A dual-domain cooperative storage health management device and method for on-board flash
By employing a dual-domain collaborative storage health management approach, combined with RS decoding-based computation and health grading decision-making, a collaborative architecture between the execution and decision-making sides is constructed. This approach solves the reliability problem of onboard Flash storage media in a radiation environment, enabling autonomous perception and proactive management, and extending the lifespan of the storage system.
Patent Information
- Application Number
- CN202611298415.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-26
- Publication Date
- 2026-11-06
AI Technical Summary
Existing technologies lack the ability to autonomously sense the degradation trend of onboard Flash storage media in orbit. Error correction coding is disconnected from storage health status, wear leveling strategies lack real-time error statistics, and there is a lack of coordination between time-domain degradation trends and spatial-domain wear distribution. The lack of a closed-loop management system across the entire chain leads to insufficient reliability of storage systems in radiation environments.
A dual-domain collaborative storage health management method is adopted. Real-time error statistics are obtained through RS decoding and associated computation. The degradation rate is assessed based on the global average bit error rate, dynamic wear leveling and health classification decision-making are carried out to achieve autonomous perception and proactive intervention. A collaborative architecture between the execution side (Flash FPGA) and the decision-making side (floating point DSP) is constructed to form a closed-loop management system.
It significantly extends the reliable service life of the spaceborne Flash storage system. Through joint evaluation in the time and space domains, it dynamically adjusts error correction coding and writing strategies to form a positive feedback closed loop, thereby improving the radiation resistance of the storage system.
Smart Images
Figure CN123047883A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of spaceborne electronic systems and data storage reliability technology, and in particular to a dual-domain collaborative storage health management device and method for spaceborne Flash memory. Background Technology
[0002] To reduce manufacturing costs, commercial spaceflight widely adopts industrial-grade Nor Flash memory as the onboard storage medium. Compared to traditional radiation-hardened devices, industrial-grade Flash has a significant cost advantage, but it faces serious reliability challenges in the space radiation environment.
[0003] The damage caused by space radiation to Flash memory mainly manifests in two different failure mechanisms: The first type is gradual degradation caused by the total ionization dose effect. High-energy particles in space (such as protons, electrons, and rays) accumulate trap charges in the floating gate oxide layer of the Flash, leading to floating gate charge leakage and a monotonically increasing bit error rate over time. This is a slow but irreversible process.
[0004] The second type is sudden failures caused by single-event effects. When high-energy heavy ions or protons strike a memory cell, they may instantaneously change the charge state of the floating gate cell, causing data bits to flip randomly. This type of failure is characterized by its sporadic and random nature, making it difficult to predict in advance.
[0005] The core contradiction between these two failure mechanisms lies in the fact that traditional aerospace relies on expensive radiation-hardened devices to resist radiation damage, while commercial aerospace uses low-cost industrial-grade Flash memory that lacks the ability to autonomously sense degradation trends and proactively intervene in health management on-orbit. Once the storage medium fails due to radiation accumulation, it may lead to the loss of function of the entire onboard system.
[0006] In existing technologies, the reliability of onboard storage systems mainly relies on the following methods: (1) Static error correction coding technology. Existing solutions use error correction codes such as Reed-Solomon (hereinafter referred to as RS code), BCH code, or LDPC code to provide fixed error correction capabilities. However, this error correction capability is fixed at the design stage and cannot be dynamically adjusted with changes in the health status of the storage medium. When the storage medium degrades to a certain extent, the fixed error correction capability will be unable to correct the continuously increasing new errors. This type of solution belongs to "passive error correction" rather than "active management" and lacks linkage with the dynamic health status of the storage medium.
[0007] (2) Storage wear leveling technology. Some solutions balance the wear level of each storage area by distributing data during writes. However, existing wear leveling strategies are usually based on simple polling or random selection, and do not use real-time error statistics of each partition as a scheduling basis, so they cannot actively guide write operations to the partition with the best current health status.
[0008] (3) Spacecraft health management research. Existing research on spacecraft fault prediction and health management (hereinafter referred to as PHM) is mostly focused on mechanical components and power systems, with very little in-depth research on degradation modeling and lifetime prediction of Flash storage media. Even if there are solutions involving storage systems, they fail to cover both the temporal trend and spatial distribution of Flash degradation at the same time, and have not established a collaborative feedback mechanism between the evaluation results of the two domains.
[0009] In summary, the existing technology has the following problems: (1) Lack of online perception and prediction capabilities for the degradation trend of storage media. Ground personnel cannot obtain the real-time health status of the storage system on orbit, let alone predict its remaining effective lifespan, and can only respond passively after a failure occurs.
[0010] (2) Error correction coding and storage health status are disconnected. The error correction code parameters are fixed, making it impossible to use efficient and lightweight coding during the healthy period and switch to strong error correction coding during the degradation period, resulting in both resource waste and insufficient capacity.
[0011] (3) The wear leveling strategy lacks health status awareness. The write scheduling does not take the real-time error statistics of each partition as a basis, and cannot effectively slow down the global degradation rate.
[0012] (4) There is a lack of coordination between the time domain degradation trend and the spatial domain wear distribution. Existing solutions either focus only on life prediction or only on wear equilibrium. The two are independent of each other and fail to form a positive feedback loop of "cost reduction → life extension → early warning → aggressive scheduling".
[0013] (5) Lack of closed-loop management from perception to execution. Without a complete closed-loop architecture of "perception → assessment → decision → execution", it is impossible to automatically trigger execution actions such as code switching, backup switching, and isolation protection based on health assessment results. Summary of the Invention
[0014] The purpose of this invention is to provide a dual-domain collaborative storage health management device and method for spaceborne Flash memory, enabling industrial-grade Flash memory to autonomously sense degradation trends and proactively intervene in orbit, achieving a reliable service life close to that of radiation-hardened devices with low-cost hardware.
[0015] To achieve the above objectives, this invention provides a dual-domain collaborative storage health management method for onboard Flash memory, comprising: By using RS decoding and associated calculation, real-time error statistics and global average bit error rate of each partition of the storage medium can be obtained; The degradation rate of the storage medium is assessed based on the historical trend of the global average bit error rate to estimate its remaining effective lifetime. Based on the cumulative error count of each partition, a mask value for high error zones and availability zones is maintained. During each write operation, the partition with the fewest cumulative errors in the availability zone is selected as the write target to achieve dynamic wear leveling. The dual-domain collaborative arbitrator comprehensively arbitrates the estimated remaining effective lifetime and the saturation of the high error region, and makes a decision according to the preset health classification to output the execution instruction set; According to the execution instruction set, one or more actions including refresh scheduling, RS encoding switching, backup switching, and isolation protection are executed in parallel.
[0016] Preferably, the RS decoding accompanying computation includes: deploying two sets of RS codecs, corresponding to a lightweight error correction mode and a strong error correction mode respectively; when Flash data in the storage medium is read, the RS decoding process generates accompanying computation results, which are output as byproducts: real-time error statistics information and global average bit error rate for each partition; for static data that is not actively read, adaptive refresh scheduling is used to periodically trigger readout and verification to achieve full-coverage health monitoring of the storage medium.
[0017] Preferably, the step of assessing the degradation rate of the storage medium based on the historical trend of the global average bit error rate includes: receiving the global average bit error rate through a time-domain evaluator, maintaining a historical data window, tracking the trend of bit error rate changes, in order to assess the current degradation rate and estimate its remaining effective lifetime.
[0018] Preferably, the dynamic wear leveling includes: selecting the partition with the fewest cumulative errors in the available area as the write target through a spatial domain evaluator, and continuously directing write traffic to the low-wear partition to achieve uniform wear across the entire chip.
[0019] Preferably, the preset health classification includes: During the healthy period, lightweight error correction coding and a regular refresh cycle are used; During the sub-healthy period, switch to strong error correction coding and encrypt the refresh cycle; During the warning period, a backup pre-switch is triggered and the availability zone range is tightened. At the end of its lifespan, isolation protection is triggered.
[0020] Preferably, when the remaining effective lifetime is shortened to the threshold of the warning period, the available zone range is tightened by the spatial domain evaluator, that is, the number of available zones is reduced, sacrificing some storage capacity for a longer survival time; after the backup pre-switch is triggered during the warning period, the original risky Flash enters read-only mode until the data migration is completed, and then the isolation protection power-off is triggered; the RS encoding is switched from lightweight error correction mode to strong error correction mode, the error correction capability is improved, and the switch is completed during the refresh interval; the refresh cycle is dynamically adjusted according to the health level, with a regular interval during the healthy period, a shorter interval during the sub-healthy period, and a dense interval during the warning period; the partition to be refreshed for the write target is specified by the evaluation result of the spatial domain evaluator.
[0021] The present invention also provides a dual-domain collaborative storage health management device for onboard Flash memory, which executes the dual-domain collaborative storage health management method for onboard Flash memory as described above, including: The execution-side main control unit, implemented using Flash FPGA, is used to read data from the storage medium, perform RS decoding and calculate the syndrome, output partition error statistics and global average bit error, and receive execution instructions to complete refresh scheduling and RS encoding switching. The decision-side evaluation unit, implemented using a floating-point DSP, is connected to the execution-side main control unit and is used to run the time-domain evaluator, the spatial-domain evaluator, and the dual-domain collaborative arbitrator. It generates health assessments and decisions based on the received partition error statistics and global average bit error rate. The storage unit, connected to the execution-side main control unit, uses dual Nor Flash storage media to form a backup for each other and is used to store data. The actuator unit, connected to the decision-side evaluation unit and the execution-side main control unit, includes an isolation protection switch and a switching switch for performing backup switching and power-off isolation.
[0022] Preferably, the execution-side main control unit integrates a dual-mode RS codec, corresponding to a lightweight error correction mode and a powerful error correction mode, respectively. The execution-side main control unit is connected to the decision-side evaluation unit via a parallel bus for uplink transmission of the partition error statistics and global average bit error rate, and downlink reception of the execution instructions. The execution-side main control unit is connected to the storage unit and the service load unit via a parallel bus for data read / write and refresh operations, and for writing corrected data to the service load unit during refresh operations. The service load unit is implemented using SRAM FPGA, used to carry onboard service logic, and is the final user of the data inside the Flash memory.
[0023] Preferably, the time-domain evaluator maintains a historical data window based on the global average bit error rate, tracks the trend of bit error rate changes over time, evaluates the current degradation rate, and calculates the remaining effective lifetime using the maximum number of bit errors that the strong error correction mode can correct as the lifetime end threshold; the spatial domain evaluator divides the storage medium into multiple partitions, maintains the cumulative error count of each partition, and defines high error areas and availability areas. Each write operation selects the partition with the fewest cumulative errors in the availability area as the write target.
[0024] Preferably, the dual-domain collaborative arbitrator makes a decision based on a preset health level, combining the estimated remaining effective lifetime and the saturation of the high-error partition, and converts the decision result into the execution instruction set and sends it to the execution-side main control unit; when adjusting the refresh cycle, the execution-side main control unit dynamically adjusts the refresh interval according to the health level; a normal interval is used during the healthy period, the interval is shortened during the sub-healthy period, and a dense interval is used during the warning period; during the sub-healthy period, synchronous refresh of the dual-chip storage medium Nor Flash is initiated.
[0025] Compared with the prior art, the present invention has the following advantages and technical effects: 1. This invention adopts a dual-chip collaborative architecture of "execution side + decision side": the execution side (Flash FPGA) is responsible for RS encoding and decoding, error statistics and execution actions; the decision side (floating-point DSP) is responsible for running the health management algorithm and issuing instructions. The two domains independently evaluate and make collaborative decisions, complementarily covering the temporal trend and spatial distribution characteristics of Flash degradation.
[0026] 2. This invention jointly evaluates the health status of Flash memory from both temporal and spatial dimensions. The temporal domain evaluation tracks the trend of bit error rate changes over time, assesses the degradation rate, and estimates the remaining effective lifespan. The spatial domain evaluation divides the storage space into several partitions, counts the number of errors in each partition, and prioritizes writing to partitions with fewer errors to even out wear across the entire chip. After converging information from both dimensions, the system makes a comprehensive judgment on the health status.
[0027] 3. This invention employs a health-level driven adaptive execution strategy. Based on the dual-domain collaborative evaluation results, the storage health status is divided into a healthy period, a sub-healthy period, a warning period, and a lifespan end period. Different levels correspond to different execution strategies: the healthy period uses lightweight error correction coding and a regular inspection cycle; the sub-healthy period switches to strong error correction coding and encrypted inspection; the warning period actively switches to a backup chip; and the lifespan end period implements isolation protection and reports to the ground.
[0028] 4. This invention establishes a dual-domain closed-loop feedback mechanism. In the spatial domain, by directing writes to low-wear partitions, the rate of increase in the global bit error rate is effectively slowed, providing an additional margin for the remaining lifetime predicted in the time domain. When the remaining lifetime in the time domain shortens to the warning threshold, the decision unit instructs the spatial domain to tighten the available partition range, sacrificing some storage capacity for a longer lifetime. This positive feedback chain makes the overall gain exceed the simple sum of the effects of two single domains.
[0029] 5. This invention adopts an architecture in which the execution side (Flash FPGA) and the decision-making side (floating-point DSP) work together. The Flash FPGA is responsible for encoding and decoding and statistical reporting with high real-time requirements, while the floating-point DSP is responsible for computationally intensive health assessment algorithms. The division of labor is clear and each performs its own duties, which reduces the software complexity and resource consumption of a single chip. Attached Figure Description
[0030] To more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.
[0031] Figure 1 This is a hardware structure block diagram of the dual-domain collaborative storage health management device provided by the present invention.
[0032] Figure 2 The flowchart illustrates the dual-domain collaborative storage health management method provided by this invention.
[0033] Figure 3 This is a diagram illustrating the overall framework of dual-domain collaboration provided by the present invention.
[0034] Figure 4 This is a diagram illustrating the dual-domain closed-loop feedback mechanism provided by the present invention. Detailed Implementation
[0035] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0036] This invention enables industrial-grade Flash memory to autonomously sense degradation trends and proactively intervene in the on-orbit environment by constructing a dual-domain collaborative health management system, thereby achieving a reliable service life close to that of radiation-hardened devices with low-cost hardware.
[0037] The system employs two chips working collaboratively: the Flash FPGA (execution side) interacts directly with the Flash memory, periodically running RS codec to write / read data from the Flash for external controller use, periodically performing RS codec to write / read data from the Flash for self-testing and counting errors in each region, executing refresh and RS code mode switching actions, and executing Flash switching actions; the floating-point DSP (decision side) runs the health management algorithm, receives error statistics, assesses the Flash's health status, and issues decision instructions. Each chip performs its specific function: the execution side ensures real-time performance, while the decision side ensures algorithm accuracy. Through the collaborative work of the execution side (Flash FPGA) and the decision side (floating-point DSP), a fully closed-loop autonomous health management system—from "perception → assessment → decision → execution"—is achieved, significantly extending the effective service life of the storage system.
[0038] As a first aspect of the present invention, a dual-domain collaborative storage health management method for spaceborne applications is provided, the method comprising the following steps: Step S1: Construct a closed-loop management architecture of "Perception → Evaluation → Decision → Execution". The architecture includes a three-layer logical structure: the perception layer is completed by the execution side (Flash FPGA) for RS decoding, accompanying calculation and partition error statistics; the evaluation and decision layer is completed by the decision side (floating-point DSP) for running the time domain evaluator, spatial domain evaluator and dual-domain collaborative arbitrator; the execution layer is completed by the execution side and peripheral circuits for refresh scheduling, RS encoding switching, backup switching, isolation triggering and other actions.
[0039] Step S2: The perception layer collects real-time error statistics data of the storage medium. Two RS codecs are deployed in the Flash FPGA, corresponding to lightweight error correction mode and strong error correction mode, respectively. When Flash data is read, the RS decoding process generates accompanying calculation results, which are output as byproducts: error statistics information for each partition and the global average bit error rate, and reported to the floating-point DSP. For static data that is not actively read, readout and verification are periodically triggered through subsequent refresh scheduling to achieve full-coverage health monitoring of the storage array.
[0040] Step S3: The time-domain evaluator assesses the degradation trend and calculates the remaining effective lifetime. The decision-making side (floating-point DSP) receives the global average bit error rate reported by the execution side, maintains a historical data window of a certain length, tracks the trend of bit error rate changes over time, and assesses the current degradation rate. Using the maximum number of bit errors that can be corrected in brute-force error correction mode as the lifetime end threshold, the remaining effective lifetime (RUL) that the Flash can still operate normally under the current degradation rate is calculated. This evaluation result is output to the dual-domain collaborative arbitrator.
[0041] Step S4: The spatial domain evaluator implements error-aware dynamic wear leveling. The Flash storage space is divided into multiple partitions, and the decision-making side maintains the cumulative error count for each partition. High-error partitions (sensitive areas) and available partitions (active areas) are defined. During each write operation, the spatial domain evaluator selects the partition with the fewest cumulative errors in the active area as the write target, continuously directing write traffic to the low-wear partition to achieve uniform wear across the entire chip. The partition wear distribution results are output to the dual-domain collaborative arbitrator.
[0042] Step S5: The dual-domain collaborative arbitrator performs health grading and decision mapping. The arbitrator integrates the received time-domain RUL prediction and the spatial-domain sensitive area saturation, and makes a decision according to the preset health grading matrix. The grading logic is as follows: Healthy period: Ample remaining lifespan, low proportion of sensitive areas. The system adopts lightweight error correction coding (lower error correction capability but fast decoding speed) and a regular refresh cycle, and write operations are performed according to the principle of minimizing errors in the active area.
[0043] Sub-health period: Remaining lifespan shortens to a certain range, and the proportion of sensitive areas increases. The system switches to strong error correction encoding (error correction capability doubles but decoding speed decreases), while encrypting the refresh cycle and limiting the write speed.
[0044] Warning period: The remaining lifespan is relatively short, and the proportion of sensitive areas is high. Based on switching strong error correction encoding and encrypted refresh, the system actively triggers a backup pre-switch to the backup Flash, while tightening the available partition range of the instruction space domain (reducing the number of active areas, sacrificing some capacity for a longer lifespan).
[0045] End of life: The remaining lifespan is about to expire. The system triggers isolation protection, completely powering off and isolating the risky flash memory, and reporting to the ground.
[0046] The arbitrator transforms the decision into a set of execution instructions and sends them down to the execution layer.
[0047] Step S6: The execution layer completes multi-level adaptive execution actions. After receiving the instruction, the execution layer executes the following actions in parallel: (1) Adaptive refresh scheduling: The refresh cycle for Flash is dynamically adjusted according to the health level (normal interval is used during the healthy period → the interval is shortened during the sub-healthy period → dense interval is used during the warning period). The refresh target partition is specified by the spatial domain evaluator. During the sub-healthy period, dual Flash synchronous refresh is started.
[0048] (2) Adaptive switching of dual-mode RS encoding: switching from lightweight error correction mode to powerful error correction mode, the error correction capability is doubled, and the switching is completed during the refresh interval.
[0049] (3) Flash pre-switching and isolation protection: During the early warning period, it actively switches to the backup Flash, and the original risk Flash triggers the isolation protection to cut off power; at the end of its life, it is directly isolated and scrapped.
[0050] Step S7: Dual-domain closed-loop positive feedback loop. The above steps constitute a complete closed-loop feedback loop: the minimum error replacement in the spatial domain will be written to the low-wear partition (step S4), slowing down the rise in the global bit error rate and effectively extending the remaining lifetime predicted in the time domain (step S3); when the remaining lifetime is shortened to the warning threshold, the arbitrator instructs the spatial domain to tighten the active area range (step S5). The joint evaluation of the two domains drives all execution actions (step S6). The execution actions (refresh, write) in turn affect the error statistics of each partition (step S2), forming a continuous closed-loop feedback loop.
[0051] As a second aspect of the present invention, a dual-domain collaborative storage health management device for spaceborne applications is provided, the device comprising: The execution side main control unit is implemented using Flash FPGA and integrates a dual-mode RS codec for: (1) performing RS decoding on Flash read data and calculating the syntactic expression, outputting partition error statistics and global average bit error; (2) receiving execution instructions issued by the decision side and completing operations such as refresh scheduling and RS encoding switching.
[0052] The decision-side evaluation unit, implemented using a floating-point DSP, is used for: (1) running a time-domain evaluator to calculate the RUL based on the historical trend of the bit error rate; (2) running a spatial-domain evaluator to achieve dynamic wear leveling based on minimum error replacement; and (3) running a dual-domain collaborative arbitrator to make decisions based on the health classification and issue execution instructions.
[0053] The storage unit uses two Nor Flash chips in an equal architecture, serving as backups for each other, to store onboard system programs and data.
[0054] The service load unit, implemented using SRAM FPGA, is used to carry onboard service logic, accept refresh data from the execution side, and is the final user of the data inside the Flash memory.
[0055] The actuator unit includes an isolation protection switch and a switching switch, used to perform power-off isolation of the risky Flash and switching of the backup Flash under the instruction of the arbitrator.
[0056] Furthermore, the execution-side master control unit is connected to the decision-side evaluation unit via a parallel bus for uplink transmission of error statistics and downlink reception of execution instructions.
[0057] Furthermore, the execution-side main control unit is connected to the dual Nor Flash chips via a parallel bus for data read / write and refresh operations.
[0058] Furthermore, the execution-side main control unit and the service load unit are connected via a parallel bus, which is used to write the corrected data into the service load unit during the refresh operation.
[0059] This embodiment specifically provides a dual-domain collaborative storage health management device and method for spaceborne applications.
[0060] like Figure 1 As shown, the device includes a Flash FPGA (execution-side main control unit), a floating-point DSP (decision-side evaluation unit), dual Nor Flash chips (memory unit), an SRAM FPGA (service load unit), an isolation protection switch, and a switching switch (execution mechanism unit). The execution side is connected to the decision side via a parallel bus, and is also connected to the dual Nor Flash chips and the service load unit via a parallel bus.
[0061] like Figure 2 As shown, the method includes the following steps: Step S1: The RS decoder in the execution-side FPGA decodes the Flash read data, performs adjoint calculation to generate partition error statistics and global average bit error, and reports to the decision-side DSP.
[0062] Step S2: The decision-side time domain evaluator receives the global average bit error rate, maintains a historical data window, tracks the trend of bit error rate changes, assesses the degradation rate, and calculates the remaining effective lifetime.
[0063] Step S3: The decision-side spatial domain evaluator receives error statistics for each partition, maintains high-error zone masks and availability zone masks, and selects the partition with the fewest errors in the availability zone as the write target for each write operation.
[0064] Step S4: The arbitrator comprehensively assesses the remaining lifetime and the saturation of the sensitive area, makes a decision based on the health classification matrix, and outputs the execution instruction set.
[0065] Step S5: The execution layer receives instructions and performs actions such as refresh cycle adjustment, RS encoding switching, backup switching, and isolation protection in parallel.
[0066] Step S6: The execution of actions (refresh, write) affects the partition error statistics and is fed back to step S1, forming a continuous closed loop.
[0067] like Figure 3As shown, the overall framework of the dual-domain collaborative system adopts a four-layer structure from top to bottom: the first layer is the sensing input layer (global average bit error rate and partition error vector); the second layer is the evaluation layer (the time domain evaluator outputs the remaining lifetime estimate, and the spatial domain evaluator outputs the partition wear distribution); the third layer is the decision layer (the dual-domain collaborative arbitrator outputs the execution instruction set); and the fourth layer is the execution layer (refresh scheduling, RS encoding switching, Flash switching, and isolation protection). A dashed loop leads back up from the bottom of the execution layer to the sensing layer, labeled "closed-loop feedback (error statistics update)".
[0068] like Figure 4 As shown, the dual-domain closed-loop feedback mechanism adopts a structure of left and right domains plus bottom execution: the left spatial domain achieves "cost throttling and life extension" through "minimum error replacement → slowing down the rise of global bit error rate → reducing the variance of partition wear", and feeds back to the right to the time domain (extending remaining lifespan); when the remaining lifespan is shortened to the warning threshold, the right time domain feeds back to the left to the spatial domain ("aggressive scheduling", tightening the range of available partitions, sacrificing capacity for survival time); the states of the left and right domains are simultaneously merged into the bottom "dual-domain collaborative execution" module, triggering actions such as RS switching, backup switching, and isolation protection.
[0069] Preferably, the two RS encoding and decoding modes correspond to lightweight error correction (fast decoding speed and low resource consumption) and powerful error correction (doubled error correction capability and decreased decoding speed), respectively, and the decision-making side adaptively switches between the two according to the health level.
[0070] Preferably, spatial wear leveling divides the Flash memory into multiple partitions, and during writing, the partition with the fewest errors is selected first.
[0071] Preferably, after the backup pre-switch is triggered during the early warning period, the original Flash enters read-only mode until the data migration is completed, and then the isolation protection power-off is triggered.
[0072] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.
Claims
1. A dual-domain collaborative storage health management method for spaceborne Flash, characterized in that, include: By using RS decoding and associated calculation, real-time error statistics and global average bit error rate of each partition of the storage medium can be obtained; The degradation rate of the storage medium is assessed based on the historical trend of the global average bit error rate to estimate its remaining effective lifetime. Based on the cumulative error count of each partition, a mask value for high error zones and availability zones is maintained. During each write operation, the partition with the fewest cumulative errors in the availability zone is selected as the write target to achieve dynamic wear leveling. The dual-domain collaborative arbitrator comprehensively arbitrates the estimated remaining effective lifetime and the saturation of the high error region, and makes a decision according to the preset health classification to output the execution instruction set; According to the execution instruction set, one or more actions including refresh scheduling, RS encoding switching, backup switching, and isolation protection are executed in parallel.
2. The dual-domain collaborative storage health management method for spaceborne Flash of claim 1, wherein, The RS decoding accompanying computation includes: deploying two sets of RS codecs, corresponding to lightweight error correction mode and strong error correction mode respectively; when Flash data in the storage medium is read, the RS decoding process generates accompanying computation results, which are output as byproducts: real-time error statistics information and global average bit error rate for each partition; for static data that is not actively read, adaptive refresh scheduling is used to periodically trigger readout and verification to achieve full-coverage health monitoring of the storage medium.
3. The dual-domain collaborative storage health management method for spaceborne Flash of claim 1, wherein, The method of assessing the degradation rate of the storage medium based on the historical trend of the global average bit error rate includes: receiving the global average bit error rate through a time domain evaluator, maintaining a historical data window, tracking the trend of bit error rate changes, assessing the current degradation rate, and estimating its remaining effective lifetime.
4. The dual-domain collaborative storage health management method for spaceborne Flash of claim 1, wherein, The dynamic wear leveling process includes: selecting the partition with the fewest cumulative errors in the available area as the write target through the spatial domain evaluator, and continuously directing the write traffic to the low-wear partition to achieve uniform wear across the entire chip.
5. The dual-domain collaborative storage health management method for spaceborne Flash of claim 4, wherein, The preset health classification includes: During the healthy period, lightweight error correction coding and a regular refresh cycle are used; During the sub-healthy period, switch to strong error correction coding and encrypt the refresh cycle; During the warning period, a backup pre-switch is triggered and the availability zone range is tightened. At the end of its lifespan, isolation protection is triggered.
6. The dual-domain collaborative storage health management method for spaceborne Flash of claim 5, wherein, When the remaining effective lifetime is shortened to the threshold of the warning period, the availability zone range is tightened by the spatial domain evaluator, that is, the number of availability zones is reduced, in order to sacrifice some storage capacity for a longer survival time; after the backup pre-switch is triggered during the warning period, the original risky Flash enters read-only mode until the data migration is completed, and then the isolation protection power-off is triggered. The RS encoding is switched from lightweight error correction mode to strong error correction mode, which improves the error correction capability, and the switching is completed during the refresh interval. The refresh cycle is dynamically adjusted according to the health level. A regular interval is used during the healthy period, the interval is shortened during the sub-healthy period, and a dense interval is used during the warning period. The partition to be refreshed for the writing target is specified by the evaluation result of the spatial domain evaluator.
7. A device for satellite Flash-oriented dual in-circuit storage health management, which executes a method for satellite Flash-oriented dual in-circuit storage health management according to any one of claims 1-6, characterized in that, include: The execution-side main control unit, implemented using Flash FPGA, is used to read data from the storage medium, perform RS decoding and calculate the syndrome, output partition error statistics and global average bit error, and receive execution instructions to complete refresh scheduling and RS encoding switching. The decision-side evaluation unit, implemented using a floating-point DSP, is connected to the execution-side main control unit and is used to run the time-domain evaluator, the spatial-domain evaluator, and the dual-domain collaborative arbitrator. It generates health assessments and decisions based on the received partition error statistics and global average bit error rate. The storage unit, connected to the execution-side main control unit, uses dual Nor Flash storage media to form a backup for each other and is used to store data. The actuator unit, connected to the decision-side evaluation unit and the execution-side main control unit, includes an isolation protection switch and a switching switch for performing backup switching and power-off isolation.
8. The dual-domain collaborative storage health management apparatus for spaceborne Flash of claim 7, wherein, The execution-side main control unit integrates a dual-mode RS codec, corresponding to lightweight error correction mode and strong error correction mode respectively. The execution-side main control unit is connected to the decision-side evaluation unit via a parallel bus for uplink transmission of partition error statistics and global average bit error rate, and downlink reception of execution instructions. The execution-side main control unit is connected to the storage unit and the service load unit via a parallel bus for data read / write and refresh operations, and for writing corrected data to the service load unit during refresh operations. The service load unit is implemented using SRAM FPGA, used to carry onboard service logic, and is the final user of the data inside the Flash memory.
9. The dual-domain collaborative storage health management apparatus for spaceborne Flash of claim 7, wherein, The time-domain evaluator maintains a historical data window based on the global average bit error rate, tracks the trend of bit error rate changes over time, evaluates the current degradation rate, and calculates the remaining effective lifetime using the maximum number of bit errors that the strong error correction mode can correct as the lifetime end threshold. The spatial domain evaluator divides the storage medium into multiple partitions, maintains the cumulative error count of each partition, and defines high error zones and availability zones. Each write operation selects the partition with the fewest cumulative errors in the availability zone as the write target.
10. The dual-domain collaborative storage health management apparatus for spaceborne Flash of claim 7, wherein, The dual-domain collaborative arbitrator makes a decision based on a preset health level, combining the estimated remaining effective lifetime and the saturation of the high-error partition, and converts the decision result into the execution instruction set and sends it to the execution-side main control unit; when the execution-side main control unit adjusts the refresh interval according to the health level, the refresh interval is dynamically adjusted. During the healthy period, a regular interval is used; during the sub-healthy period, the interval is shortened; and during the warning period, a dense interval is used. During the sub-healthy period, synchronous refresh of dual-chip storage medium Nor Flash is initiated.