A spaceborne value-driven data lifecycle management architecture and method

Abstract

The application provides a kind of spaceborne value-driven data life cycle management architecture, including multidimensional value evaluator, it is configured to assess the timeliness of raw data package, scarcity, task correlation, commercial value and scientific value in orbit, and generate integrated value score;Data life cycle manager, it is configured to compare the integrated value score of multidimensional value evaluator output received and decision threshold;And automatically execute one of at least two life cycle paths predefined;Cross-satellite value negotiation protocol engine, data fingerprint and value negotiation message are exchanged by inter-satellite link and adjacent satellite;Resource state perception and threshold self-adapting module, it is configured to monitor the storage space margin on satellite, computing load, communication window state and current task mode;Ground interface and model updating module, it is configured to receive the updated value evaluation model, task priority instruction and user request on ground station.

Description

Technical Field

[0001] This invention relates to the fields of satellite electronic systems and onboard data processing technology. Specifically, it relates to an onboard data lifecycle management architecture and its implementation method for satellite platforms (including single satellites and satellite constellations), driven by data value. This invention can be used on various satellite platforms such as remote sensing satellites, communication satellites, and scientific exploration satellites, and is particularly suitable for resource-constrained low-Earth orbit satellite constellation scenarios. Background Technology

[0002] With the rapid development of LEO mega-constellation technology, the data acquisition capability of a single satellite has increased exponentially. Taking high-resolution remote sensing satellites as an example, a single satellite can generate several terabytes of raw observation data daily. However, the on-board resources of the satellite platform (storage capacity, computing power, and communication bandwidth) are subject to strict constraints on size, weight, and power consumption (SWaP), creating an increasingly acute contradiction between "data deluge vs. resource bottleneck".

[0003] Currently, most satellites adopt a simple pipeline mode of "acquisition-storage-download": all data collected by the sensors are stored indiscriminately in a large-capacity storage device, and are downloaded in batches to the ground station through the data transmission channel when passing over the ground, and are then processed and analyzed by the ground system. This mode has the following prominent problems: (1) Serious waste of resources: a large amount of low-value or redundant data (such as cloud cover images and repeated observation data) occupy valuable on-board storage space and download bandwidth. (2) Delay in emergency response: high-value emergency data (such as observation results of disaster areas) need to queue up and wait for the download window, and cannot be delivered to ground users in a timely manner. (3) Lack of autonomous decision-making ability: satellites do not have the ability to judge the value of data in orbit, and rely entirely on the post-screening of the ground, which makes it impossible to dynamically optimize the allocation of on-board resources according to mission requirements. (4) Insufficient satellite group coordination: when multiple satellites observe the same area, there is a large amount of duplicate data, and there is a lack of value negotiation mechanism between satellites, making it impossible to achieve data deduplication and mission collaborative optimization in orbit.

[0004] The shortcomings of existing technologies in addressing this problem include: (1) Traditional data compression methods (such as JPEG2000 and CCSDS compression standards) only focus on reducing the amount of data, without involving the evaluation and screening of data value dimensions, and cannot distinguish between high-value and low-value data.

[0005] (2) Existing on-board mission scheduling methods (such as CN118331591B and US20220374262A1) focus on the allocation and optimization of computing resources, and do not establish a value-driven joint decision-making mechanism for storage, computing and transmission from the perspective of the entire data life cycle.

[0006] (3) On-board semantic communication technologies (such as CN119380729A and CN121217213B) focus on semantic compression in the transmission link, but are limited to the single link of "data transmission" and have not built a unified value assessment and decision-making framework covering the entire link of storage, computing and transmission.

[0007] (4) Existing star cluster collaboration methods (such as the blockchain task allocation of CN115510250B) focus on the recording and allocation of tasks, and do not provide a lightweight cross-star value negotiation protocol for data value exchange and collaborative decision-making between stars.

[0008] In conclusion, the satellite industry urgently needs an architectural solution that can fundamentally resolve the contradiction between "data overload and resource constraints"—a solution that should enable satellites to autonomously assess data value in orbit, make joint storage-computation-transmission decisions based on value, and maximize the value of constellation-level data through inter-satellite negotiation. Summary of the Invention

[0009] In view of this, the present invention provides a spaceborne value-driven data lifecycle management architecture, characterized in that it includes: A multidimensional value evaluator is configured to evaluate the timeliness, scarcity, mission relevance, commercial value, and scientific value of raw data packets in orbit and generate a comprehensive value score. A data lifecycle manager is configured to receive a comprehensive value score output by a multidimensional value evaluator; compare the value score with a decision threshold; and automatically execute one of at least two predefined lifecycle paths. The cross-satellite value negotiation protocol engine exchanges data fingerprints and value negotiation messages with neighboring satellites through inter-satellite links to perform data deduplication and value collaboration at the constellation level. The resource status awareness and threshold adaptive adjustment module is configured to monitor on-board storage space availability, computing load, communication window status and current task mode, as well as adjust the evaluation weights of the multidimensional value evaluator and the decision thresholds of the data lifecycle manager. The ground interface and model update module is configured to receive updated value assessment models, task priority instructions, and user requests from the ground station to perform data lifecycle iterations.

[0010] The formula for calculating the comprehensive value score by the multidimensional value assessor is as follows: in, For timeliness value function, This represents the time elapsed since the data was collected. The attenuation rate constant; The result is calculated based on the number of matches between data fingerprints and on-board historical data. The lower the number of matches, the higher the scarcity. The result is calculated using the cosine similarity between the data semantic labels and the active task target vector; For predicting the economic value of data based on ground market models; The result is calculated based on the difference between the data content and the distribution of historical data; As a weighting factor for timeliness assessment, As a weight for scarcity assessment, As the weight of task relevance, As a measure of commercial value, Weighting based on scientific value.

[0011] The data lifecycle manager maintains the relationship between four decision thresholds, namely, the first threshold is greater than the second threshold, which is greater than the third threshold, which is greater than the fourth threshold; the lifecycle path includes: When the comprehensive value score is greater than or equal to the first threshold, the lifecycle path is the first path, that is, the original data packet is immediately transmitted to the ground station via the Ka-band link or the laser link; When the comprehensive value score is less than the first threshold and greater than or equal to the second threshold, the lifecycle path is the second path, that is, the original data packet is deeply processed on the satellite and then transmitted to the ground station. When the comprehensive value score is less than the second threshold and greater than or equal to the third threshold, the lifecycle path is the third path, that is, the original data packet is compressed without loss or near loss and stored in the on-board storage module, waiting for the regular downlink window to be downlinked to the ground station. When the comprehensive value score is less than the third threshold and greater than or equal to the fourth threshold, the lifecycle path is the fourth path, that is, the original data packet is discarded after generating the thumbnail or summary metadata. When the comprehensive value score is less than the fourth threshold, the lifecycle path becomes the fifth path, i.e., the original data packet is discarded.

[0012] The deep processing includes at least one of the following: AI inference, feature extraction, and data fusion.

[0013] The cross-satellite value negotiation protocol engine is configured to perform negotiation through the following process: Step one: The data holding satellite calculates the fingerprint of the newly acquired data and broadcasts it to neighboring satellites; Step 2: After receiving the data fingerprint, the neighboring satellites match and search the local satellite's data fingerprint database. If the same or highly similar data is found, an interest confirmation message is sent along with the local satellite's existing value assessment of the data. Step 3: After receiving the interest confirmation message, the data holding satellite sends data metadata to neighboring satellites, including the size, type and timeliness of the data, to provide a basis for value assessment. Step four: The data-holding satellite and neighboring satellites exchange their respective value assessments of the data, and obtain a consensus value score through a negotiation algorithm; Step 5: Based on the consensus value score, both parties adjust their respective lifecycle path decisions; if one party has already stored the same data, the other party selects the fourth or fifth path to process the data in order to perform data deduplication at the constellation level. Step six: Record the consensus value score in the satellite's distributed ledger for use by the ground station for auditing and commercial settlement; Step 7: After receiving and confirming the consensus value score record, the ground station sends confirmation and scheduling instructions to the data holding satellite to guide the data distribution task and complete the closed loop of the entire negotiation and scheduling.

[0014] The data fingerprint generation method includes at least one of the following: hash algorithm, semantic feature vector extraction, and spatiotemporal bounding box cross method.

[0015] The evaluation weight adjustment rule is as follows: in emergency mode, the task relevance weight is automatically increased to 0.4-0.5; in energy-saving mode, the scarcity evaluation weight and scientific value weight are reduced to reduce computational overhead.

[0016] The decision threshold adjustment rule is that when on-board storage space is tight, the third and fourth thresholds are raised so that more data is directed to the fourth or fifth path.

[0017] When a specific event signal is detected, the system switches from normal mode to emergency mode, triggering a reconfiguration of evaluation weights and decision thresholds; the specific event includes at least one of the following: earthquake early warning and military alert.

[0018] A method for managing the lifecycle of spaceborne value-driven data, characterized by comprising the following steps: The multidimensional value evaluator performs a multidimensional value evaluation on the input data packet and generates a comprehensive value score; The comprehensive value score is received by the data lifecycle manager; The data lifecycle manager selects to execute at least one of two lifecycle paths based on a comparison of the overall value score and the decision threshold. The cross-satellite value negotiation protocol engine interacts with other satellites via inter-satellite links and feeds back data fingerprints and value negotiation results to the multi-dimensional value evaluator and data lifecycle manager to perform collaborative optimization of the various modules of the data lifecycle management architecture. The resource status perception and threshold adaptive adjustment module continuously monitors the satellite resource status and adjusts the evaluation weight of the multidimensional value evaluator and the decision threshold of the data lifecycle manager to form a closed-loop adaptive adjustment for resource perception. The ground interface and model update module receive value assessment model updates, task priority instructions, and user requests from the ground via the satellite-to-ground link to perform data lifecycle iterations.

[0019] The present invention has the following beneficial effects: (1) This invention improves the effective utilization rate of on-board storage space by 40-60% through value-driven joint decision-making of storage-computation-transmission, and focuses on high-value data processing on-board computing resources, thus avoiding the waste of computing power.

[0020] (2) Under the same downlink bandwidth conditions, the present invention enables more valuable information to be transmitted within a limited downlink window by prioritizing the downlink of high-value data and removing unnecessary data on track.

[0021] (3) This invention enables high-value emergency data to be transmitted instantly through the first path, reducing the response delay from the traditional hour level to the second level, which is particularly suitable for time-sensitive scenarios such as disaster monitoring and military reconnaissance.

[0022] (4) This invention proposes an on-orbit real-time evaluation model based on multidimensional value, which significantly reduces the time occupied by ground stations, improves the autonomous operation capability of the constellation, and reduces the dependence on the ground station network and the operating manpower cost.

[0023] (5) This invention provides a cross-star value negotiation protocol, which enables the data deduplication rate within the constellation to reach 30-60%, avoids the waste of resources from repeated observation of the same area by multiple stars, and improves the overall data acquisition efficiency and information coverage of the constellation.

[0024] (6) The architecture proposed in this invention is a "double-independent" architecture of mission mode and data type, which can be flexibly adapted to different mission scenarios (remote sensing, communication, scientific exploration) and different satellite platforms (microsatellites, large satellites, constellations), and has broad application prospects. Attached Figure Description

[0025] To further illustrate the above and other advantages and features of the various embodiments of the present invention, a more specific description of the various embodiments of the present invention will be presented with reference to the accompanying drawings. It is to be understood that these drawings depict only typical embodiments of the invention and are therefore not intended to limit its scope. In the drawings, identical or corresponding parts will be indicated by identical or similar reference numerals for clarity.

[0026] Figure 1 This is the overall architecture diagram of the spaceborne value-driven data lifecycle management system of the present invention.

[0027] Figure 2 This is a flowchart of the multidimensional value assessment and path decision-making process for data in this invention.

[0028] Figure 3 This is the data lifecycle path decision tree of the present invention.

[0029] Figure 4 This is a timing diagram of the inter-satellite value negotiation protocol of the present invention.

[0030] Figure 5 This is a deployment diagram of the remote sensing constellation value-driven data processing system of the present invention. Detailed Implementation

[0031] It should be noted that the components in the various figures may be shown exaggeratedly for illustrative purposes and are not necessarily to scale. In each figure, the same reference numerals are used for components that are identical or have the same function.

[0032] In this invention, unless otherwise specified, "arranged on," "arranged above," and "arranged on" do not exclude the possibility of an intermediate element between them. Furthermore, "arranged on or above" merely indicates the relative positional relationship between two components, and in certain cases, such as when the product orientation is reversed, it can also be converted to "arranged below or under," and vice versa.

[0033] In this invention, the various embodiments are merely intended to illustrate the solutions of the invention and should not be construed as limiting.

[0034] In this invention, unless otherwise specified, the quantifiers “a” and “one” do not exclude scenarios involving multiple elements.

[0035] It should also be noted that, in the embodiments of the present invention, only a portion of the components or parts may be shown for clarity and simplicity. However, those skilled in the art will understand that, under the teachings of the present invention, necessary components or parts can be added as needed for specific scenarios. Furthermore, unless otherwise stated, features in different embodiments of the present invention can be combined with each other. For example, a feature in the second embodiment can replace a corresponding or functionally identical or similar feature in the first embodiment, and the resulting embodiment will also fall within the scope of disclosure or description of this application.

[0036] It should also be noted that, within the scope of this invention, the terms "same," "equal," and "equal to" do not imply that the two values ​​are absolutely equal, but rather allow for a certain reasonable margin of error. In other words, the terms also encompass "substantially the same," "substantially equal," and "substantially equal to." Similarly, in this invention, the directional terms "perpendicular to," "parallel to," etc., also encompass the meanings of "substantially perpendicular to" and "substantially parallel to."

[0037] Furthermore, the numbering of the steps in the methods of the present invention does not limit the execution order of the method steps. Unless otherwise specified, the method steps may be executed in different orders.

[0038] In this invention, the modules of the system according to the invention can be implemented using software, hardware, firmware, or a combination thereof. When a module is implemented using software, its function can be implemented through computer program flow. For example, the module can be implemented using code segments (such as code segments in languages ​​like C and C++) stored in a storage device (such as a hard disk, memory, etc.), wherein the corresponding function of the module can be implemented when the code segment is executed by a processor. When a module is implemented using hardware, its function can be implemented by setting a corresponding hardware structure. For example, the module's function can be implemented by hardware programming a programmable device such as a field-programmable gate array (FPGA), or by designing an application-specific integrated circuit (ASIC) that includes multiple transistors, resistors, capacitors, and other electronic devices. When a module is implemented using firmware, the module's function can be written into a read-only memory such as an EPROM or EEPROM in the form of program code, and the corresponding function of the module can be implemented when the program code is executed by a processor. In addition, some functions of the module may need to be implemented by separate hardware or in cooperation with the hardware. For example, the detection function is implemented by the corresponding sensor (such as a proximity sensor, accelerometer, gyroscope, etc.), the signal transmission function is implemented by the corresponding communication device (such as Bluetooth device, infrared communication device, baseband communication device, Wi-Fi communication device, etc.), the output function is implemented by the corresponding output device (such as display, speaker, etc.), and so on.

[0039] The present invention will be further described below with reference to specific embodiments and the accompanying drawings.

[0040] Figure 1 This is a diagram illustrating the overall architecture of the spaceborne value-driven data lifecycle management system of this invention. Figure 1As shown, the Value-Driven Data Lifecycle Management Architecture (V-DLMA) is deployed on a satellite platform and includes the following core functional units: A Multi-Dimensional Value Assessor (MDVA) 210, configured to evaluate the timeliness, scarcity, mission relevance, commercial value, and scientific value of data packets in orbit and generate a comprehensive value score; a Data Lifecycle Manager (DLM) 220, configured to receive the comprehensive value score output by the MDVA; compare the value score with a threshold; and automatically execute one of several predefined lifecycle paths; and a Cross-Satellite Value Negotiation Protocol Engine (CSVN-PE) 230, which exchanges data fingerprints and value negotiation messages with neighboring satellites via inter-satellite links to perform data deduplication and value collaboration at the constellation level. The Resource-Aware Threshold Adaptor (RATA) module 240 is configured to monitor the status information of the satellite resource layer 300, including the storage capacity of the onboard storage unit 310, the computing load of the onboard computing unit 320, the communication window status of the inter-satellite / satellite-to-ground communication unit 330, and the current task mode identified by the sensor 340, and to dynamically adjust the evaluation weights of the multi-dimensional value evaluator 210 and the decision thresholds of the data lifecycle manager 220. The Ground Interface & Model Update (GIMU) module 250 is configured to receive value evaluation model updates from the ground model training terminal, task priority instructions from the ground operations center, and user requests from the user terminal to execute iterative strategies for satellite-to-ground collaboration. The ground station network 110, the ground operations center 120, the user terminal 130, and the ground model training terminal 140 together constitute the ground segment 100 of this architecture.

[0041] This embodiment uses a hash algorithm to generate data fingerprints. For situations involving multiple sensors or different viewpoints, ordinary hashing may fail to identify "similar but different" data. In such cases, semantic feature vector extraction can be used. This involves using an autoencoder or CNN to extract semantic feature vectors from the image data (Embedding) and calculating the cosine similarity between the vectors. If the similarity exceeds a set threshold (e.g., 90%), the data is considered highly overlapping. Alternatively, a spatiotemporal bounding box cross-validation method (a lightweight alternative) can be used. For nanosatellites with extremely limited computing resources, image hashing is not calculated; instead, only "observation time window + latitude and longitude polygon bounding boxes" are broadcast. If the IoU (Intersection over Union) of the bounding boxes of two satellites exceeds a set threshold and the time interval is extremely short, the data is considered homogeneous, triggering deduplication negotiation.

[0042] Based on the above modules, the spaceborne value-driven data lifecycle management method includes the following steps: The multidimensional value evaluator 210 performs multidimensional value evaluation on the input data packet and generates a comprehensive value score; The comprehensive value score is received by the data lifecycle manager 220; The data lifecycle manager 220 selects one of five lifecycle paths to execute based on a comparison of the comprehensive value score and the decision threshold. The cross-satellite value negotiation protocol engine 230 interacts with other satellites via inter-satellite links and feeds back the data fingerprint and value negotiation results to the multi-dimensional value evaluator 210 and the data lifecycle manager 220 to perform collaborative optimization of the various modules of the data lifecycle management architecture. The resource status perception and threshold adaptive adjustment module 240 continuously monitors the satellite resource status and adjusts the evaluation weight of the multi-dimensional value evaluator 210 and the decision threshold of the data lifecycle manager 220 to form a closed-loop adaptive adjustment for resource perception. The ground interface and model update module 250 receive value assessment model updates, mission priority instructions, and user requests from the ground via the satellite-ground link to perform data lifecycle iterations.

[0043] Figure 2 This is a flowchart illustrating the multidimensional value assessment and path decision-making process for data in this invention. Figure 2 As shown, MDVA evaluates each data packet based on its timeliness, scarcity, task relevance, commercial value, and scientific value, calculating its comprehensive value score. The calculation formula is: in, For timeliness value function, This represents the time elapsed since the data was collected. The attenuation rate constant; The result is calculated based on the number of matches between data fingerprints and on-board historical data. The lower the number of matches, the higher the scarcity. The result is calculated using the cosine similarity between the data semantic labels and the active task target vector; For predicting the economic value of data based on ground market models; The result is calculated based on the difference between the data content and the distribution of historical data; As a weighting factor for timeliness assessment, As a weight for scarcity assessment, As the weight of task relevance, As a measure of commercial value, Weighting based on scientific value. The evaluation weights are dynamically and adaptively adjusted by RATA based on the task mode. The adjustment rule for the evaluation weights is as follows: In emergency mode, Automatically increases to 0.4-0.5 to ensure that data with high task relevance is prioritized. In energy-saving mode, and Reduce appropriately to decrease computational overhead.

[0044] This embodiment uses a linear weighted formula to calculate the comprehensive value score. In other embodiments, a machine learning-based nonlinear evaluation method can be used. Specifically, MDVA can be replaced with a pre-trained lightweight neural network (such as a decision tree, random forest, or lightweight MLP model), with a five-dimensional feature vector as input and a value probability value between 0 and 1 as output. Alternatively, a reinforcement learning agent can be used, introducing deep reinforcement learning (DRL). This uses "maximizing the total value of the downlinked data" as the reward function, allowing the onboard agent to automatically output action strategies based on the current environmental state, rather than relying on fixed empirical weights. .

[0045] Figure 3 This is the data lifecycle path decision tree of the present invention. For example... Figure 3 As shown, DLM maintains the relationship between four dynamic decision thresholds, namely the first threshold. And determine the data processing path based on the comprehensive value score S: when the comprehensive value score is greater than or equal to In this case, the lifecycle path is the first path, meaning the original data packet is immediately transmitted to the ground station via a high-priority Ka-band link or laser link. This path consumes transmission bandwidth, has a latency in the order of seconds, and retains 100% of the data, making it suitable for extremely high-value data such as disaster monitoring and military targets. When the comprehensive value score is less than... And greater than or equal to In this case, the lifecycle path is the second path, where the original data packet undergoes deep processing on-board and is then transmitted to the ground station. This path consumes on-board computing power, with a data retention rate of 10%-30%, and is suitable for scenarios requiring rapid output of analysis results, such as AI extraction or feature processing. When the comprehensive value score is less than... And greater than or equal to In this case, the lifecycle path is the third path, where the original data packet is compressed losslessly or near-losslessly and stored in the onboard high-capacity storage module, awaiting download to the ground station during the regular download window. This path only consumes onboard storage space and is suitable for routine observation data. When the comprehensive value score is less than... And greater than or equal to In this case, the lifecycle path is the fourth path, which involves discarding the original data package after generating thumbnails or summary metadata. This path has extremely low resource consumption and is suitable for scenarios where the index needs to be retained but the value of the original data is limited. When the comprehensive value score is less than... In this case, the lifecycle path is the fifth path, which involves discarding the original data packets. This path consumes no resources and is suitable for valueless data such as cloud-covered images and invalid observations, thus saving onboard resources.

[0046] The threshold It is not a fixed value, but rather dynamically adjusted by RATA based on onboard resource status and mission mode. For example, when storage space is scarce, and Automatic ascent directs more data to the fourth or fifth path. The onboard resource status includes storage utilization. Calculation load Download window remaining time The current task mode M ∈ {Normal, Emergency, Energy Saving}. Based on these indicators, RATA dynamically adjusts the evaluation weights of the multidimensional value evaluator and the decision thresholds of the data lifecycle manager. Furthermore, RATA can introduce an "energy and orbital position prediction model." The system can combine satellite ephemeris data to calculate the length of time it will be in the shadow region (without solar charging). If it is about to enter a long shadow region, the system proactively reduces high-power processing in advance (reducing the probability of triggering the second path, lowering...). (To force more data to sleep or be discarded), enabling forward-looking power consumption optimization based on orbital illumination cycles.

[0047] The decision threshold adjustment rule is as follows: In other words, the more scarce the storage, the higher the threshold for emergency download, and only the most valuable data triggers immediate download. When storage is strained, even more data is discarded directly. For storage utilization The baseline threshold at that time The storage sensitivity coefficient serves as the threshold for emergency downloads, controlling the extent to which storage utilization raises the emergency download threshold. The baseline threshold for data discarding. The storage sensitivity coefficient serves as the threshold for data discarding, and is used for dynamic adjustment. To release on-board storage.

[0048] When a specific event signal is detected, such as an earthquake early warning or a military alarm, the system automatically switches from normal mode to emergency mode, triggering a rapid reconfiguration of weights and thresholds.

[0049] In addition, regarding the comprehensive value score The commercial value component in the calculation formula and weight The ground system provides a data request bidding API. When there is a sudden business need (such as a financial institution urgently needing real-time cargo ship images of a port), the ground segment's user terminal can submit a high-priority "bidding order" to the constellation. Once this order is received by the GIMU, it will instantly boost the relevant data for that area. The weighting caused data that might have otherwise taken the third path to suddenly break through. The threshold is that the original data packets are immediately transmitted to the ground station via a high-priority Ka-band link or laser link, thereby realizing the data-as-a-service business model of "dynamic real-time reconfiguration of on-board resources according to ground capital / demand".

[0050] Furthermore, the value assessment calculation process in MDVA can be embedded in the hardware logic gates of the FPGA (forming line-speed processing), enabling the comprehensive value score to be obtained in nanoseconds as soon as the raw data packet leaves the sensor. Only data determined to follow the second path will be moved to NPU or GPU memory via DMA (Direct Memory Access) channels. This hardware-level transformation ensures that value assessment itself does not become a bottleneck for on-board computing, achieving "zero additional burden" value assessment.

[0051] This embodiment uses a strict hard threshold comparison for DLM path determination. In other embodiments, fuzzy logic control can also be used, employing a fuzzy rule inference system. The comprehensive value score S and resource status are input into the fuzzy controller, which outputs the "membership degree" of each path. The path is selected according to the principle of maximum membership degree, avoiding data value jumps at threshold edges. Alternatively, a priority queue-based soft scheduling mechanism can be used. Instead of setting a hard threshold, all non-discardable data is stored in a hierarchical priority queue. When the download window arrives, data is dequeued and downloaded in descending order of score; when storage overflows, data is forcibly popped from the tail of the queue (lowest score) and discarded.

[0052] Figure 4This is a timing diagram of the inter-satellite value negotiation protocol of the present invention. Figure 4 As shown, CSVN-PE operates a lightweight value negotiation protocol between neighboring satellites. To prevent "Byzantine nodes" (e.g., a satellite's sensor malfunction causing it to incorrectly broadcast a high-value fingerprint, leading to other satellites mistakenly deleting data) during cross-satellite value negotiation, an onboard lightweight distributed ledger is further introduced. Whenever a satellite executes the fifth or fourth path, it uses its decision-making criteria (including data fingerprints broadcast by other satellites and consensus value scores) to determine its path. The data (including the timestamp and location at the time) is packaged into micro-blocks and uploaded to the blockchain. When receiving data, the ground station can verify the records on the blockchain, trace and audit the legality of all discarded data, and provide commercial settlement credentials for data sharing among multiple satellite operators. The cross-satellite value negotiation protocol engine is configured to perform negotiation through the following process: Step 1 (Fingerprint Broadcast): The data holding satellite calculates the hash fingerprint of the newly collected data and broadcasts it to neighboring satellites; Step 2 (Interest Expression): After receiving the fingerprint, the neighboring satellite matches and searches the fingerprint database of this satellite. If the same or highly similar data is found, an interest confirmation message is sent along with the existing value assessment of the data by this satellite. Step 3 (Meta-information interaction): After receiving the interest confirmation message, the data holding satellite further sends detailed data metadata to neighboring satellites, including the data size, type, and timeliness, to provide a basis for accurate assessment.

[0053] Step 4 (Value Negotiation): The data-holding satellite and neighboring satellites exchange their respective value assessments of the data (including value scores and their components), and reach a consensus value score S through a negotiation algorithm; Step 5 (Collaborative Decision-Making): Based on the consensus value score, both parties adjust their respective lifecycle path decisions; if one party has already stored the same data, the other party selects the fourth or fifth path to process the data in order to perform data deduplication at the constellation level. Step Six (Record on Chain): Record the consensus value score in the satellite's distributed ledger (optional) for use by ground stations for auditing and commercial settlement.

[0054] Step 7 (Download Instructions): After receiving and confirming the record of the consensus value score, the ground station sends confirmation and download scheduling instructions to the data holding satellite to guide the data distribution task of the data holding satellite, thus completing the closed loop of the entire negotiation and scheduling.

[0055] The application scenarios of the present invention are illustrated below through three specific examples.

[0056] Example 1: Remote sensing constellation value-driven data processing system. Figure 5is a deployment diagram of the remote sensing constellation value-driven data processing system of the present invention. As shown in Figure 5 , in this embodiment, the architecture proposed by the present invention is deployed in a constellation composed of 6 low-orbit remote sensing satellites (Satellite 1, Satellite 2, Satellite 3, Satellite 4, Satellite 5, Satellite 6). Each satellite is equipped with a Synthetic Aperture Radar (SAR) and an optical sensor, and operates at an orbital altitude of 500-600km. Each satellite is deployed with a complete V-DLMA architecture, and the core hardware configuration is: 1 ARM architecture CPU for value assessment and protocol processing, 1 NPU (Neural-network Processing Unit, for on-board AI inference, corresponding to the second path), 2TB NVMe SSD storage (for compressed storage of the third path), and an inter-satellite laser communication terminal (10Gbps, for protocol interaction of CSVN-PE). This system includes two satellite-to-ground data downlink paths, including a Ka-band downlink link (1Gbps), which is a link for the first path to urgently transmit data to the ground station; and an X-band downlink link (300Mbps), which is a link for the third path to routinely transmit data to the ground station. MDVA, DLM, CSVN-PE and RATA run on the real-time operating system in the form of containerized microservices.

[0057] The typical working scenarios of this embodiment include: Scenario 1: Rapid disaster response.

[0058] Rapid disaster response: An earthquake disaster suddenly occurs in a certain area. Satellite 3 in the constellation collects SAR images of this area during overpass. MDVA assesses that the data has an extremely high correlation with the "disaster emergency" task (the task correlation component V_mission=0.95), and the comprehensive value score . DLM triggers the first path, and Satellite 3 immediately transmits the SAR images to the ground emergency command center through the Ka-band emergency link, with a total delay of less than 5 seconds. At the same time, CSVN-PE broadcasts a message that this area has been observed to subsequent Satellite 4 and Satellite 5. The subsequent satellites downgrade their own observation data of this area to the third path or the fourth path according to the negotiation result, so as to avoid redundant storage.

[0059] Scenario 2: Conventional marine monitoring.

[0060] In the normal operation mode, the marine monitoring data collected by Satellite 1 has an evaluation score S=0.45, which falls in the interval of T3≤S<T2, triggering the third path to perform compression and then storage. The data is subjected to near-lossless compression on the satellite using the CCSDS 123.0-B-2 standard, with a compression ratio of about 3:1, and is stored in the SSD to wait for the conventional X-band downlink window 4 hours later.

[0061] Scenario 3: Automatic filtering of cloud-covered images.

[0062] The optical images acquired by Satellite 2, analyzed by MDVA, showed a cloud coverage rate >80%, with both scientific and commercial value below the threshold, resulting in a composite score of [insert score here]. This triggers the fifth path to discard the data directly. This data did not occupy any storage or download resources. Simulation estimates suggest that this scenario can save 25-40% of unnecessary storage and transmission overhead in typical optical remote sensing tasks.

[0063] Performance simulation results: Simulation tests based on historical remote sensing mission data show that after deploying the V-DLMA architecture, under the same on-board resource conditions: the effective data downlink volume is increased by 4.2 times (core indicator), the on-board storage utilization efficiency is increased by 55%, the emergency mission response latency is reduced from an average of 12 minutes to 2.3 seconds, and the data deduplication rate within the constellation reaches 45%.

[0064] Example 2: Dynamic Bandwidth Allocation System for Communication Satellite Constellations. This example applies the architecture of the present invention to a Medium Earth Orbit (MEO) constellation consisting of 12 communication satellites. In this scenario, the value assessment object of V-DLMA is abstracted from "data packets" to "communication link requests," evaluating the value of each user's link request (based on user priority, service type, latency sensitivity, service level agreement, etc.), and dynamically allocating scarce on-board bandwidth and transponder resources according to the value score. The cross-satellite value negotiation protocol evolves into a "cross-satellite bandwidth exchange protocol" in this example: when the value of a user request in a certain area of ​​a satellite is low while the value of a corresponding area of ​​an adjacent satellite is high, the beam coverage is dynamically adjusted or the user connection is transferred through inter-satellite links to achieve constellation-level bandwidth resource optimization. Simulation results show that the overall throughput of the constellation is increased by 28%, and the service quality guarantee rate for high-priority users reaches 99.7%.

[0065] Example 3: Autonomous Management of Scientific Satellite Data. This example applies the architecture of this invention to deep space scientific exploration missions (such as Mars orbiters or astronomical observation satellites). In this scenario, the value assessment of scientific data focuses on... The system utilizes anomaly detection algorithms to automatically identify rare scientific phenomena (such as gamma-ray bursts and exoplanet transit signals) and automatically triggers the first-path emergency downlink for data of extremely high scientific value. The cross-satellite value negotiation protocol has evolved into "scientific discovery negotiation": when multiple satellites observe the same scientific phenomenon, they negotiate their respective data acquisition strategies to avoid duplicate observations and achieve three-dimensional coverage by collaboratively adjusting observation angles. This embodiment particularly emphasizes the mission independence of the V-DLMA architecture; in this embodiment, the weights of the five value dimensions of MDVA are reconfigured as follows: This demonstrates the architecture's ability to flexibly adapt to different task scenarios.

[0066] The nodes of this invention are not limited to satellites. This architecture is fully scalable to heterogeneous air-space-ground networks that include high-altitude platforms (HAPS, such as stratospheric airships and high-altitude drones) and relay satellites (GEO / MEO). For example, data assessed as "high-value but without a direct ground link" by a low-Earth orbit satellite is routed via a first path to an upper-level GEO relay satellite, utilizing the GEO's continuous downlink window for emergency backhaul.

[0067] While some embodiments of the invention have been described in this application, those skilled in the art will understand that these embodiments are merely illustrative. Numerous variations, alternatives, and improvements will arise in those skilled in the art under the teachings of this invention without departing from its scope. The appended claims are intended to define the scope of the invention and thereby cover the methods and structures within the scope of the claims themselves and their equivalents.

Claims

1. A spaceborne value-driven data lifecycle management architecture, characterized in that, include: A multidimensional value evaluator is configured to evaluate the timeliness, scarcity, mission relevance, commercial value, and scientific value of raw data packets in orbit and generate a comprehensive value score. A data lifecycle manager is configured to receive a comprehensive value score output by a multidimensional value evaluator and compare the value score with a decision threshold. And automatically execute one of at least two predefined lifecycle paths; The cross-satellite value negotiation protocol engine exchanges data fingerprints and value negotiation messages with neighboring satellites through inter-satellite links to perform data deduplication and value collaboration at the constellation level. The resource status awareness and threshold adaptive adjustment module is configured to monitor on-board storage space availability, computing load, communication window status and current task mode, as well as adjust the evaluation weights of the multidimensional value evaluator and the decision thresholds of the data lifecycle manager. The ground interface and model update module is configured to receive updated value assessment models, task priority instructions, and user requests from the ground station to perform data lifecycle iterations.

2. The architecture according to claim 1, characterized in that, The formula for calculating the comprehensive value score by the multidimensional value assessor is as follows: in, For timeliness value function, This represents the time elapsed since the data was collected. The attenuation rate constant; The result is calculated based on the number of matches between data fingerprints and on-board historical data. The lower the number of matches, the higher the scarcity. The result is calculated using the cosine similarity between the data semantic labels and the active task target vector; For predicting the economic value of data based on ground market models; The result is calculated based on the difference between the data content and the distribution of historical data; As a weighting factor for timeliness assessment, As a weight for scarcity assessment, As the weight of task relevance, As a measure of commercial value, Weighting based on scientific value.

3. The architecture according to claim 1, characterized in that, The data lifecycle manager maintains the relationship between four decision thresholds, namely, the first threshold is greater than the second threshold, which is greater than the third threshold, which is greater than the fourth threshold; the lifecycle path includes: When the comprehensive value score is greater than or equal to the first threshold, the lifecycle path is the first path, that is, the original data packet is immediately transmitted to the ground station via the Ka-band link or the laser link; When the comprehensive value score is less than the first threshold and greater than or equal to the second threshold, the lifecycle path is the second path, that is, the original data packet is deeply processed on the satellite and then transmitted to the ground station. When the comprehensive value score is less than the second threshold and greater than or equal to the third threshold, the lifecycle path is the third path, that is, the original data packet is compressed without loss or near loss and stored in the on-board storage module, waiting for the regular downlink window to be downlinked to the ground station. When the comprehensive value score is less than the third threshold and greater than or equal to the fourth threshold, the lifecycle path is the fourth path, that is, the original data packet is discarded after generating the thumbnail or summary metadata. When the comprehensive value score is less than the fourth threshold, the lifecycle path becomes the fifth path, i.e., the original data packet is discarded.

4. The architecture according to claim 3, characterized in that, The deep processing includes at least one of the following: AI inference, feature extraction, and data fusion.

5. The architecture according to claim 1, characterized in that, The cross-satellite value negotiation protocol engine is configured to perform negotiation through the following process: Step one: The data holding satellite calculates the fingerprint of the newly acquired data and broadcasts it to neighboring satellites; Step 2: After receiving the data fingerprint, the neighboring satellites match and search the local satellite's data fingerprint database. If the same or highly similar data is found, an interest confirmation message is sent along with the local satellite's existing value assessment of the data. Step 3: After receiving the interest confirmation message, the data holding satellite sends data element information to neighboring satellites, including the size, type and timeliness of the data, to provide a basis for value assessment. Step four: The data-holding satellite and neighboring satellites exchange their respective value assessments of the data, and obtain a consensus value score through a negotiation algorithm; Step 5: Based on the consensus value score, both parties adjust their respective lifecycle path decisions; if one party has already stored the same data, the other party selects the fourth or fifth path to process the data in order to perform data deduplication at the constellation level. Step six: Record the consensus value score in the satellite's distributed ledger for use by the ground station for auditing and commercial settlement; Step 7: After receiving and confirming the consensus value score record, the ground station sends confirmation and scheduling instructions to the data holding satellite to guide the data distribution task and complete the closed loop of the entire negotiation and scheduling.

6. The architecture according to claim 1, characterized in that, The data fingerprint generation method includes at least one of the following: hash algorithm, semantic feature vector extraction, and spatiotemporal bounding box cross method.

7. The architecture according to claim 1, characterized in that, The evaluation weight adjustment rule is as follows: in emergency mode, the task relevance weight is automatically increased to 0.4-0.5; in energy-saving mode, the scarcity evaluation weight and scientific value weight are reduced to reduce computational overhead.

8. The architecture according to claim 1, characterized in that, The decision threshold adjustment rule is that when on-board storage space is tight, the third and fourth thresholds are raised so that more data is directed to the fourth or fifth path.

9. The architecture according to claim 1, characterized in that, When a specific event signal is detected, the system switches from normal mode to emergency mode, triggering a reconfiguration of evaluation weights and decision thresholds; the specific event includes at least one of the following: earthquake early warning and military alert.

10. A method for spaceborne value-driven data lifecycle management, characterized in that, Includes the following steps: The multidimensional value evaluator performs a multidimensional value evaluation on the input data packet and generates a comprehensive value score; The comprehensive value score is received by the data lifecycle manager; The data lifecycle manager selects to execute at least one of two lifecycle paths based on a comparison of the overall value score and the decision threshold. The cross-satellite value negotiation protocol engine interacts with other satellites via inter-satellite links and feeds back data fingerprints and value negotiation results to the multi-dimensional value evaluator and data lifecycle manager to perform collaborative optimization of the various modules of the data lifecycle management architecture. The resource status perception and threshold adaptive adjustment module continuously monitors the satellite resource status and adjusts the evaluation weight of the multidimensional value evaluator and the decision threshold of the data lifecycle manager to form a closed-loop adaptive adjustment for resource perception. The ground interface and model update module receive value assessment model updates, task priority instructions, and user requests from the ground via the satellite-to-ground link to perform data lifecycle iterations.

Images