the task type is sensor data transmission, calculating the target computing power requirement of the target service request based on the task type includes: Acquire several sensor transmission links of sensor data to be transmitted; Obtaining the operation type and data volume of the sensor data to be transmitted; to determine the computing power requirements of each sensor transmission link; Based on the computing power requirements of the transmission links of each sensor, the target computing power requirement of the target service request is calculated.

5. The method according to claim 1, characterized in that In a case where the task type is file data transmission, calculating the target computing power requirement of the target service request based on the task type includes: Obtain several file transfer links of file data to be transferred; Obtaining the file protocol type of the file data to be transferred to determine the computing power requirements of each file transfer link; Based on the computing power requirements of the various file transfer links, the target computing power requirements of the target service request are calculated.

6. The method according to claim 1, characterized in that Determining a target route that meets the target computing power requirement includes: Based on the task type, obtaining the delay requirement of the target service request to determine the routing parameters of the target route; the routing parameters include the number of satellites and the service requirements of a single satellite; Selecting a plurality of target satellites from a preset satellite set; wherein the target satellites meet the service requirements of the single satellite; A target route is established based on the routing parameters and the position of the target satellite.

7. The method according to claim 6, characterized in that The determining of the routing parameters of the target route includes: Constructing a routing delay model; the routing delay model includes satellite data processing delay and routing transmission delay; The delay requirement of the target service request is substituted into the routing delay model to determine the routing parameters of the target route.

8. The method according to claim 1, characterized in that The method further comprises: Based on the preset time interval, the basic information of the satellites in the preset satellite set is collected and updated.

9. The method according to claim 1, characterized in that After executing the target service request, the satellite on the target route generates a return result and sends the return result to the sending end of the target service request.

10. A computing resource control device based on high-throughput satellite, characterized in that: The device comprises: A receiving module, configured to receive a target service request and obtain data characteristics of the target service request; A determination module, configured to send the data features to a prediction model to determine the task type of the target service request; A calculation module, configured to calculate a target computing power requirement of the target service request based on the task type; The execution module is used to determine a target route that meets the target computing power requirement and send a target instruction to the satellite on the target route to drive the satellite on the target route to execute the target service request.

Images