A storage information verification system based on consensus mechanism

Abstract

The present invention relates to the field of distributed storage verification technology, and discloses a storage information verification system based on a consensus mechanism. The system includes consensus network communication, storage feature analysis, verification decision core, timing prediction, decision compensation, and verification execution engine modules. The consensus network communication module connects to distributed storage nodes and receives storage verification requests; the storage feature analysis module extracts multi-dimensional feature vectors of target data from them to form a feature descriptor set; the verification decision core module combines the embedded consensus verification space and dynamic confidence constraint operation to output an initial verification decision plan; the timing prediction module divides the time window to generate a set of predicted feature descriptors; the decision compensation module performs deviation compensation based on this to obtain an optimized plan; the verification execution engine module sends instructions to trigger distributed nodes to perform verification operations. The dynamic verification system constructed by the system adapts to the dynamic characteristics and heterogeneous environments of distributed storage.

Description

Technical Field

[0001] The present invention relates to the field of distributed storage verification technology, and in particular to a storage information verification system based on a consensus mechanism. Background Art

[0002] In distributed storage systems, the surge in the number of nodes and the exponential growth in data volumes pose significant challenges to the authenticity and consistency of stored information. Traditional centralized verification models rely on a single trusted node for data verification. When a node encounters a malicious attack or malfunctions, the entire verification system can be paralyzed, and it struggles to adapt to cross-regional, highly dynamic network environments.

[0003] Existing distributed verification schemes often rely on static consensus mechanisms, making decisions based solely on current data characteristics while ignoring the evolution of data over time. For example, some systems perform verification by simply comparing data hash values, but are unable to account for fluctuations in data characteristics caused by normal updates or transmission delays, leading to increased false positives. Furthermore, most schemes lack effective bias compensation mechanisms. When network topology changes or node computing power becomes uneven, verification decisions are prone to bias, compromising overall system reliability.

[0004] In traditional verification processes, feature extraction is limited in dimension, often focusing only on data integrity characteristics while ignoring multi-dimensional information such as data access frequency, modification history, and node reputation. This results in incomplete verification results. When faced with complex attack methods such as data tampering and Sybil attacks, existing systems struggle to quickly identify abnormal data, causing erroneous data to spread across the network, further exacerbating storage risks. With the integration of edge computing and IoT technologies, the heterogeneity of distributed storage nodes has become increasingly prominent. Hardware performance and network bandwidth vary significantly between nodes, making traditional verification mechanisms difficult to adapt to this heterogeneous environment, resulting in low verification efficiency and wasted resources. Summary of the Invention

[0005] The purpose of the present invention is to provide a storage information verification system based on a consensus mechanism to solve the problems raised in the above background technology.

[0006] To achieve the above objectives, the present invention provides a storage information verification system based on a consensus mechanism, the system comprising:

[0007] Consensus network communication module, used to connect to distributed storage nodes and receive storage verification requests;

[0008] a storage feature analysis module, configured to extract a multi-dimensional feature vector of target storage data based on the storage verification request and generate a storage data feature descriptor set;

[0009] A verification decision core module is used to perform consensus decision operations based on the embedded consensus verification space and dynamic confidence constraints, combined with the stored data feature descriptor set, and output an initial verification decision plan;

[0010] A time series prediction module is used to segment the stored data feature descriptor set into time windows, predict the feature evolution path in the future time window through historical verification records, and generate a predicted feature descriptor set;

[0011] A decision compensation module, configured to perform deviation compensation calculation on the initial verification decision plan based on the prediction feature descriptor set to generate an optimized verification decision plan;

[0012] The verification execution engine module is used to send execution instructions to the consensus network communication module according to the optimized verification decision scheme, triggering the distributed nodes to perform storage verification operations.

[0013] Preferably, upon receiving a storage verification request, the consensus network communication module executes:

[0014] Load the node topology distribution diagram of the target storage data;

[0015] Dynamically generate a data collection strategy based on the node topology distribution map, wherein the data collection strategy includes a collection frequency classification table and a priority weight matrix;

[0016] According to the acquisition frequency classification table and the priority weight matrix, the distributed nodes are scheduled to synchronously transmit and store data slices and metadata check codes.

[0017] Preferably, the storage feature analysis module performs:

[0018] Analyze the physical storage structure and logical access path of the target storage data;

[0019] Constructing a storage topology feature model according to the physical storage structure and the logical access path;

[0020] Inputting the storage data slice into the storage topology feature model to perform space mapping operation and output a storage data feature descriptor set;

[0021] The storage data feature descriptor set includes a block hash fingerprint sequence and a data integrity mark matrix.

[0022] Preferably, the construction process of the storage feature analysis module includes:

[0023] Configuring a cross-node feature analysis library, wherein the cross-node feature analysis library includes a feature extraction rule library for heterogeneous storage devices;

[0024] Loading a historical verification feature sample library, wherein the historical verification feature sample library contains a feature template set of verified stored data;

[0025] Training a topology-aware feature analysis model based on the cross-node feature analysis library and the historical verification feature sample library;

[0026] The topology-aware feature analysis model is incrementally optimized through real-time verification of data streams to generate a final storage feature analysis module.

[0027] Preferably, the incremental parameter optimization process includes:

[0028] Inject simulated verification data streams within the preset test cycle;

[0029] Calculating a feature detection sensitivity index output by the topology-aware feature analysis model;

[0030] When the feature detection sensitivity index is lower than a preset sensitivity threshold, activating the adaptive learning unit;

[0031] The adaptive learning unit extracts incremental feature samples from the real-time verification data stream, updates the historical verification feature sample library and retrains model parameters.

[0032] Preferably, the verification decision core module performs:

[0033] Probabilistically associating and matching the stored data feature descriptor set with the consensus verification space;

[0034] generating a dynamic decision triggering interval according to the matching result, wherein the dynamic decision triggering interval includes a plurality of candidate verification strategies;

[0035] Perform confidence evaluation operations on each candidate verification strategy and output a strategy confidence score;

[0036] Screen candidate verification strategies that meet dynamic confidence constraints and combine them to form an initial verification decision plan.

[0037] Preferably, the probabilistic association matching process includes:

[0038] Traverse the historical decision records in the consensus verification space and extract sample feature descriptors and corresponding verification strategies;

[0039] Calculate the topological feature similarity depth coefficient between the current stored data feature descriptor set and the sample feature descriptor;

[0040] The similar depth coefficients of multiple topological features are weightedly fused according to the preset weight distribution rules;

[0041] When the weighted fusion value exceeds the consensus association threshold, the corresponding verification strategy is added to the association strategy pool.

[0042] Preferably, the decision compensation module performs:

[0043] Inputting the predicted feature descriptor set into the consensus verification space for secondary association matching;

[0044] Generate a compensation decision trigger interval based on the secondary matching result;

[0045] Perform multiple rounds of iterative optimization operations within the compensation decision triggering interval to generate a compensation verification strategy set;

[0046] The compensation verification strategy set is combined with the initial verification decision plan to perform strategy fusion calculation and output an optimized verification decision plan.

[0047] Preferably, the verification execution engine module executes synchronously during execution:

[0048] Monitor the verification response delay and result consistency indicators of distributed nodes;

[0049] generating an execution efficiency evaluation report based on the verification response delay;

[0050] Calculate the verification stability coefficient based on the result consistency index;

[0051] When the verification stability coefficient is lower than the dynamic threshold, the time series prediction module is triggered to regenerate the feature evolution path.

[0052] Preferably, the system further comprises:

[0053] a stability fuse module, configured to continuously receive the verification stability coefficient output by the verification execution engine module;

[0054] When the accumulated abnormal stability signals received within a preset monitoring period exceed the fuse threshold, a system self-check instruction is generated;

[0055] The system self-check instruction triggers the consensus network communication module to suspend new verification requests and start the node status diagnosis process.

[0056] Compared with the prior art, the present invention has the following beneficial effects:

[0057] The consensus network communication module enables efficient connectivity of distributed storage nodes, enabling rapid responses to storage verification requests. This breaks the geographical limitations and single-point dependency of traditional centralized verification, making the verification process more scalable. The storage feature analysis module extracts feature vectors of target storage data from multiple dimensions. The resulting feature descriptor sets cover multiple aspects of data integrity, access patterns, node associations, and more. Compared to single-dimensional feature extraction, these can more comprehensively reflect the true state of the data and reduce misjudgments due to missing information.

[0058] The core verification decision module embeds a consensus verification space and dynamic confidence constraints, and combines multi-dimensional feature descriptors for consensus decision operations. This overcomes the limitations of traditional static decision-making models and can adjust decision parameters based on the real-time network status, making the initial verification decision plan more tailored to the current network environment. The time series prediction module uses time window segmentation technology to predict the evolution path of stored data features. The generated set of predicted feature descriptors captures the changing trends of data over time, allowing verification decisions to no longer be limited to the current state and to proactively respond to future fluctuations in data features.

[0059] The decision compensation module calculates deviation compensation for the initial verification decision plan based on the predicted feature descriptors. This effectively offsets decision deviations caused by factors such as network latency and node performance differences, making the optimized verification decision plan more adaptable. The verification execution engine module triggers verification operations on distributed nodes based on the optimization plan, ensuring that verification instructions can be efficiently executed across heterogeneous nodes, adapting to different hardware performance and network environments, and avoiding resource waste.

[0060] By combining multi-dimensional feature analysis with time series prediction, the entire system enhances its ability to identify complex attacks, enabling timely detection of risks such as data tampering and abnormal access, and reducing the spread of erroneous data. Furthermore, dynamic consensus decision-making and deviation compensation mechanisms enable the system to maintain stable operation despite changes in network topology and fluctuations in node computing power, enhancing the adaptability and flexibility of data verification in distributed storage environments. The coordinated operation of various modules makes the verification process more aligned with the dynamic characteristics of distributed systems, enabling efficient resource allocation in heterogeneous node environments and reducing redundant operations during the verification process. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 This is a diagram showing the working principle of the storage information verification system based on the consensus mechanism of the present invention;

[0062] Figure 2 This is the flowchart of the consensus network communication module;

[0063] Figure 3 Flowchart constructed for the storage feature analysis module;

[0064] Figure 4 Flowchart for verifying the decision-making core module;

[0065] Figure 5 Flowchart of the decision compensation module. DETAILED DESCRIPTION

[0066] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0067] See also Figure 1 The present invention provides a storage information verification system based on a consensus mechanism, which includes: a consensus network communication module, a storage feature analysis module, a verification decision core module, a timing prediction module, a decision compensation module, and a verification execution engine module. The specific implementation is as follows:

[0068] The consensus network communication module connects to distributed storage nodes and receives storage verification requests.

[0069] The storage feature analysis module extracts the multi-dimensional feature vector of the target storage data based on the storage verification request and generates a storage data feature descriptor set.

[0070] The verification decision core module performs consensus decision operations based on the embedded consensus verification space and dynamic confidence constraints, combined with the stored data feature descriptor set, and outputs the initial verification decision plan.

[0071] The time series prediction module divides the stored data feature descriptor set into time windows, predicts the feature evolution path in the future time window through historical verification records, and generates a predicted feature descriptor set.

[0072] The decision compensation module performs deviation compensation calculation on the initial verification decision plan based on the predicted feature descriptor set to generate an optimized verification decision plan.

[0073] The verification execution engine module sends execution instructions to the consensus network communication module according to the optimized verification decision plan, triggering the distributed nodes to perform storage verification operations.

[0074] Example 1: See Figure 1 and Figure 2When receiving a storage verification request, the consensus network communication module first loads the node topology distribution map of the target storage data. The node topology distribution map contains parameters such as the physical location information of the distributed storage nodes, network connection status, data storage capacity, and data transmission bandwidth between nodes. These parameters are obtained and dynamically updated by real-time scanning of the distributed network. Based on the node topology distribution map, the system will dynamically generate a data collection strategy. The collection frequency classification table in the strategy is graded according to the node load, data importance, and historical transmission stability. For example, nodes are divided into three levels: high frequency, medium frequency, and low frequency according to the data access frequency. Different levels correspond to different collection time intervals. The priority weight matrix is ​​constructed based on the data integrity requirements, real-time requirements, and node reliability scores. Each element in the matrix represents the priority weight value of the corresponding node in the specific data slice collection process.

[0075] According to the collection frequency classification table and priority weight matrix, the consensus network communication module schedules distributed nodes to synchronously transmit and store data slices and metadata check codes. During the scheduling process, the system will first detect the current operating status of the node, including CPU usage, memory occupancy, and network latency, to ensure that the nodes participating in the transmission have sufficient processing power. For nodes with a high-frequency collection level, data transmission instructions are sent at shorter time intervals and assigned higher priority weights, and they are scheduled to transmit key data slices first; for nodes with a low-frequency collection level, longer time intervals are used, and their transmission tasks are appropriately delayed when network resources are tight. During the transmission process, each data slice is accompanied by a corresponding metadata check code. The metadata check code is generated by the content of the data slice through a specific algorithm and is used to verify the integrity of the data slice at the receiving end.

[0076] When processing target storage data, the storage feature analysis module first analyzes its physical storage structure and logical access path. The physical storage structure includes information such as the data's sector distribution, block size, storage media type, and data redundancy within the storage device. This information is obtained by reading the storage device's hardware interface data and firmware information. The logical access path encompasses the data's file directory structure, indexing methods, permission settings, and user access records. This information is obtained by parsing the file system's metadata and access logs.

[0077] Based on the analyzed physical storage structure and logical access paths, the storage feature analysis module constructs a storage topology feature model. This model integrates physical storage distribution characteristics and logical access association characteristics, establishing a multi-dimensional mapping relationship to associate physical storage locations with logical access nodes. The model includes multiple submodules, each processing different types of feature information. For example, the physical structure submodule handles parameters such as sector distribution and block size, while the logical path submodule handles file directories and access permissions.

[0078] The stored data slices are input into the storage topology feature model for spatial mapping. During this process, the model performs a multi-dimensional analysis of the data slice's content, including data format, length, content characteristics, and relationships with other data slices. Through spatial mapping, the physical storage and logical access characteristics of the data slices are converted into standardized feature vectors. These feature vectors are arranged according to specific rules to form a set of stored data feature descriptors. The block hash fingerprint sequence in this set is calculated for each data slice using a hash algorithm. Each hash fingerprint corresponds to a unique data slice content and changes dynamically as the data slice is updated. The data integrity tag matrix records the status of each data slice during transmission and storage, including whether it has been modified, whether there are transmission errors, and whether it passes verification. Each row in the matrix represents a data slice, and each column represents an integrity status indicator, representing the specific status in binary or numerical form.

[0079] Throughout the entire process, the consensus network communication module and the storage feature analysis module maintain real-time data interaction. The consensus network communication module sends the transmitted storage data slices and metadata checksums to the storage feature analysis module in real time. The storage feature analysis module then feeds back the key parameters of the physical storage structure and logical access path obtained through analysis to the consensus network communication module for optimizing the data collection strategy. When the network status of the distributed nodes changes, the node topology distribution map is updated in real time, and the consensus network communication module adjusts the collection frequency classification table and priority weight matrix accordingly to ensure the efficiency and accuracy of data collection. At the same time, the storage topology feature model will also continuously adjust its internal parameters based on the newly received data slices to improve the accuracy of the spatial mapping operation, so that the generated storage data feature descriptor set can more accurately reflect the characteristics of the target storage data.

[0080] Example 2: See Figure 3 When building the storage feature analysis module, we first configure the cross-node feature analysis library. This library covers feature extraction rules for various heterogeneous storage devices. These rules are tailored to different types of storage media (such as solid-state drives, mechanical hard drives, and distributed storage clusters), and include information such as data block reading methods, metadata parsing formats, and feature extraction start and end locations. The rule library is organized using a categorized index, grouping storage devices by interface type, storage protocol, and data processing capabilities. This facilitates the rapid invocation of corresponding extraction rules when analyzing different devices.

[0081] Load the historical validation feature sample library, which contains a large number of verified storage data feature templates. These templates contain storage data feature information for different scenarios, such as the distribution of long-term archived data and the changing patterns of frequently accessed data. Each template is annotated with the corresponding storage environment parameters, data type, and validation results. The sample library uses a distributed storage method, spread across multiple nodes, and enables fast retrieval through an index table that records the template's feature keywords, storage location, and update time.

[0082] The topology-aware feature analysis model is trained based on a cross-node feature analysis library and a historically validated feature sample library. During training, different types of feature templates are selected from the sample library as training samples. These samples are preprocessed according to the rules in the cross-node feature analysis library to extract their key feature parameters. The preprocessed samples are then fed into the model for iterative training. The model's weight parameters are adjusted to enable the model to accurately identify data features under different storage topologies. During training, the deviation between the model's output and the sample annotations is recorded in real time, and this deviation is reduced through multiple iterations.

[0083] The topology-aware feature analysis model is incrementally optimized using real-time validation data streams to generate the final storage feature analysis module. During this incremental parameter optimization process, the system injects simulated validation data streams into the model within a pre-set test cycle. These simulated data streams contain various preset feature change patterns, such as feature data for scenarios like data block corruption, access path changes, and storage device failures. After processing the simulated data streams, the model outputs a feature detection sensitivity index, which reflects the model's ability to identify different feature changes.

[0084] When the feature detection sensitivity index falls below the preset sensitivity threshold, the adaptive learning unit is activated. It extracts incremental feature samples from the real-time validation data stream. These samples represent new feature data that the model has not encountered in previous training, such as characteristics of new storage devices or unknown data access patterns. The incremental feature samples are added to the historical validation feature sample library, the library contents are updated, and the model parameters are retrained. During retraining, only some model parameters are adjusted for the newly added sample data, avoiding large-scale modifications to existing parameters to maintain model stability.

[0085] Throughout the optimization process, the system records changes in model performance before and after each parameter adjustment, including changes in feature extraction speed and feature descriptor accuracy. Through continuous incremental optimization, the model adapts to changing storage environments and data characteristics, ensuring that the storage feature analysis module maintains effective feature extraction capabilities over the long term. Furthermore, the cross-node feature analysis library is regularly updated based on newly added storage device types and feature extraction requirements, supplemented with new extraction rules to expand the model's applicability.

[0086] Example 3: See Figure 4 When processing the set of stored data feature descriptors, the verification decision core module first performs a probabilistic correlation match with the consensus verification space. The consensus verification space contains a large number of historical verification decision cases, each of which records the corresponding stored data feature descriptors, verification strategies, and decision results. During the matching process, the system compares each feature item in the stored data feature descriptor set one by one, analyzing its similarity with the feature items of each case in the consensus verification space. By calculating the association probability between the feature items, the degree of match between the current stored data features and historical cases is determined, forming a preliminary association probability distribution.

[0087] Based on the matching results, the system generates a dynamic decision trigger interval. This interval is a range containing multiple candidate verification strategies, ranked from highest to lowest based on the degree of match. Each candidate verification strategy includes specific verification steps, the range of nodes involved, and data verification methods, designed to address different stored data characteristics and verification requirements. For example, for scenarios with a high degree of feature match, the candidate strategy may use a simplified verification process; for scenarios with a low degree of match, it may include more verification steps and node participation.

[0088] A confidence assessment is performed on each candidate verification strategy, outputting a strategy confidence score. During the assessment, the system analyzes the candidate strategy from multiple perspectives, including the strategy's successful execution history, its applicability to similar storage data characteristics, and the resource consumption required during execution. By integrating this information across these dimensions, a confidence score is calculated for each strategy. A higher score indicates a more suitable strategy for the current scenario.

[0089] Candidate verification strategies that meet dynamic confidence constraints are screened and combined to form an initial verification decision plan. Dynamic confidence constraints are dynamically adjusted based on the importance of the stored data, the real-time requirements for verification, and the current operating status of the system. For example, for the verification of core business data, the constraints will set a higher confidence threshold; for non-critical data, the threshold can be appropriately lowered to improve verification efficiency. During the screening process, the system will retain candidate strategies whose confidence scores meet or exceed the constraint threshold. These strategies are then combined based on their coverage and execution efficiency to ensure that the initial verification decision plan can fully cover the verification needs of the stored data while avoiding conflicts and redundancies between strategies.

[0090] Throughout the entire process, the Verification Decision Core Module maintains data interaction with the Storage Feature Analysis Module, acquiring real-time updates to the stored data feature descriptor set and adjusting candidate verification strategies and confidence assessment results based on the new feature data. When stored data features change significantly, the system re-executes the probabilistic association matching and confidence assessment steps to generate a new initial verification decision. Furthermore, the consensus verification space is continuously updated as new verification cases are completed, incorporating new feature descriptors, verification strategies, and decision results.

[0091] The score calculation involved in the confidence assessment operation can be achieved through the following formula:

[0092] ;

[0093] in, represents the strategy confidence score; 、 、 represent the weight coefficients of historical success rate, resource adaptability and resource consumption respectively, and ; Indicates the number of successful times of the strategy in historical verification; Indicates the total number of executions of the strategy; Indicates the number of adaptation items between the policy and the current storage environment; Indicates the total number of check items for the storage environment; Indicates the resource consumption required for policy execution; Indicates the maximum resource consumption allowed by the system.

[0094] Example 4: See Figure 5During the probabilistic association matching process, the system traverses historical decision records in the consensus verification space. These records contain various data generated during past storage information verification processes, including the characteristics of the stored data involved in each verification, the verification strategy used, and the final verification results. Sample feature descriptors and corresponding verification strategies are extracted from these records. Sample feature descriptors are structured representations of the multi-dimensional characteristics of historical storage data, while corresponding verification strategies represent the operational processes and methods used to complete verification at the time.

[0095] Calculate the topological similarity depth coefficient between the current set of stored data feature descriptors and the sample feature descriptors. This coefficient measures the degree of similarity between the two in terms of storage topology-related features. The calculation considers multiple perspectives, including storage node connectivity, data distribution patterns, and access path associations. The similarity calculation results from each perspective are integrated to form the final coefficient value.

[0096] The similarity depth coefficients of multiple topological features are weighted and fused according to preset weighting rules. These weighting rules are based on the influence of different topological features in the verification decision, with more influential features receiving greater weights. Through weighted fusion, multiple coefficients are combined into a single composite value that more comprehensively reflects the overall similarity between the currently stored data features and the sample features.

[0097] When the weighted fusion value exceeds the consensus association threshold, the corresponding verification strategy is added to the association strategy pool. The association strategy pool is used to centrally store historical verification strategies that have a high similarity with the current stored data characteristics.

[0098] The decision compensation module inputs the predicted feature descriptor set into the consensus verification space for secondary correlation matching. The predicted feature descriptor set is the stored data features within the future time window predicted by the time series prediction module based on historical data. The secondary correlation matching process is similar to probabilistic correlation matching, and similarity coefficients are also calculated and integrated to find a historical verification strategy that matches the predicted features.

[0099] A compensation decision trigger interval is generated based on the secondary matching results. This interval encompasses a range of possible compensation strategies. These strategies are designed to address potential deviations between predicted and current features, aiming to compensate for potential deficiencies in the initial verification decision solution in responding to future feature changes.

[0100] Multiple rounds of iterative optimization are performed within the compensation decision trigger interval to generate a set of compensation verification strategies. Each round of iteration adjusts and optimizes the strategy based on the results of the previous round. Through continuous iteration, the compensation strategy can better adapt to the evolution path of the prediction characteristics, ensuring its effectiveness and pertinence.

[0101] The compensation verification strategy set is combined with the initial verification decision plan through a strategy fusion calculation to output an optimized verification decision plan. During the fusion calculation process, the strategies in the two strategy sets are compared and integrated, retaining the effective ones and removing any duplicate or conflicting ones to form a more complete and adaptable verification decision plan.

[0102] The following is an example table of weight distribution involved in the weighted fusion process of topological feature similarity depth coefficients:

[0103] Table 1: Topological feature type weight distribution table

[0104] ;

[0105] Through this process, the decision compensation module optimizes the initial verification decision plan based on future predictions. This optimized plan not only adapts to the current stored data characteristics but also effectively addresses potential future changes in these characteristics, thereby improving the reliability and adaptability of the entire storage information verification system. Throughout the implementation process, all modules maintain real-time data interaction and information sharing, ensuring that each step is based on the latest and most comprehensive data, thereby guaranteeing the rationality and effectiveness of the resulting optimized verification decision plan.

[0106] Example 5: During the execution process, the verification execution engine module synchronously monitors the verification response delay and result consistency index of the distributed nodes. The verification response delay refers to the time interval between the verification execution engine module sending the execution instruction and the distributed node returning the verification result. By embedding timestamps in the instruction and result data packets to achieve record keeping, the system will continuously track the response time changes of each node. The result consistency index is used to measure the degree of consistency of the verification results of the same stored data by different distributed nodes. It is calculated by comparing the key information such as the hash value of the verification data slice returned by each node, the metadata check code, etc. When there are differences in the verification results of multiple nodes, the indicator will change accordingly.

[0107] Generate an execution efficiency evaluation report based on the verification response latency. This report includes the average response time, maximum response time, minimum response time, and response time distribution for each distributed node. Nodes are categorized by their storage partition and data type. The report also identifies nodes with response times outside the normal range and records their corresponding verification task type and data characteristics.

[0108] The verification stability coefficient is calculated based on the result consistency index. This coefficient comprehensively considers the changing trend of the result consistency index within a certain time window, as well as the frequency of deviations in verification results between different nodes. During the calculation process, consistency indicators at multiple consecutive time points are weighted, with the weight increasing over time. More recent indicators have a greater impact on the coefficient, reflecting the current stability of the system.

[0109] When the verification stability coefficient falls below the dynamic threshold, the timing prediction module is triggered to regenerate the feature evolution path. This dynamic threshold is adjusted dynamically based on the system's operating load, the importance of the stored data, and historical stability data. It is relaxed when the system load is high and tightened when processing critical data. When regenerating the feature evolution path, the timing prediction module incorporates the latest verification response latency, result consistency indicators, and node status data, adjusting the prediction model parameters to make the generated path more consistent with the current system's actual operation.

[0110] The system includes a stability fusing module, which continuously receives the verification stability coefficient output by the verification execution engine module. During operation, the stability fusing module sets a counter for each verification cycle. When the verification stability coefficient falls below the dynamic threshold, the counter records an abnormal stability signal. The preset monitoring period is a continuous time interval. The length of the interval is determined by the system's verification frequency and data update rate, and may be several hours or a day.

[0111] When the cumulative number of abnormal stability signals received within a preset monitoring period exceeds the fuse threshold, the stability fuse module generates a system self-test instruction. The fuse threshold is determined based on the system's fault tolerance and business continuity requirements, and different types of storage data verification tasks correspond to different thresholds. The system self-test instruction triggers the consensus network communication module to suspend new verification requests. At this time, the consensus network communication module will return a suspended response to the node that initiated the new request and record relevant information about the request for subsequent processing. At the same time, the system initiates the node status diagnosis process, which includes checking the node's hardware operating status (such as disk read and write speed, memory usage, network interface status, etc.), software service logs (such as verification program error records, process operation status, etc.), and data storage integrity (such as whether there is data block corruption, index errors, etc.). The diagnostic results will be recorded and used for subsequent troubleshooting and system recovery.

[0112] It should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus.

[0113] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A storage information verification system based on a consensus mechanism, characterized in that: include: Consensus network communication module, used to connect to distributed storage nodes and receive storage verification requests; a storage feature analysis module, configured to extract a multi-dimensional feature vector of target storage data based on the storage verification request and generate a storage data feature descriptor set; The verification decision core module is used to perform consensus decision operations based on the embedded consensus verification space and dynamic confidence constraints, combined with the set of stored data feature descriptors, and output an initial verification decision plan; the consensus verification space contains a large number of historical verification decision cases, each of which records the corresponding stored data feature descriptors, verification strategies, and decision results; A time series prediction module is used to segment the stored data feature descriptor set into time windows, predict the feature evolution path in the future time window through historical verification records, and generate a predicted feature descriptor set; A decision compensation module, configured to perform deviation compensation calculation on the initial verification decision plan based on the prediction feature descriptor set to generate an optimized verification decision plan; A verification execution engine module is used to send an execution instruction to the consensus network communication module according to the optimized verification decision scheme, triggering the distributed nodes to perform storage verification operations; The verification decision core module performs: Performing a probabilistic association matching process on the stored data feature descriptor set and the consensus verification space; generating a dynamic decision triggering interval according to the matching result, wherein the dynamic decision triggering interval includes a plurality of candidate verification strategies; Perform confidence evaluation operations on each candidate verification strategy and output a strategy confidence score; Screen candidate verification strategies that meet dynamic confidence constraints and combine them to form an initial verification decision plan; The probabilistic association matching process includes: Traverse the historical decision records in the consensus verification space and extract sample feature descriptors and corresponding verification strategies; Calculate the topological feature similarity depth coefficient between the current stored data feature descriptor set and the sample feature descriptor; The similar depth coefficients of multiple topological features are weightedly fused according to the preset weight distribution rules; When the weighted fusion value exceeds the consensus association threshold, the corresponding verification strategy is added to the association strategy pool.

2. The storage information verification system based on the consensus mechanism according to claim 1, characterized in that: The consensus network communication module executes, upon receiving a storage verification request: Load the node topology distribution diagram of the target storage data; Dynamically generate a data collection strategy based on the node topology distribution map, wherein the data collection strategy includes a collection frequency classification table and a priority weight matrix; According to the acquisition frequency classification table and the priority weight matrix, the distributed nodes are scheduled to synchronously transmit and store data slices and metadata check codes.

3. The storage information verification system based on the consensus mechanism according to claim 1, characterized in that: The storage feature analysis module performs: Analyze the physical storage structure and logical access path of the target storage data; Constructing a storage topology feature model according to the physical storage structure and the logical access path; Inputting the storage data slice into the storage topology feature model to perform space mapping operation and output a storage data feature descriptor set; The storage data feature descriptor set includes a block hash fingerprint sequence and a data integrity mark matrix.

4. The storage information verification system based on the consensus mechanism according to claim 3 is characterized in that: The construction process of the storage feature analysis module includes: Configuring a cross-node feature analysis library, wherein the cross-node feature analysis library includes a feature extraction rule library for heterogeneous storage devices; Loading a historical verification feature sample library, wherein the historical verification feature sample library contains a feature template set of verified stored data; Training a topology-aware feature analysis model based on the cross-node feature analysis library and the historical verification feature sample library; An incremental parameter optimization process is performed on the topology-aware feature analysis model through real-time verification data flow to generate a final storage feature analysis module.

5. The storage information verification system based on the consensus mechanism according to claim 4 is characterized in that: The incremental parameter optimization process includes: Inject simulated verification data streams within the preset test cycle; Calculating a feature detection sensitivity index output by the topology-aware feature analysis model; When the feature detection sensitivity index is lower than a preset sensitivity threshold, activating the adaptive learning unit; The adaptive learning unit extracts incremental feature samples from the real-time verification data stream, updates the historical verification feature sample library and retrains model parameters.

6. The storage information verification system based on the consensus mechanism according to claim 1, characterized in that: The decision compensation module performs: Inputting the predicted feature descriptor set into the consensus verification space for secondary association matching; Generate a compensation decision trigger interval based on the secondary matching result; Perform multiple rounds of iterative optimization operations within the compensation decision triggering interval to generate a compensation verification strategy set; The compensation verification strategy set is combined with the initial verification decision plan to perform strategy fusion calculation and output an optimized verification decision plan.

7. The storage information verification system based on the consensus mechanism according to claim 1, characterized in that: The verification execution engine module executes synchronously during the execution process: Monitor the verification response latency and result consistency indicators of distributed nodes; generating an execution efficiency evaluation report based on the verification response delay; Calculate the verification stability coefficient based on the result consistency index; When the verification stability coefficient is lower than the dynamic threshold, the time series prediction module is triggered to regenerate the feature evolution path.

8. The storage information verification system based on the consensus mechanism according to claim 7, characterized in that: The system further comprises: a stability fuse module, configured to continuously receive the verification stability coefficient output by the verification execution engine module; When the accumulated abnormal stability signals received within a preset monitoring period exceed the fuse threshold, a system self-check instruction is generated; The system self-check instruction triggers the consensus network communication module to suspend new verification requests and start the node status diagnosis process.

Images