Data tiering method and system for space-grade solid state drive

Abstract

The application discloses a data hierarchical storage method and system applied to a spaceflight solid state disk, and relates to the technical field of data hierarchical storage. A plurality of storage levels are integrated in the solid state disk, and the data to be stored is acquired. The access features of the data to be stored are collected in real time. Meanwhile, the context information of a spaceflight task is collected. The attribute label of the data to be stored is determined according to the identification of the context information of the spaceflight task. The access features and the attribute label are input into a pre-constructed storage strategy evaluation space. The matching scores of the data to be stored in different storage levels are determined in the storage strategy evaluation space in combination with the current on-orbit environment state and system load condition of the spacecraft. The data to be stored is transferred from the original storage level to the target storage level, and the global data mapping table is updated. Meanwhile, the data refresh strategy and error correction content are dynamically adjusted according to the physical characteristics of the target storage level, so that the accuracy of the data hierarchical storage sequence of the solid state disk is improved.

Description

Technical Field

[0001] This invention relates to the technical field of hierarchical data storage, and more particularly to a method and system for hierarchical data storage of solid-state drives used in aerospace applications. Background Technology

[0002] With the rapid development of aerospace technology, the amount of data generated by spacecraft during their on-orbit operation is growing exponentially, covering various types of data, from high-priority flight control data and telemetry parameters to massive amounts of scientific experimental payload data. These data vary significantly in terms of access frequency, data value, retention time, and reliability requirements. To meet these complex storage needs, solid-state drives (SSDs) have gradually replaced traditional mechanical storage devices and become the core component of spacecraft data storage systems due to their advantages such as high throughput, low latency, and shock resistance.

[0003] Most existing aerospace-grade solid-state drives (SSDs) use a single type of storage medium for unified storage, usually requiring the use of high-performance, high-reliability SLC (Surface Mount Cell) NAND flash memory. However, this results in high storage costs and limited capacity density, making it difficult to meet the ever-increasing demand for large-capacity data storage. Conversely, if high-density TLC (Thin-Layer Cell) or QLC (Quick-Layer Cell) media are used to increase capacity, it is difficult to meet the stringent requirements of critical control data for read / write speed and anti-interference capabilities. Although existing technologies have proposed the concept of hybrid storage, they often lack technical solutions for building multi-level storage units within a single SSD and establishing efficient and intelligent data migration channels, resulting in low accuracy of the data hierarchical storage sequence of SSDs. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of the prior art. This invention provides a data hierarchical storage method and system for solid-state drives used in aerospace applications.

[0005] This invention provides a data tiered storage method for solid-state drives used in aerospace applications, comprising: Multiple storage tiers are integrated within the solid-state drive, an intelligent data migration channel is established between each storage tier, and the multiple storage tiers are configured differently, so that the different storage tiers present a gradient distribution from performance to capacity. Acquire the data to be stored, collect the access characteristics of the data to be stored in real time, and collect the context information of the space mission. Determine the attribute labels of the data to be stored based on the identification of the context information of the space mission. The access features and attribute labels are input into the pre-built storage strategy evaluation space. Within the storage strategy evaluation space, the matching score of the data to be stored under different storage levels is determined by combining the current on-orbit environment status of the spacecraft and the system load. The corresponding optimization target content is then combined to generate data migration instructions. In response to migration decision instructions, the data to be stored is moved from the original storage tier to the target storage tier, and the global data mapping table is updated. At the same time, the data refresh strategy and error correction content are dynamically adjusted according to the physical characteristics of the target storage tier, thereby determining the data tiered storage sequence of the solid-state drive.

[0006] This invention provides a data tiered storage system for solid-state drives (SSDs) used in aerospace applications. This system is applied to the aforementioned data tiered storage method for SSDs used in aerospace applications. The data tiered storage system for SSDs used in aerospace applications includes: The storage module is used to integrate multiple storage tiers within a solid-state drive, establish intelligent data migration channels between different storage tiers, and perform differentiated configurations for multiple storage tiers, so that different storage tiers present a gradient distribution from performance to capacity. The data recognition module is used to acquire the data to be stored, collect the access characteristics of the data to be stored in real time, and collect the context information of the space mission. Based on the recognition of the context information of the space mission, the attribute labels of the data to be stored are determined. The data migration module is used to input access features and attribute labels into a pre-built storage strategy evaluation space, and within the storage strategy evaluation space, combine the current on-orbit environment status of the spacecraft and the system load to determine the matching score of the data to be stored under different storage levels, and further combine the corresponding optimization target content to generate data migration instructions. The data tiered storage module is used to respond to migration decision commands, move the data to be stored from the original storage tier to the target storage tier, update the global data mapping table, and dynamically adjust the data refresh strategy and error correction content according to the physical characteristics of the target storage tier, thereby determining the data tiered storage sequence of the solid-state drive.

[0007] Compared with the prior art, the beneficial effects of the present invention are: (1) Multiple storage levels are integrated in the solid-state drive to establish intelligent data migration channels between various storage levels and to perform differentiated configurations for multiple storage levels so that different storage levels present a gradient distribution from performance to capacity; the data to be stored is acquired, the access characteristics of the data to be stored are collected in real time, and the context information of the aerospace mission is collected. The attribute tags of the data to be stored are determined based on the identification of the context information of the aerospace mission. Multiple storage levels are introduced to further control the access characteristics and corresponding attribute tags of the data to be stored.

[0008] (2) Input the access features and attribute labels into the pre-built storage strategy evaluation space, and determine the matching score of the data to be stored under different storage levels in the storage strategy evaluation space in combination with the current on-orbit environment status of the spacecraft and the system load. Then, generate data migration instructions in combination with the corresponding optimization target content. Respond to the migration decision instructions, move the data to be stored from the original storage level to the target storage level, and update the global data mapping table. At the same time, dynamically adjust the data refresh strategy and error correction content according to the physical characteristics of the target storage level, thereby determining the data hierarchical storage sequence of the solid-state drive, further controlling the data migration instructions, fully considering the data refresh strategy and error correction content, and improving the accuracy of the data hierarchical storage sequence of the solid-state drive. Attached Figure Description

[0009] Figure 1 This is a flowchart illustrating the data tiered storage method for solid-state drives used in aerospace applications, as described in this embodiment of the invention. Figure 2 This is a flowchart illustrating step S11 of the data tiered storage method for solid-state drives using aerospace applications in this embodiment of the invention. Figure 3 This is a flowchart illustrating step S12 in the data tiered storage method for solid-state drives using aerospace applications in this embodiment of the invention. Figure 4 This is a flowchart illustrating step S13 of the data tiered storage method for solid-state drives using aerospace applications in this embodiment of the invention. Figure 5 This is a flowchart illustrating step S14 of the data tiered storage method for solid-state drives using aerospace applications in this embodiment of the invention. Figure 6 This is a simplified schematic diagram illustrating a specific application scenario of the data hierarchical storage method for solid-state drives used in aerospace applications, as described in this embodiment of the invention. Figure 7 This is a schematic diagram of the structural composition of a data hierarchical storage system for solid-state drives used in aerospace applications, as described in this embodiment of the invention. Detailed Implementation

[0010] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0011] Please see Figures 1 to 7 A data tiered storage method for solid-state drives (SSDs) used in aerospace applications, applied to data tiered storage scenarios; the data tiered storage method for SSDs used in aerospace applications includes: Step S11: Integrate multiple storage tiers within the solid-state drive, establish intelligent data migration channels between each storage tier, and perform differentiated configurations for the multiple storage tiers, so that different storage tiers exhibit a gradient distribution from performance to capacity. Step S12: Acquire the data to be stored, collect the access characteristics of the data to be stored in real time, and at the same time, collect the context information of the space mission. Determine the attribute labels of the data to be stored based on the identification of the context information of the space mission. Step S13: Input the access features and attribute labels into the pre-built storage strategy evaluation space, and determine the matching score of the data to be stored under different storage levels in the storage strategy evaluation space in combination with the current on-orbit environment status of the spacecraft and the system load. Then, generate data migration instructions in combination with the corresponding optimization target content. Step S14: Respond to the migration decision command, move the data to be stored from the original storage tier to the target storage tier, update the global data mapping table, and dynamically adjust the data refresh strategy and error correction content according to the physical characteristics of the target storage tier, thereby determining the data tiered storage sequence of the solid-state drive.

[0012] refer to Figure 2 In step S11, the specific steps are as follows: S111: For solid-state drives used in aerospace applications, this indicates the multiple storage tiers integrated within the solid-state drive. These storage tiers include a first storage tier based on performance, a second storage tier based on capacity, and a third storage tier based on cost. S112: Establish an intelligent data migration channel based on a high-speed interconnect bus between the first storage level, the second storage level and the third storage level, and perform differentiated physical configuration of the first storage level, the second storage level and the third storage level for aerospace application environment. By adjusting the error correction strength, data refresh rate and voltage threshold, different storage levels present a continuous gradient distribution from performance to capacity.

[0013] In step S111, the specific steps are as follows: At this point, the physical storage media within the solid-state drive is identified and logically partitioned. The storage unit with the highest read / write speed, highest write endurance, and strongest resistance to radiation and interference is selected to form the first storage tier. This first storage tier is mainly implemented by configuring single-cell (SLC) flash memory chips or storage blocks that simulate SLC mode. At this time, each storage unit stores only 1 bit of data. During the marking process, the system firmware marks the physical block address range corresponding to this tier as "Tier-1" or "high-performance layer" and establishes an initial mapping table from logical address to physical address, clarifying that this tier is dedicated to carrying critical mission data with extremely high real-time requirements, extremely high data value, and a decisive role in system stability during aerospace missions.

[0014] After the first storage tier is defined, storage cells with moderate read / write speeds, moderate write tolerance, but high storage density are selected to form the second storage tier. This second storage tier mainly uses multi-cell (MLC) flash memory chips. At this time, each storage cell stores 2 bits of data, which can provide a larger storage capacity per unit area than SLC. During the marking process, the system firmware marks the physical block address range corresponding to this tier as "Tier-2" or "high capacity tier" and defines its attribute in the metadata as a tier that balances throughput and storage space. It is dedicated to carrying routine telemetry data, payload scientific data, or general mission logs that are accessed frequently in aerospace missions but whose data importance is slightly lower than that of the first tier.

[0015] The third storage tier is formed by selecting storage cells with high storage density and low cost but relatively low read / write performance and limited write endurance. This third storage tier typically uses three-cell (TLC) or four-cell (QLC) flash memory chips. In this case, each storage cell stores 3 or 4 bits of data to maximize the utilization of physical space. During the marking process, the system firmware marks the physical block address range corresponding to this tier as "Tier-3" or "low-cost tier" and defines its attributes as archive-type storage space, which is dedicated to carrying massive amounts of historical data, backup files, or non-critical environmental monitoring records that are generated for long-term preservation and accessed very infrequently during aerospace missions.

[0016] In step S1112, the specific steps are as follows: At this point, by deploying a high-speed data exchange matrix inside the solid-state drive controller and establishing a full-duplex physical connection channel between the first, second, and third storage tiers using an internal high-speed interconnect bus, this intelligent data migration channel is configured to have independent DMA control capabilities. It can perform data transfer operations between different storage tiers in parallel in the background without occupying the main controller's CPU computing resources. At the same time, this channel integrates a data integrity verification mechanism to ensure that data can be corrected in real time when it is affected by the single-event upset effect in the aerospace environment during cross-tier transmission, thus ensuring the bandwidth stability and low latency characteristics of the cross-tier migration path and supporting the dynamic flow of data.

[0017] Optionally, the intelligent data migration channel refers to the establishment of a dedicated point-to-point data transmission path between physical storage units belonging to different storage tiers (first, second, and third storage tiers) within the solid-state drive controller via an integrated high-speed serial interconnect bus, such as an Advanced Scalable Interface Bus. The "intelligence" of this channel is reflected in its embedded data migration engine, which runs independently of the main controller's central processing unit. This engine can parse data migration instructions issued from the upper layer, automatically identify the source and target physical addresses, and perform data transfer operations in parallel in the background through direct memory access, without occupying the processor core's computing resources.

[0018] In the specific setup process, the solid-state drive firmware uniformly addresses the physical address range of each storage level during the initialization phase, and assigns an independent virtual channel identifier to each level on the bus. At the same time, it configures channel priority arbitration logic to ensure that the migration operation of critical task data can obtain higher bus access rights. In addition, the migration channel also integrates real-time cyclic redundancy check logic, which performs integrity verification after data is read from the source level and before it is written to the target level, and automatically triggers a retransmission mechanism when a transmission error is detected. Through the above hardware and firmware collaborative configuration, the channel realizes automated, lossless, and low-latency data migration capabilities between different storage levels, thus constituting an intelligent data migration channel that can be specifically implemented by those skilled in the art.

[0019] To address the inherent physical differences in each storage level, and considering the high-energy particle radiation, temperature fluctuations, and vacuum environment encountered by spacecraft during on-orbit operation, the physical interface parameters of the NAND flash memory for each level are customized accordingly. This involves adjusting the register configuration of the underlying flash memory controller to modify the read voltage threshold, write programming algorithm, and charge retention level of different storage cells. The first storage level is configured for high reliability; the second for balanced performance; and the third for high density. This physical-level differentiation ensures that each level maintains its own storage medium characteristics while adapting to the specific requirements of the space environment for data retention and interference resistance.

[0020] Based on the aforementioned physical configuration, the error correction coding (ECC) strategy, data backend refresh frequency, and read / write operation voltage thresholds for each storage tier are finely adjusted. For the first storage tier, the highest-strength low-density parity-check code (LDPC) iterative algorithm is enabled, coupled with a high-frequency data refresh strategy and a strict read voltage threshold, to maximize data integrity and access speed. For the second storage tier, the number of error correction iterations is appropriately reduced and the data refresh interval is correspondingly extended, while the voltage threshold is adjusted to the standard range to balance performance and lifespan. For the third storage tier, a basic error correction strength and a long-cycle data refresh strategy are adopted, and the voltage threshold window is widened to increase storage density. Through this step-by-step parameter adjustment, the three storage tiers form a continuous gradient distribution in terms of physical performance indicators, transitioning from high performance and high reliability to large capacity and low cost.

[0021] refer to Figure 3 In step S12, the specific steps are as follows: S121: Real-time acquisition of data to be stored, determination of multiple access contents based on the identification of the data to be stored, and determination of access characteristics of the data to be stored in conjunction with a dynamic monitoring mechanism. Access characteristics include at least temporal locality, spatial locality, access frequency and read / write rate. S122: Real-time acquisition of space mission context information via the spacecraft mission bus interface. Context information includes mission phase identifiers, payload types, and command priorities. The context information is identified and parsed to determine the attribute tags of the data to be stored. The attribute tags are used to characterize the business importance, security and confidentiality level, and timeliness of the data in the mission process.

[0022] In step S121, the specific steps are as follows: At this time, the solid-state drive controller intercepts or receives the data stream to be stored from the spacecraft computing platform in real time through the host interface, and performs protocol parsing and content recognition on the received data packets. By parsing the header information, logical block address (LBA) range and associated metadata identifier of the data packets, the controller extracts the access content set corresponding to the data. The access content set contains the data source type, data size and expected access mode. By parsing the access content, the system can initially determine whether the data belongs to an immediate write of a single instruction or a sequential stream of batch data, thus providing basic context information for subsequent feature analysis.

[0023] During data reception and writing, the controller's internal dynamic monitoring engine is activated, using a preset sliding time window and address tracking algorithm to capture the data request sequence of the data to be stored and its adjacent addresses in real time. The monitoring engine calculates temporal locality by statistically analyzing the probability of repeated access to the current data block within the preset time window, that is, to determine whether the data is frequently read or updated within a very short time interval. At the same time, it calculates spatial locality by analyzing the continuity and clustering of logical block addresses (LBAs), that is, to determine whether the data belongs to a continuous large file or a data block that is frequently accessed sequentially. This process transforms the access pattern into a quantifiable locality indicator through the linkage scanning of hardware counters and logical address mapping tables.

[0024] After completing the locality analysis, the dynamic monitoring mechanism further analyzes the historical records and current operation status of the data to be stored to determine its access frequency and read / write rate. The system queries the historical access log table to obtain the number of hits of the logical address over a long period of time and calculates the access frequency per unit time. At the same time, by monitoring the current I / O queue depth and data throughput, the instantaneous transmission rate and average read / write bandwidth of the data are measured in real time. The monitoring engine combines these physical-level traffic statistics with data attributes to generate a set of multi-dimensional feature vectors. These feature vectors comprehensively characterize the activity level of the data and its consumption of storage system bandwidth, providing accurate input parameters for subsequent storage strategy evaluation.

[0025] In step S122, the specific steps are as follows: At this point, the solid-state drive controller establishes a real-time data link with the spacecraft platform and payload management unit through the integrated spacecraft mission bus interface. The controller not only receives storage write commands but also listens for and intercepts mission scheduling packets and time synchronization codes broadcast on the bus. The system extracts key context information fields from the above data packets, specifically including "the current mission phase identifier, such as launch phase, orbit insertion phase, Earth observation phase, or data return phase", "the type of payload generating data, such as star sensor, synthetic aperture radar, or scientific experiment instrument", and the command priority identifier accompanying the write command. The acquisition process adopts a hardware interrupt-driven mode to ensure strict synchronization of context information and data to be stored in terms of timestamps, providing an accurate macroscopic mission background for subsequent data attribute analysis.

[0026] After obtaining the original context information, the controller's built-in firmware parsing module performs semantic recognition and logical mapping on the received information. The parsing module determines the current system's working mode based on the task stage identifier, then identifies the business source of the data generation based on the load type, and finally determines the urgency of data processing based on the instruction priority. Through this recognition and parsing process, the system transforms discrete context parameters into a set of structured attribute tags. These attribute tags are specifically used to characterize the data's "business importance, such as core control data or auxiliary record data", "security and confidentiality level, such as top secret, confidential, or ordinary level", and the data's "timeliness in the task process, such as real-time processing data or delayed archiving data", thereby assigning a clear business attribute vector to each logical data block.

[0027] The parsed attribute tags are attached as metadata to the control block of the data to be stored and input into the storage policy evaluation space. The "business importance" parameter in the attribute tags is directly related to the reliability selection of the storage level, the "security and confidentiality level" parameter triggers the encryption and anti-interference hardening mechanism of specific storage areas, and the "timeliness" parameter determines the retention time and migration strategy of data between different levels. In this way, attribute tags enable the storage management system to make decisions not only based on the read and write characteristics of the data itself, but also to combine the overall context of the aerospace mission to achieve a precise match between data value and storage resources.

[0028] refer to Figure 4 In step S13, the specific steps are as follows: S131: Obtain access features and attribute labels, input the access features and attribute labels into the pre-built storage strategy evaluation space, and combine the multi-objective decision-making mechanism to determine the decision content of different dimensions. Further combine the current on-orbit environment status and system load of the spacecraft to perform multi-dimensional feature analysis, thereby determining the matching score of the data to be stored under different storage levels. S132: Obtain the matching score of the data to be stored under different storage levels, mark the level status of the corresponding storage level, and further perform cross-matching with the preset optimization target content. In the matching process, the optimal data migration instruction is generated in combination with the storage constraints of the data to be stored, so as to realize the steady-state data storage of the spacecraft in the current usage scenario.

[0029] In step S131, the specific steps are as follows: At this point, the system calls the pre-built storage strategy evaluation space, which is defined as a logical mapping domain containing a multi-layer rule neural network or expert system. The "access features" obtained from step S121, such as temporal locality, spatial locality, access frequency, and read / write rate, and the "attribute labels" obtained from step S122, such as business importance, security and confidentiality level, and timeliness, are standardized and combined into a high-dimensional feature vector. This feature vector is then injected into the storage strategy evaluation space as an input parameter. Within the evaluation space, the system initially extracts storage requirement constraints that match the data to be stored, such as minimum latency requirements and maximum reliability thresholds, based on a preset feature weight matrix, laying the foundation for subsequent decision calculations.

[0030] Within the storage strategy evaluation space, the system initiates a multi-objective decision-making mechanism to conduct parallel evaluations of the complex constraints of aerospace applications. This mechanism simultaneously considers multiple mutually restrictive optimization dimensions, including performance maximization, capacity utilization, power consumption control, and data retention. The multi-objective decision-making mechanism analyzes the adaptability of the data to be stored in different storage tiers (first, second, and third storage tiers) through weighted algorithms or Pareto optimal solution search, and determines the decision content for different dimensions. For example, for high real-time data, the decision-making mechanism tends to assign higher weights to the performance dimension; for non-critical archived data, it assigns higher weights to the capacity dimension, thereby generating a preliminary tier preference ranking.

[0031] After obtaining the initial hierarchical preferences, the system further collects the current on-orbit environmental status of the spacecraft and the system's own load status in real time as correction factors. The on-orbit environmental status includes space radiation flux, chip temperature profile, and power supply voltage stability. The system load status includes the current I / O queue depth, write amplification factor, and controller CPU utilization. The system performs multi-dimensional feature analysis on these environmental and load variables to determine whether the current system is in a high-stress operating mode. For example, if excessive temperature or enhanced radiation is detected, the system will increase the decision weight of error correction capability and data refresh rate. If the system load is too high, it tends to reduce the amount of data migrated across levels to maintain system stability.

[0032] The system integrates and weights the decision content generated by the multi-objective decision-making mechanism with the correction factors obtained based on environment and load analysis, and outputs the matching score of the data to be stored relative to each storage tier. The matching score is a comprehensive quantitative indicator. The calculation process involves scoring the matching degree between data features and "tier physical characteristics, such as bandwidth, IOPS, and bit error rate", and deducting the risk score caused by harsh environment or system overload. The system finally outputs a score vector, such as [Tier-1_Score:95,Tier-2_Score:60,Tier-3_Score:40]. This score vector intuitively reflects the comprehensive quality of data storage at each tier, providing an accurate quantitative basis for generating the final migration instructions.

[0033] Optionally, the matching score is a numerical indicator used to quantify the overall suitability between the data to be stored and a certain storage level. Its calculation process is implemented through a predefined multi-factor weighted scoring model. In this case, for each candidate storage level, the system first extracts five basic scoring factors based on the access characteristics and attribute tags of the data to be stored, namely: the popularity factor obtained by temporal locality quantification, the order factor obtained by spatial locality quantification, the activity factor obtained by access frequency quantification, the criticality factor obtained by business importance tag transformation, and the real-time factor obtained by timeliness tag transformation.

[0034] The value range of each factor is normalized to the interval between zero and one. At the same time, the system presets a fixed set of weight coefficients for each storage level. The weight vector of the first storage level focuses on the popularity factor and the real-time factor, the weight vector of the second storage level distributes the factors evenly, and the weight vector of the third storage level focuses on the sequential factor while weakening the real-time factor.

[0035] After obtaining the aforementioned factors and weights, the system calculates the weighted sum of each factor as the basic matching score. Based on this, the system further introduces an environmental correction coefficient and a load correction coefficient. The environmental correction coefficient is jointly determined by the radiation flux and chip temperature in the current on-orbit environment. When the radiation flux or temperature exceeds a preset threshold, this coefficient takes a value less than one, used to reduce the matching score of high-reliability data with low-error-correction-capability levels in the current environment. The load correction coefficient is jointly determined by the current system input / output queue depth and write amplification factor, used to appropriately suppress the tendency to allocate new data to levels that are already in a high-occupancy or busy state when the system is under high load.

[0036] The system multiplies the basic matching score by the environment correction factor and the load correction factor to obtain the final matching score of the data to be stored relative to the current storage tier. Through the above steps, the system outputs a unique score between zero and one for each storage tier. The higher the score, the better the overall fit between the data and the tier, thus providing a clear and reproducible quantitative basis for subsequent migration decisions.

[0037] In step S132, the specific steps are as follows: At this point, the controller reads the matching scores of the data to be stored calculated in the previous stage under different storage levels, and simultaneously queries and marks the real-time status of each storage level. This status includes the current number of remaining physical blocks, the percentage of used space, the current erase / write cycle (P / ECycle) count, and whether there are bad blocks or whether garbage collection is in progress. The system compares the matching scores with the level status information, and initially filters out a set of candidate storage levels with scores higher than a preset threshold and physical status of "available" or "non-congested". This excludes storage levels that cannot accommodate data due to insufficient space or performance degradation, ensuring that subsequent decisions are made on the basis of physical feasibility.

[0038] The system introduces preset optimization objectives, which are input as decision weight factors into the evaluation logic for cross-matching with the candidate set. The optimization objectives are dynamically set according to different stages of the space mission, including "performance priority, such as requiring the lowest latency during the orbital insertion phase," "reliability priority, such as requiring data retention in high-radiation areas," or "capacity priority, such as requiring maximum storage space during the scientific observation phase." The cross-matching process uses a multi-dimensional weighted algorithm to fit the characteristics of the data to be stored with the optimization value that each level can currently provide, and calculates a comprehensive benefit index. This index not only reflects the hardware matching degree between the data and the level, but also reflects the contribution of the current storage operation to the overall mission objective.

[0039] After determining the comprehensive benefit index, the system further extracts the storage constraints of the data to be stored, including data retention time requirements, minimum read / write bandwidth requirements, and hardening level requirements against single-event upsets (SEUs). The system uses these hard constraints as filters to perform a final verification of the cross-matching results, eliminating storage options that do not meet the constraints, and locking in the target storage level that can simultaneously satisfy the highest comprehensive benefit index and all hard constraints. The controller generates the optimal data migration instruction, which explicitly includes the source logical address, target physical level identifier, target starting physical block address, and migration trigger type, ensuring that the instruction can accurately place the data in the most suitable physical location when executed, thereby achieving steady-state data storage of the spacecraft under the current complex usage scenario.

[0040] refer to Figure 5 In step S14, the specific steps are as follows: S141: In response to the data migration command, the internal data transfer engine is invoked to transfer the data to be stored from the original storage level to the target storage level without loss, and the global data mapping table is updated in real time after the transfer is completed to ensure the consistency between logical address and physical address. S142: Mark the physical characteristics of the target storage level, and dynamically adjust the data refresh strategy and error correction content according to the physical characteristics of the target storage level. At this time, the aggressiveness of background garbage collection is adaptively adjusted and the corresponding error correction intensity is adaptively selected, so as to determine the data hierarchical storage sequence of the solid-state drive under the premise of ensuring data integrity.

[0041] In step S141, the specific steps are as follows: At this point, the solid-state drive controller receives a data migration instruction from the upper-layer decision module. This instruction contains key parameters such as the source logical address, the target storage tier identifier, and the migration length. The controller immediately decodes the instruction and verifies its validity. Then, it suspends the current low-priority background task and calls the high-priority data transfer engine inside the firmware. This data transfer engine is configured to have direct memory access (DMA) control capabilities and can run independently of the main controller CPU core. It is specifically responsible for establishing point-to-point data transfer links between physical media at different storage tiers, ensuring that the migration operation can be executed at the highest bandwidth supported by the underlying hardware.

[0042] According to the instruction parameters, the data transmission engine directly reads the user data and its associated metadata from the physical pages of the original storage tier, such as the SLC cache tier or MLC main memory area, via the internal high-speed interconnect bus. During the reading process, the engine activates its internal data integrity verification mechanism, using Cyclic Redundancy Check (CRC) or ECC checksums to verify the correctness of the read data. After verification, the engine writes the data to the pre-allocated physical blocks of the target storage tier, such as the TLC archive area. To achieve lossless transmission, the system adopts a dual confirmation strategy of "pre-write verification" and "post-write readback". That is, after the data is written to the target medium, a read operation is immediately performed and compared with the source data to ensure that the bit integrity of the data content remains absolutely consistent during cross-tier migration, even if affected by bus signal interference or differences in the charge retention force of storage cells, preventing any bit flips or data corruption.

[0043] After confirming that the data has been completely and securely written to the target storage layer and that the metadata has been synchronously updated, the controller performs an atomic operation to update the global data mapping table (FTL, FlashTranslationLayer). The system cancels the old physical page address (PPA) corresponding to the logical address (LBA) of the data to be stored in the mapping table and fills in the new physical page address. At the same time, in order to prevent address mapping errors caused by unexpected power outages during the update process, the system adopts a log-based or shadow mapping table technology to record the update operation in a temporary area first. After confirming that the write is successful, the new mapping relationship is officially flushed into the main mapping area of ​​the flash memory. This process ensures that the pointing relationship from logical address to physical address is switched immediately upon completion of the migration, guaranteeing the address continuity and consistency of subsequent access operations on the host side.

[0044] In step S142, the specific steps are as follows: At this point, after the data is successfully written to the target storage tier, the controller firmware immediately queries and marks the inherent physical characteristic parameters of the target tier. These parameters include the type of storage cell (such as SLC, MLC, TLC), the charge retention capability of the storage cell, the raw bit error rate (RBER) threshold, and the current physical temperature and program / erase (P / E) cycle count of the tier. The system encapsulates these physical characteristics into a tier state descriptor to characterize the inherent limits of the current storage medium's data retention capability, providing a basic physical constraint boundary for subsequent reliability strategy adjustments.

[0045] Based on the identified physical characteristics, the controller dynamically adapts the data maintenance strategy. For physical layers with weak charge retention or poor anti-interference capabilities, such as high-density TLC layers, the system automatically shortens the data background refresh cycle and eliminates potential charge leakage or soft errors through more frequent read-verify-rewrite operations. Simultaneously, the system adjusts the configuration of the error correction algorithm, such as enabling a combination of hard and soft decision modes for low-density parity check (LDPC) with a higher number of iterations, or increasing the parity bit length of the redundant check code (ECC). Conversely, for layers with strong physical characteristics, the standard refresh cycle and error correction strength are maintained to reduce system overhead.

[0046] The system adaptively adjusts the aggressiveness of background garbage collection based on the physical load state and write amplification factor of the target storage tier. If the target tier has low write performance, such as TLC / QLC, the controller will trigger a more aggressive active garbage collection strategy to maintain the steady-state write performance of the system and prevent long latency jitter caused by garbage collection. This strategy involves moving valid data in advance and merging fragmented blocks. If the target tier has high performance, a conservative passive garbage collection strategy is adopted to save power. During this process, the system also combines real-time physical environment monitoring. If enhanced radiation or increased temperature is detected, a higher-intensity error correction algorithm will be automatically selected to cooperate with the garbage collection operation to ensure that no new risks are introduced during data movement.

[0047] After completing all the above physical adaptation operations, the controller records the final state of this data migration, the adjusted refresh strategy parameters, the selected error correction strength, and the garbage collection trigger conditions in the metadata area, thereby logically and physically solidifying the storage location and maintenance method of the data. At this time, the system generates the final solid-state drive data hierarchical storage sequence. This sequence not only reflects the distribution topology of the data in the multi-level storage units, but also includes the customized reliability assurance mechanism corresponding to each data block. Thus, while ensuring the integrity of spacecraft data, a hierarchical storage system that can adapt to environmental changes is constructed.

[0048] Optionally, the process of dynamically adjusting the data refresh strategy and error correction content is executed based on a set of preset rule tables that are bound to the quantitative indicators of the physical characteristics of the target storage level. Specifically, a parameter mapping table is pre-installed in the firmware of the solid-state drive. Each row of this mapping table corresponds to a storage level, and each row contains three fields: the physical characteristic parameter range of that level, the refresh cycle value bound to that range, and the error correction coding strength level bound to that range.

[0049] After the data is successfully migrated to the target storage level, the controller first reads the current physical characteristic parameters of that level, including the real-time monitoring value of the original bit error rate, the number of accumulated program-erase cycles in the storage cells of that level, and the current chip operating temperature of that level. The controller compares the above three parameter values ​​with the three preset threshold ranges for the corresponding level in the mapping table, and takes the worst matching level as the current physical characteristic level.

[0050] For the adjustment of the data refresh strategy, the firmware adopts the following deterministic rules: if the physical characteristic level is excellent, the background data refresh cycle is set to 96 hours; if the level is good, it is set to 72 hours; if the level is medium, it is set to 24 hours; if the level is poor, it is set to 8 hours, and active scanning refresh is triggered at the same time, that is, the controller actively reads the data of all storage blocks in this level and performs verification and rewriting.

[0051] For adjustments to the error correction content, the firmware adopts the following grading rules: When the physical characteristic level is excellent, basic low-density parity check code is enabled, using hard decision mode and the number of iterations does not exceed two; when the level is good, standard low-density parity check code is enabled, still using hard decision mode but the number of iterations is relaxed to four; when the level is medium, enhanced low-density parity check code is enabled, and hard decision and soft decision are used in combination, with the number of soft decision iterations set to two; when the level is poor, the highest strength low-density parity check code is enabled, using soft decision mode entirely and the number of iterations set to four, with an additional erase code for secondary protection.

[0052] Please see Figure 6 The specific application scenarios and corresponding embodiments of this application are as follows: In step S111, the controller scans and marks the NAND flash memory array integrated within it. For the first storage tier in terms of performance, the solid-state drive controller divides the onboard SLC flash memory chips (e.g., industrial-grade high-temperature resistant, radiation-hardened SLC NAND) into "Region A" in the logical address space and marks them as Tier-1 in the firmware layer. This tier is designed to store real-time attitude control parameters, orbit injection commands, and critical fault detection and isolation (FDI) logs generated by the flight control computer (FCC) onboard the spacecraft, ensuring that such data can be read and written at extremely high speeds with sub-millisecond latency in the harsh space radiation environment.

[0053] For the second storage tier in terms of capacity, the solid-state drive controller divides the onboard MLC flash memory chips into "Region B" in the logical address space and marks it as Tier-2. This tier utilizes the high-density characteristics of MLC media and is configured to store high-resolution Earth observation image data, streaming data collected by scientific experimental sensors, and periodic snapshots of the entire system's health status generated during the spacecraft's on-orbit operation. These data need to be read and written frequently during the mission phase, and there are certain requirements for throughput, but the requirements for interference resistance and erase / write lifetime are slightly lower than those for Tier-1. Therefore, Tier-2 can provide sufficient storage space while ensuring a certain level of performance.

[0054] For the third storage tier, which addresses cost considerations, the solid-state drive (SSD) controller divides the onboard TLC flash memory chips into "Region C" in the logical address space and marks it as Tier-3. This tier is explicitly designated as the SSD's data "cold storage" or archive area, primarily used to store historical telemetry playback data from the early stages of aerospace missions, raw payload data that has been downloaded but needs to be archived, and non-critical system operation logs, such as power-on self-test records. This data is typically rarely modified after being written. By leveraging the large capacity of TLC chips, SSDs can meet the long-term, large-volume cumulative storage needs of aerospace missions at extremely low physical space costs, thereby achieving optimal performance and capacity configuration within the limited spacecraft payload and power consumption budget.

[0055] In step S112, for the solid-state drive, the controller chip internally establishes a data transmission channel between the "SLC cache area: first level", "MLC main memory area: second level" and "TLC archive area: third level" through the integrated AXI high-performance bus interconnect architecture. This channel is configured with a dedicated data migration engine. When the solid-state drive receives a data migration instruction, such as needing to migrate attitude control data from the SLC area to the MLC area, the engine can establish a high-speed direct link on the internal bus through DMA to complete the data transfer at a transmission rate of over 5GB / s. At the same time, this link has a real-time cyclic redundancy check (CRC) function to prevent transmission bit errors caused by space radiation.

[0056] For the aerospace environment where solid-state drives operate in orbit, differentiated physical configurations are implemented for each level. For the first storage level (SLC), given that it stores critical flight control commands, the controller configures its physical interface to "aerospace rugged mode," which significantly improves the voltage accuracy of programming / erasing operations and forces the active data scrubbing refresh mechanism to be enabled every 4 hours to eliminate possible accumulated single-event soft errors and ensure data security.

[0057] For the second storage tier (MLC), considering that it stores high-definition image data from Earth observation, it is necessary to balance write speed and capacity. The controller is configured as "standard balanced mode", the error correction algorithm adopts soft decision LDPC but limits the number of iterations, the data refresh cycle is set to once every 24 hours, and the voltage threshold is adjusted to a moderate range that can tolerate slight charge leakage, thereby maximizing the utilization of storage space while ensuring the reliability of image data.

[0058] For the third storage tier (TLC), used to store massive amounts of historical telemetry logs, the controller is configured as a "high-density deep storage mode." In this mode, simplified hard-decision error correction codes are used to reduce latency overhead, the data refresh cycle is extended to every 72 hours or even longer, and the read voltage threshold window is widened to accommodate the narrow charge distribution characteristics of TLC cells. Through this series of specific parameter adjustments, the solid-state drive has successfully built a continuous performance gradient from the ultra-fast and highly reliable first tier to the large-capacity and durable third tier, adapting to the hierarchical storage needs of different stages and types of data in aerospace missions.

[0059] In step S121, when the spacecraft's payload computer sends a set of high-resolution raw image data of Earth observation to it, the main control interface of the solid-state drive captures the data stream in real time; the controller parses the header of the data packet, identifies the access content of the data as "large-capacity sequential write image data", marks its source as "visible light camera payload", and then starts the dynamic monitoring process for the data.

[0060] The dynamic monitoring engine inside the solid-state drive performs spatial and temporal locality analysis on this batch of image data. Since the observed images are usually written continuously in the order of shooting, the monitoring engine detects that the logical block addresses (LBAs) exhibit highly continuous characteristics. Therefore, it is determined that the spatial locality of the data is extremely strong, belonging to typical sequential stream data. At the same time, the monitoring engine analyzes the historical records of the data before writing and finds that the region has not been read repeatedly in past orbital cycles. Therefore, it is determined that its temporal locality is weak, that is, "hot data" that is not frequently accessed.

[0061] Based on the above identification results, the SSD's controller chip further quantifies and calculates the access characteristics of the data; the monitoring circuit counts that the write rate of this batch of data remains near the saturation bandwidth of the PCIe bus, determining that its read and write rates belong to a high-speed sequence; at the same time, combined with the task plan, the system identifies that this data is only generated during the imaging stage and transmitted back when passing through, and its access frequency is marked as a one-time or low-frequency access type. Through this series of feature extractions, the SSD accurately characterizes this batch of image data as cold data with the characteristics of "high spatial locality, low temporal locality, high write rate, and low access frequency", providing a solid technical basis for subsequent decisions to store it in a large-capacity third storage tier.

[0062] In step S122, the controller monitors the spacecraft's mission data stream in real time through the onboard SpaceWire bus interface. When the solid-state drive receives a flight telemetry data write request from the attitude control subsystem, the controller simultaneously captures the mission scheduling packet on the bus, extracts the mission phase identifier of "orbit insertion ignition control" from it, identifies the data source as the "high-precision fiber optic gyroscope" payload, and parses the "critical control" priority identifier carried by the write command.

[0063] The solid-state drive's main controller firmware performs in-depth analysis of the aforementioned context information. Since the mission phase is "orbit insertion and ignition," the firmware determines that any deviation during this phase could lead to mission failure, so it sets the business importance label of this data to "highest level." At the same time, considering that attitude control data involves the core safety of the spacecraft, its security level label is set to "top secret." Furthermore, given that this data needs to be read by the navigation computer within milliseconds for closed-loop control, its timeliness label is set to "real-time and highly timely."

[0064] Based on these generated attribute tags, the solid-state drive responds during the storage strategy evaluation phase. According to the attribute tags of "highest level of business importance" and "real-time and high timeliness", combined with the high-performance characteristics of "first storage level: SLC" established in S11, the system decides to forcibly allocate the flight telemetry data to the first storage level. At the same time, according to the "top secret" security level tag, the controller automatically enables the hardware encryption engine before writing the data and uses the high-strength LDPC error correction algorithm configured in the first storage level to strengthen the storage, thereby ensuring the security and immediate availability of critical attitude control data in the harsh environment of severe spacecraft vibration and high-energy particle radiation.

[0065] In step S131, the controller acquires the feature data collected in the previous stage; assuming the data to be stored is "hyperspectral imager payload", its feature vector is displayed as "high spatial locality, low temporal locality, high write rate", and the attribute label is displayed as "business importance: medium, security level: high, timeliness: non-real-time"; the controller inputs the vector into the storage strategy evaluation space, and the algorithm in the evaluation space determines according to the rules that the data belongs to the type of large capacity, high throughput but non-instant response, and initially tends to the second or third storage level with larger capacity.

[0066] The multi-objective decision-making mechanism of solid-state drives intervenes and begins to weigh the decision content of each dimension. The mechanism analysis found that although the data has a large throughput, it is not a key control data. Therefore, its weight in the performance dimension is appropriately reduced, while its weight in the capacity and power consumption dimensions is given higher weight. The initial decision tends to use the third storage tier (TLC) to maximize space utilization and reduce storage power consumption per bit.

[0067] The solid-state drive's sensor module provides feedback on the current on-orbit environmental status and system load. Assuming the spacecraft is currently in the South Atlantic Anomaly (SAA), where space radiation flux is significantly increased and the solid-state drive's main control chip temperature reaches the high-temperature threshold of 75 degrees Celsius, the multi-dimensional feature analysis module identifies the current environment as "high-risk." The high voltage threshold and weak anti-interference capability of the TLC unit may increase the risk of data rollover under the current environment. Therefore, the analysis module sends a correction factor to the decision logic, requesting an increase in the weight of reliability and anti-interference capability.

[0068] Based on the above analysis and corrections, the solid-state drive (SSD) calculated its final matching score. Although it initially leaned towards Tier 3, the reliability score of Tier 3 was significantly reduced due to the harsh environment, for example, from 70 points to 30 points. In contrast, Tier 2 storage (MLC), which has been hardened for aerospace applications, has stronger radiation resistance (BCH code hardening) and moderate temperature adaptability. Its overall matching score surpasses that of Tier 3 after weighting, for example, Tier 2 scores 85 points and Tier 3 scores 30 points. Based on this, the system determined that the optimal matching level for the hyperspectral imaging data is Tier 2 storage (MLC), thus ensuring the large-capacity storage requirements while avoiding the risk of data loss caused by radiation and high temperatures.

[0069] In step S132, the controller obtains the matching scores of the data to be stored—such as the "real-time telemetry log of the attitude control system"—at three levels: 95 points for the first storage level (SLC), 70 points for the second storage level (MLC), and 40 points for the third storage level (TLC). At the same time, the controller marks the current status of each level and detects that although the first storage level has the highest score, its remaining available space is only 5%, which is in a "sub-healthy" state with high occupancy. The second storage level has sufficient remaining space and is idle. Although the third storage level has the largest space, it is undergoing background garbage collection, resulting in a high response delay.

[0070] Solid-state drives (SSDs) are cross-matched with the current preset optimization goals. Assuming the spacecraft is currently in the critical mission phase of "attitude maneuvering adjustment," the preset optimization goals are set as a dual weighted average of "performance priority" and "reliability priority." During the cross-matching process, the system finds that although the first storage level is space-constrained, its sub-millisecond low latency and high anti-interference capability match the current optimization goals. Through weighted algorithm calculation, even considering space pressure, the comprehensive benefit index of the first storage level is still higher than that of other levels, especially its high reliability attribute, which is given a very high weight during the maneuvering phase.

[0071] The system generates the optimal migration instruction based on the storage constraints of the attitude control log. These constraints explicitly require that "data retention time be no less than 72 hours" and "read latency be less than 1ms," which directly excludes the third storage level, which is busy with garbage collection and has fluctuating latency. To address the issue of limited space in the first storage level, the controller adds a "pre-emptive" operation when generating the migration instruction. This triggers a local cold data eviction instruction, which migrates a batch of early non-critical archived data from the first storage level to the second storage level, thereby freeing up sufficient physical blocks. The final optimal data migration instruction points to the emptied area in the first storage level, ensuring that the attitude control log is written with the highest reliability. This achieves steady-state operation of the storage system and data security under high-maneuverability operation scenarios of spacecraft.

[0072] In step S141, assuming the controller receives an instruction to migrate "critical telemetry data during the orbit insertion phase" from the high-performance first storage tier (SLC) to the high-capacity second storage tier (MLC), the controller immediately responds to the instruction, suspends the current defragmentation process, and activates the internally integrated "intelligent data transfer engine," which takes over bus control and prepares to establish a high-speed data transfer channel between the SLC storage unit and the MLC storage unit.

[0073] The solid-state drive's data transfer engine begins performing a lossless transfer operation. The engine reads the telemetry data from the physical block in the SLC region via the internal bus, and performs ECC error correction checks in real time during the read to eliminate any possible single-event upset (SEU) errors. The engine writes the cleaned data to the target physical block in the MLC region. To ensure high reliability in aerospace environments, the engine immediately initiates a "read-back verification" process after the write is complete, which involves reading the data just written to the MLC and comparing it byte by byte with the original data in the SLC. Only after confirming that the data bits are completely identical is the transfer considered successful, thus ensuring the integrity and security of the data during the migration from a high-interference-resistance level to a normal capacity level.

[0074] The SSD controller performs a global data mapping table update. After confirming the data is correct, the controller accesses the FTL cache table in RAM and modifies the physical pointers corresponding to logical addresses 0x1000 to 0x1FFF from SLC physical page PPA_SLC_01 to MLC physical page PPA_MLC_05. The controller writes this mapping change record to the mapping table area of ​​the NAND flash memory to prevent data loss in case of power failure. At this point, the SSD has successfully completed the cross-level data transfer. The next time the host system reads this telemetry data, it will directly retrieve the data from the MLC area, while the physical block corresponding to the SLC area will be marked as "invalid" and await subsequent erasure and recycling. This achieves precise synchronization of storage state at both the logical and physical levels.

[0075] In step S142, it is assumed that a batch of "post-orbit engineering parameter telemetry data" has been migrated from the high-performance first storage level (SLC) to the high-capacity third storage level (TLC) by the decision module; the controller immediately marks the physical characteristics of the target level (TLC), identifies that the level has the physical characteristics of "low charge retention force" and "high raw bit error rate", and the current sensor shows that the chip temperature in this area is high. These characteristics are marked as high-risk conditions.

[0076] Solid-state drives dynamically adjust maintenance strategies based on the aforementioned high-risk characteristics. For the characteristics of TLC level, the controller forcibly adjusts the background data refresh strategy from the default "refresh once every 72 hours" to "actively refresh once every 24 hours" and enables the enhanced LDPC error correction algorithm, increasing the number of soft decision iterations to the maximum value to offset the risk of charge leakage and the impact of high bit error rate under high temperature environment, and ensure the long-term integrity of telemetry data.

[0077] The SSD's adaptive management module intervenes, adjusting the aggressiveness of background garbage collection. Due to the poor random write performance of the TLC level, to avoid subsequent write operations being blocked while waiting for garbage collection, the controller adjusts the garbage collection strategy for this TLC area to "moderately aggressive," that is, actively defragmenting when the system is idle to maintain a high free block level. At the same time, during the data transfer process involved in garbage collection, the controller adaptively selects high-intensity error correction protection that matches the target level to prevent data corruption caused by write interference during data block migration.

[0078] Through the aforementioned series of physical adaptations and strategy adjustments, the solid-state drive determined the final storage form of the telemetry data at the TLC level, generating a data hierarchical storage sequence that includes physical address, refresh cycle (24H), error correction level, and retrieval strategy. This sequence ensures that even if the data is stored in the relatively less reliable TLC medium, through enhanced management at the software level, it can still meet the stringent requirements of aerospace missions for long-term preservation and highly reliable reading of engineering parameter telemetry data, achieving a balance between storage capacity and data integrity.

[0079] Please see Figure 7 The data tiered storage system for aerospace-grade solid-state drives is applied to the aforementioned data tiered storage method for aerospace-grade solid-state drives; the data tiered storage system for aerospace-grade solid-state drives includes: Storage module 21 is used to integrate multiple storage tiers within a solid-state drive, establish intelligent data migration channels between different storage tiers, and perform differentiated configurations for multiple storage tiers, so that different storage tiers present a gradient distribution from performance to capacity. The data recognition module 22 is used to acquire the data to be stored, collect the access characteristics of the data to be stored in real time, and collect the context information of the space mission. Based on the recognition of the context information of the space mission, the attribute labels of the data to be stored are determined. Data migration module 23 is used to input access features and attribute labels into a pre-built storage strategy evaluation space, and determine the matching score of the data to be stored under different storage levels in the storage strategy evaluation space in combination with the current on-orbit environment status and system load of the spacecraft, and further generate data migration instructions in combination with the corresponding optimization target content. The data tiered storage module 24 is used to respond to migration decision instructions, move the data to be stored from the original storage tier to the target storage tier, update the global data mapping table, and dynamically adjust the data refresh strategy and error correction content according to the physical characteristics of the target storage tier, thereby determining the data tiered storage sequence of the solid-state drive.

[0080] It should be noted that although multiple modules are mentioned in the detailed description above, this division is not mandatory; the above description is only a specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the protection scope of the present invention; therefore, the protection scope of the present invention should be determined by the protection scope of the claims.

Claims

1. A data tiered storage method for solid-state drives used in aerospace applications, characterized in that, include: Multiple storage tiers are integrated within the solid-state drive, an intelligent data migration channel is established between the various storage tiers, and the multiple storage tiers are configured differently, so that the different storage tiers present a gradient distribution from performance to capacity. Acquire the data to be stored, collect the access characteristics of the data to be stored in real time, and collect the context information of the space mission. Determine the attribute labels of the data to be stored based on the identification of the context information of the space mission. The access features and attribute labels are input into the pre-built storage strategy evaluation space. Within the storage strategy evaluation space, the matching score of the data to be stored under different storage levels is determined by combining the current on-orbit environment status of the spacecraft and the system load. The corresponding optimization target content is then combined to generate data migration instructions. In response to migration decision instructions, the data to be stored is moved from the original storage tier to the target storage tier, and the global data mapping table is updated. At the same time, the data refresh strategy and error correction content are dynamically adjusted according to the physical characteristics of the target storage tier, thereby determining the data tiered storage sequence of the solid-state drive.

2. The data tiered storage method for solid-state drives used in aerospace applications according to claim 1, characterized in that, The process involves integrating multiple storage tiers within a solid-state drive (SSD), establishing intelligent data migration channels between these tiers, and configuring these tiers differently to create a gradient distribution from performance to capacity. This includes: For solid-state drives (SSDs) used in aerospace applications, multiple storage tiers integrated within the SSD are marked. These storage tiers include a first storage tier for performance, a second storage tier for capacity, and a third storage tier for cost.

3. The data tiered storage method for solid-state drives used in aerospace applications according to claim 2, characterized in that, The method of integrating multiple storage tiers within a solid-state drive, establishing intelligent data migration channels between these tiers, and configuring the tiers differently to create a gradient distribution from performance to capacity across them, also includes: A smart data migration channel based on a high-speed interconnect bus is established between the first, second, and third storage tiers. The first, second, and third storage tiers are configured with differentiated physical configurations for aerospace applications. By adjusting the error correction strength, data refresh rate, and voltage threshold, the different storage tiers exhibit a continuous gradient distribution from performance to capacity.

4. The data tiered storage method for solid-state drives used in aerospace applications according to claim 1, characterized in that, The process involves acquiring data to be stored, collecting access characteristics of the data in real time, and simultaneously collecting context information from the space mission. Based on the identification of the space mission's context information, the attribute tags of the data to be stored are determined, including: The system acquires data to be stored in real time, identifies multiple access contents based on the identification of the data to be stored, and determines the access characteristics of the data to be stored by combining a dynamic monitoring mechanism. The access characteristics include at least temporal locality, spatial locality, access frequency, and read / write rate.

5. The data tiered storage method for solid-state drives used in aerospace applications according to claim 4, characterized in that, The process of acquiring data to be stored, collecting access characteristics of the data to be stored in real time, and simultaneously collecting context information of the space mission, and determining the attribute tags of the data to be stored based on the identification of the space mission context information, also includes: The context information of the space mission is collected in real time through the spacecraft mission bus interface. The context information includes mission phase identifiers, payload types and command priorities. The context information is identified and parsed to determine the attribute tags of the data to be stored. The attribute tags are used to characterize the business importance, security and confidentiality level and timeliness of the data in the mission process.

6. The data tiered storage method for solid-state drives used in aerospace applications according to claim 1, characterized in that, The process involves inputting access features and attribute labels into a pre-constructed storage strategy evaluation space, and within that space, determining the matching score of the data to be stored at different storage levels based on the spacecraft's current on-orbit environment and system load. Further, it combines this with the corresponding optimization target content to generate data migration instructions, including: The system acquires access features and attribute labels, inputs them into a pre-built storage strategy evaluation space, and combines a multi-objective decision-making mechanism to determine decision content in different dimensions. Furthermore, it performs multi-dimensional feature analysis based on the spacecraft's current on-orbit environment and system load to determine the matching score of the data to be stored at different storage levels.

7. The data tiered storage method for solid-state drives used in aerospace applications according to claim 6, characterized in that, The process of inputting access features and attribute labels into a pre-constructed storage strategy evaluation space, determining the matching score of the data to be stored at different storage levels within the storage strategy evaluation space based on the current on-orbit environment status and system load of the spacecraft, and further generating data migration instructions based on the corresponding optimization target content, also includes: The system obtains the matching scores of the data to be stored under different storage levels and marks the level of the corresponding storage level. It then performs cross-matching in combination with the preset optimization target content. In the matching process, it generates the optimal data migration instructions based on the storage constraints of the data to be stored, so as to realize the steady-state data storage of the spacecraft in the current usage scenario.

8. The data tiered storage method for solid-state drives used in aerospace applications according to claim 1, characterized in that, The response migration decision command moves the data to be stored from the original storage tier to the target storage tier and updates the global data mapping table. Simultaneously, it dynamically adjusts the data refresh strategy and error correction content based on the physical characteristics of the target storage tier, thereby determining the data tiered storage sequence for the solid-state drive, including: In response to data migration instructions, the internal data transfer engine is invoked to transfer the data to be stored from the original storage level to the target storage level without loss. After the transfer is completed, the global data mapping table is updated in real time to ensure the consistency between logical addresses and physical addresses.

9. The data tiered storage method for solid-state drives used in aerospace applications according to claim 8, characterized in that, The response migration decision instruction moves the data to be stored from the original storage tier to the target storage tier and updates the global data mapping table. Simultaneously, it dynamically adjusts the data refresh strategy and error correction content based on the physical characteristics of the target storage tier, thereby determining the data tiered storage sequence for the solid-state drive. The method also includes: The physical characteristics of the target storage tier are marked, and the data refresh strategy and error correction content are dynamically adjusted according to the physical characteristics of the target storage tier. At this time, the aggressiveness of background garbage collection is adaptively adjusted and the corresponding error correction intensity is adaptively selected, so as to determine the data hierarchical storage sequence of the solid-state drive while ensuring data integrity.

10. A data tiered storage system for solid-state drives used in aerospace applications, characterized in that, The data tiered storage system for solid-state drives used in aerospace applications is applied to the data tiered storage method for solid-state drives used in aerospace applications as described in any one of claims 1-9; the data tiered storage system for solid-state drives used in aerospace applications includes: The storage module is used to integrate multiple storage tiers within a solid-state drive, establish intelligent data migration channels between different storage tiers, and perform differentiated configurations for multiple storage tiers, so that different storage tiers present a gradient distribution from performance to capacity. The data recognition module is used to acquire the data to be stored, collect the access characteristics of the data to be stored in real time, and collect the context information of the space mission. Based on the recognition of the context information of the space mission, the attribute labels of the data to be stored are determined. The data migration module is used to input access features and attribute labels into a pre-built storage strategy evaluation space, and within the storage strategy evaluation space, combine the current on-orbit environment status of the spacecraft and the system load to determine the matching score of the data to be stored under different storage levels, and further combine the corresponding optimization target content to generate data migration instructions. The data tiered storage module is used to respond to migration decision commands, move the data to be stored from the original storage tier to the target storage tier, update the global data mapping table, and dynamically adjust the data refresh strategy and error correction content according to the physical characteristics of the target storage tier, thereby determining the data tiered storage sequence of the solid-state drive.

Images