Digital right near-field real-time cancel-after-verification system and method based on edge calculation

Abstract

The invention relates to the technical field of information, and discloses a digital right near-field real-time cancel-after-verification system based on edge computing, which comprises a space-time coding module used for generating a unique cancel-after-verification identifier according to the geographic position and time of a user, and transmitting the identifier to a dynamic weight module; a dynamic weight module which is connected with the space-time coding module, generates a right and interest distribution strategy based on a federated learning model and an optimization algorithm, and outputs the optimized strategy to a block chain fragmentation module; the block chain fragment module is connected with the dynamic weight module, dynamically divides a consensus network according to the real-time cancel-after-verification density, executes distributed verification in fragments and generates a cancel-after-verification authorization instruction; and the cancel-after-verification execution module is used for distributing cancel-after-verification tasks according to the authorization instruction and synchronously updating the multi-node inventory state. By adopting the technical scheme of gridding coordinate mapping, time window division and nonlinear hash of the space-time coding module, the technical effect that the verification identifier is unpredictable and uniquely bound is achieved.

Description

Technical Field

[0001] The present invention relates to the field of information technology, and particularly to a digital rights near-field real-time verification system and method based on edge computing. Background Art

[0002] The digital rights near-field real-time verification system and method based on edge computing is a technical system for offline high-frequency transaction scenarios. By integrating the distributed computing power of edge computing nodes, the immutable characteristics of blockchain, and dynamic optimization algorithms, it realizes the rapid verification and secure authentication of digital rights such as electronic coupons and points. Its core goal is to solve the response delay and reliability bottlenecks of traditional centralized systems in high-concurrency and dynamic scenarios through geographical location awareness, real-time resource scheduling, and multi-node collaboration while ensuring privacy and anti-attack capabilities.

[0003] Existing digital rights verification systems based on edge computing usually adopt predefined area sharding and static task allocation strategies. Specifically, the physical space is divided into grid areas of fixed size, and each edge node is responsible for processing verification requests within a fixed grid, and a centralized server uniformly generates verification identifiers. At the resource allocation level, a global weight model is trained based on historical data, and the node task load is adjusted in a periodic batch update manner. The blockchain part adopts a fixed sharding strategy, and the main node is preselected to complete transaction verification and block generation.

[0004] However, the existing solutions have significant defects in a dynamic business environment: the static sharding and fixed weight allocation mechanisms are difficult to adapt to the real-time fluctuating verification request density, resulting in serious uneven load among edge nodes. Task backlogs and response timeouts frequently occur in high-density areas, while low-density areas are in a state of long-term resource idleness. This resource mismatch problem is particularly prominent in sudden traffic scenarios such as commercial promotions and holidays, and has become a key obstacle restricting the practicality and scalability of the system. Summary of the Invention

[0005] Aiming at the deficiencies of the prior art, the present invention provides a digital rights near-field real-time verification system and method based on edge computing, which solves the problems of high response delay, serious resource mismatch, and insufficient anti-forgery ability in dynamic business scenarios caused by static sharding, fixed weight allocation, and centralized scheduling in existing digital rights verification systems based on edge computing.

[0006] To achieve the above objectives, the present invention is realized through the following technical solutions: A digital rights near-field real-time verification system based on edge computing, including: A spatio-temporal encoding module, configured to generate a unique verification identifier according to the user's geographical location and time, and transmit the identifier to the dynamic weight module; The dynamic weight module, connected to the spatio-temporal coding module, generates an interest distribution strategy based on the federated learning model and the optimization algorithm, and outputs the optimized strategy to the blockchain sharding module; The blockchain sharding module, connected to the dynamic weight module, dynamically divides the consensus network according to the real-time cancellation density, and performs distributed verification within the shard and generates a cancellation authorization instruction; The cancellation execution module, connected to the blockchain sharding module, allocates cancellation tasks according to the authorization instruction and synchronously updates the inventory status of multiple nodes.

[0007] Preferably, the spatio-temporal coding module performs the following steps: Map the user coordinates to the standard grid to generate a spatio-temporal vector , where: ; Among them, is the grid resolution, is the time window length; Perform non-linear coding and hash operation on the spatio-temporal vector to generate a cancellation identifier CID.

[0008] Preferably, the dynamic weight module includes: The federated learning unit aggregates the weight increments of the edge nodes and injects differential privacy noise. The update formula is: ; Among them: ; The quantum annealing unit constructs and solves an optimization model containing interest distribution constraints.

[0009] Preferably, the optimization model is a quadratic unconstrained binary optimization model, and its Hamiltonian is: ; Among them: is the penalty coefficient, .

[0010] Preferably, the sharding radius of the blockchain sharding module satisfies: ; Among them, is the minimum sharding radius, is the adjustment constant; is the cancellation density per unit area.

[0011] Preferably, the consensus process of the blockchain sharding module includes: Elect the master node based on the spatio-temporal identification distance; Use the SM9 algorithm to sign the cancellation request, and the signature data includes the hash digest of the weight matrix.

[0012] Preferably, the write-off execution module allocates the write-off quantity through the following formula: ; where: is the total quantity of rights requests, is the node allocation quantity.

[0013] Preferably, the generation formula of the SM9 signature is: ; where: is the private key conforming to the GM / T 0044 standard.

[0014] The present invention also provides a digital rights near-field real-time write-off method based on edge computing, including the following steps: Step 1, generate a write-off identifier bound by time and space; Step 2, dynamically optimize the rights allocation strategy; Step 3, dynamically slice blockchain consensus; Step 4, multi-node collaborative write-off and inventory synchronization.

[0015] The present invention also provides a computer-readable storage medium storing a computer program, and when the program is executed by a processor, the steps of the method are implemented.

[0016] The present invention provides a digital rights near-field real-time write-off system and method based on edge computing. It has the following beneficial effects: 1. By adopting the grid coordinate mapping, time window division and non-linear hashing technical solutions of the space-time coding module, the present invention achieves the technical effect that the write-off identifier is unpredictable and uniquely bound. Compared with the identification schemes using static or simple encryption in the prior art, it solves the deficiencies of being vulnerable to replay attacks and brute-force cracking, and effectively resists scalping and forged write-offs through random number perturbation and dynamic time windows.

[0017] 2. Through the federated learning aggregation and quantum annealing path solving of the dynamic weight module, the present invention realizes the global optimization and dynamic adaptation of the rights allocation strategy. In the prior art, fixed weights or centralized optimization easily lead to resource rigidity and privacy leakage. The present invention injects differential noise into the local training of edge nodes and combines quantum annealing for fast solution, taking into account both privacy protection and computational efficiency.

[0018] 3. By dynamically adjusting the sharding radius based on the real-time write-off density and electing the master node through space-time proximity, the present invention significantly reduces the network latency and communication overhead. Compared with the traditional fixed sharding or random sharding mechanisms, it solves the problems of uneven load and high cross-region communication cost, and enables the throughput in high-concurrency scenarios to adapt to the change of service density.

[0019] 4. The present invention synchronizes through the write-off execution module with the reverse entropy increment through the two-phase commit protocol to ensure the atomicity and eventual consistency of the inventory status of multiple nodes. In the prior art, asynchronous submission or full synchronization is likely to cause overselling and data conflicts. The present invention locks resources in the pre-submission stage and quickly repairs differences based on the version vector to achieve highly reliable inventory management. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 is the main framework diagram of the present invention; Figure 2 is the method flow chart of the present invention; Figure 3 is the schematic structural diagram of the computer device of the present invention.

[0021] Among them, 1. Computer device; 11. Processor; 12. Memory; 13. Storage medium. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0022] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0023] Embodiment: Please refer to the attached Figure 1 , the embodiment of the present invention provides a digital rights near-field real-time write-off system based on edge computing, including: A spatio-temporal encoding module for generating a unique write-off identifier according to the user's geographical location and time and transmitting the identifier to the dynamic weight module; A dynamic weight module, connected to the spatio-temporal encoding module, generating an interest distribution strategy based on the federated learning model and the optimization algorithm, and outputting the optimized strategy to the blockchain sharding module; A blockchain sharding module, connected to the dynamic weight module, dynamically dividing the consensus network according to the real-time write-off density, performing distributed verification within the shard and generating a write-off authorization instruction; A write-off execution module, connected to the blockchain sharding module, allocating write-off tasks according to the authorization instruction and synchronously updating the inventory status of multiple nodes.

[0024] The spatio-temporal encoding module performs the following steps: Mapping the user coordinates to the standard grid to generate a spatio-temporal vector , where: ; Among them, is the grid resolution, and is the time window length; Perform non - linear encoding and hashing operations on the spatio - temporal vector to generate the cancellation identifier CID.

[0025] The dynamic weight module includes: The federated learning unit aggregates the weight increments of the edge nodes and injects differential privacy noise. The update formula is: ; Where: ; The quantum annealing unit constructs and solves an optimization model that includes the equity distribution constraint.

[0026] The optimization model is a quadratic unconstrained binary optimization model, and its Hamiltonian is: ; Where: is the penalty coefficient, .

[0027] The sharding radius of the blockchain sharding module satisfies: ; Among them, is the minimum sharding radius, is the adjustment constant; is the cancellation density per unit area.

[0028] The consensus process of the blockchain sharding module includes: Elect the master node based on the spatio - temporal identification distance; Use the SM9 algorithm to sign the cancellation request, and the signature data includes the hash digest of the weight matrix.

[0029] The cancellation execution module allocates the cancellation quantity through the following formula: ; Where: is the total quantity of equity requests, is the node allocation quantity.

[0030] The generation formula of the SM9 signature is: ; Where: is the private key that conforms to the GM / T 0044 standard.

[0031] As an option, the spatio - temporal encoding module first normalizes the original geographical coordinates obtained by the user terminal. Specifically, the original coordinates The positioning sensor (e.g., GPS or WiFi fingerprint positioning module) from the user device is discretely mapped through preset reference points and grid resolution. Exemplarily, let the regional reference point coordinates be , and the grid side length be and , then the calculation method of the standardized grid coordinates is: ; It should be noted that the grid resolution can be dynamically configured according to the actual application scenario. For example, it can be set to 50 meters in urban dense areas and 200 meters in suburban areas to balance positioning accuracy and computational complexity. Preferably, the reference point usually selects the coordinates of the regional center point or landmark building to ensure the uniformity of grid division.

[0032] In a possible implementation, the module further discretizes the time dimension. Specifically, the current timestamp (accurate to the millisecond level) is obtained, and it is divided according to the preset time window length to generate the time window number , and its calculation method is: ; It can be understood that the setting of the time window length needs to balance the dynamic update frequency and system load. For example, it can be shortened to 3 minutes during peak hours to improve anti-counterfeiting ability, and extended to 10 minutes during off-peak hours to reduce computational overhead.

[0033] Exemplarily, the module combines the standardized spatial coordinates with the time window number into a three-dimensional spatio-temporal vector , as the basic input for subsequent encoding. It should be noted that the construction of the three-dimensional vector realizes the strong binding of the spatial and time dimensions, making the identifiers generated at the same physical location in different time windows have significant differences.

[0034] As an option, the module enhances the features of the spatio-temporal vector through non-linear transformation. Specifically, a single-layer neural network model is used to map the three-dimensional vector to a high-dimensional hidden space, and its calculation method is: ; Among them, is the weight matrix, is the bias vector, and the hyperbolic tangent function (tanh) is selected as the activation function to limit the output range between [-1, 1]. Preferably, the weight matrix is initialized using the orthogonal initialization method to avoid the problem of gradient disappearance, while the bias term Initialize to a zero vector. It should be noted that the purpose of this non-linear transformation is to disrupt the linear structure of the original spatio-temporal vector and enhance the irreversibility of the encoding result.

[0035] In one possible implementation, the module randomizes the hidden vector to further enhance security. Specifically, a 64-bit random number is generated , and it is bitwise XOR (XOR) with the hidden vector and then input into the hash function to generate the final verification identifier , and its calculation method is: ; It can be understood that the random number is dynamically generated by a cryptographically secure pseudo-random number generator (CSPRNG) and is unique within each time window. By introducing random interference, even in adjacent grids within the same time window, the generated CID will have high randomness due to being different, thus effectively resisting brute-force cracking attacks.

[0036] It should be noted that the SHA3-256 algorithm is selected for the hash function, and its output length is 256 bits, meeting the requirements of cryptographic strength. Preferably, the input data is padded before the hash operation to make its length meet the algorithm input requirements. For example, when the hidden vector has a length of 128 bits, it is concatenated with a 64-bit random number and padded to 256 bits, and then the hash operation is performed.

[0037] In one possible implementation, the module also includes an exception handling mechanism. Specifically, when a positioning signal loss or time synchronization anomaly is detected, a backup encoding strategy is enabled: generating a spatio-temporal vector based on the last valid coordinate and superimposing a timestamp offset to maintain dynamics. Exemplarily, if the current timestamp synchronization fails, then calculate: ; where is the last valid timestamp, is a preset fixed offset (e.g., 30 seconds). It should be noted that this mechanism ensures that a legal CID can still be generated in a weak network environment, avoiding interruption of the verification service.

[0038] It can be understood that the output result CID of the spatio-temporal encoding module will be used as a globally unique identifier throughout subsequent processes such as dynamic weight calculation, blockchain sharding consensus, and verification execution. Through the multiple technology superposition of space discretization, time windowing, non-linear transformation, and randomization, this module realizes strong identity binding and anti-tampering capabilities for verification requests.

[0039] As an option, the dynamic weight module first aggregates the local weight updates of each edge node through the federated learning framework. Specifically, each edge node maintains a local LSTM prediction model, with the input being the write-off records and corresponding spatio-temporal vectors of the previous time window , and the output being the weight increment . It should be noted that the training data of the local model includes features such as the success rate and response delay of the node in processing specific rights and interests types to reflect the real-time processing ability of the node. Exemplarily, the global aggregation process of federated learning adopts a differential privacy protection mechanism. Specifically, noise injection is performed on the local model updates during the gradient aggregation stage, and its update formula is: ; where is the learning rate, is the gradient clipping threshold, is the noise standard deviation. It can be understood that the gradient clipping operation (clip) is used to limit the update amplitude of a single node to prevent outliers from interfering with the global model, while the addition of Gaussian noise ensures individual data privacy. Preferably, the noise standard deviation is calculated according to the preset privacy budget to meet the strict mathematical definition of differential privacy.

[0040] In a possible implementation, the module further optimizes the rights and interests allocation path through the quantum annealing algorithm. Specifically, the write-off node selection problem is modeled as a quadratic unconstrained binary optimization (QUBO) model, and its Hamiltonian is defined as: ; where indicates whether to select node to process the rights and interests , is the constraint penalty coefficient. It should be noted that the first term of the objective function maximizes the overall utility of the weight matrix , and the second term enforces that each rights and interests request is mainly processed by a single node to avoid resource conflicts. Preferably, the value of the constraint coefficient λ needs to balance the optimization objective and the strength of constraint satisfaction. For example, when the node resources are tight, λ can be increased to strengthen the constraint.

[0041] Exemplarily, the quantum annealing solution process is executed on a dedicated quantum computing device. Specifically, the Hamiltonian is mapped to the coupling field and bias field of qubits, and the ground state solution is found through the annealing algorithm It is understandable that the ground state solution corresponds to the configuration with the lowest energy, that is, the globally optimal node allocation scheme. It should be noted that parameters such as the annealing time and the temperature scheduling curve affect the solution accuracy. Preferably, the annealing time is set to the minimum value allowed by the device to improve the calculation efficiency.

[0042] As an option, the module probabilistically renormalizes the quantum annealing result. Specifically, the binary solution is converted into a weight matrix in the form of a continuous probability distribution, and its conversion formula is: ; where represents the Hamming distance between the solution and the center of the ideal distribution , and is the temperature parameter. It should be noted that the purpose of the probabilistic processing is to introduce flexibility in the hard decision (0 / 1 selection), allowing edge nodes to dynamically adjust the nuclear sales volume under sudden loads. Preferably, the value of the temperature parameter is negatively correlated with the real-time load of the system. For example, when the node load is high, is reduced to expand the difference in the allocation probability.

[0043] In a possible implementation, the module includes a version control mechanism for the weight matrix. Specifically, each updated weight matrix is attached with a version number , and is bound to the CID generated by the spatio-temporal encoding module and stored in the blockchain shard. It is understandable that the version number is used to track the evolution history of the weight matrix and match the corresponding allocation strategy during the verification of write-off. Exemplarily, when it is detected that the version of the weight matrix does not match the CID time window, an abnormal rollback process is triggered, and the federated learning aggregation and quantum annealing solution are restarted.

[0044] As an option, the blockchain shard module first dynamically adjusts the shard radius according to the spatio-temporal write-off density. Specifically, the shard radius is calculated by the formula: ; where represents the number of real-time write-off requests per unit area (unit: times per square meter per second), is the preset minimum shard radius, and is the adjustment constant. It should be noted that the square root term in the formula is used to balance the non-linear relationship between the shard scale and the request density, ensuring that the shards in the high-density area are refined to reduce the load of a single shard, while the shards in the low-density area are expanded to reduce the consensus overhead. Preferably, the minimum shard radius The setting needs to consider the coverage requirements of the physical scenario. For example, in an indoor scenario, it can be set to 30 meters to match the building structure.

[0045] In a possible implementation, the module elects a consensus master node within a shard based on the spatio-temporal identity distance. Specifically, each candidate node generates a spatio-temporal identity according to its geographical location SHA3-256 , and calculates the XOR distance with the of the user cancellation request: ; Among them, represents the Euclidean norm of the vector. It should be noted that the distance calculation ensures that the master node is highly adjacent to the user request in the spatio-temporal dimension, thereby reducing network transmission latency. Preferably, the master node election period is synchronized with the time window of the spatio-temporal coding module, for example, re-elected every 5 minutes to adapt to the dynamic environment.

[0046] As an option, the module uses the SM9 identity-based cryptographic algorithm to verify the signature of the cancellation request. Specifically, the signature data includes the spatio-temporal vector , the weight matrix version number , and the hash digest of the cancellation identifier , and its generation formula is: ; Among them, is the private key of the edge node, which conforms to the GM / T 0044 standard, is the SHA3-256 hash function. It can be understood that the signature mechanism strongly binds the weight matrix version number to the cancellation request, preventing malicious nodes from launching replay attacks using expired policies. Exemplarily, when the weight matrix is updated to version , the old version signature automatically becomes invalid, and the cancellation authorization process needs to be initiated again.

[0047] In a possible implementation, the module reaches consensus within a shard through a lightweight Byzantine Fault Tolerance (BFT) protocol. Specifically, after the master node collects the cancellation requests, it broadcasts a pre-prepare message to other nodes within the shard, and the message format includes: Block height, request list, ; After the slave nodes verify the validity of the master node signature, they enter the prepare and commit phases, and generate respectively: Block height, digest ; Block height, final state ; It should be noted that the consensus protocol only needs a majority of nodes within a shard (e.g., more than 2 / 3) to reach an agreement to complete block confirmation, significantly reducing the communication complexity of the traditional PBFT algorithm. Preferably, the spatio-temporal shard identifier is embedded in the pre-prepared message to ensure that the message is only propagated within the current shard and avoid cross-shard interference.

[0048] As an option, the write-off execution module first allocates the write-off volume of each node according to the dynamic weight matrix. Specifically, for the total request volume of the equity type , the allocation volume of node is calculated as follows: ; It should be noted that the weight matrix comes from the optimized output of the dynamic weight module, and its element represents the allocation weight of node processing the equity . It can be understood that by performing a floor operation, the allocation volume is ensured to be an integer, avoiding inventory fragmentation caused by fractional write-off volumes. Preferably, the remaining unallocated volume is processed by the cloud backup node to ensure complete response to requests.

[0049] In a possible implementation, the module realizes atomic inventory update through the two-phase commit protocol. Specifically, it includes the following steps: Pre-commit phase: After node receives the allocation volume , it locks the corresponding quantity in the local inventory, generates a pre-commit log and returns a confirmation signal to the master node.

[0050] Commit phase: When the master node receives confirmations from more than half of the nodes, it broadcasts a commit instruction, and each node actually deducts the inventory and generates a transaction completion record.

[0051] Exemplarily, if a node fails after pre-commit, the master node will trigger a timeout rollback mechanism to release the locked inventory and restart the allocation process. It should be noted that the two-phase protocol ensures the atomicity of inventory operations in a distributed environment and prevents overselling or data inconsistency As an option, the module uses the reverse entropy algorithm to synchronize the inventory status across shards. Specifically, each node maintains a version vector , recording the timestamp and content of its latest inventory change. Periodically exchange incremental data packets with adjacent nodes, and its format is defined as: , , ; Among them, is a timestamp, is the inventory change amount, which is jointly signed by at least 3 nodes within the shard. It can be understood that the signature mechanism ensures the authenticity and integrity of the incremental data and prevents malicious nodes from forging inventory changes. Preferably, the version vector adopts the Vector Clock algorithm to track the state dependency relationship among multiple nodes through logical timestamps.

[0052] In a possible implementation, the module includes an anomaly detection and adaptive recovery mechanism. Specifically, when it is detected that the node inventory is inconsistent with the global state, the following operations are performed: Difference localization: By comparing the local version vector with the version vectors of neighbor nodes, locate the inconsistent right types and time windows; Data repair: Pull the correct inventory data from the majority of node replicas to overwrite the local abnormal state; Log traceability: Roll back the unconfirmed operations based on the pre-commit log and re-execute the allocation process.

[0053] It should be noted that the mechanism ensures that the repaired state is consistent with the consensus result within the shard through the majority principle, avoiding data loss caused by a single point of failure.

[0054] It can be understood that the output of the write-off execution module (i.e., the inventory change record) will be used as the input for subsequent transaction verification. Through the coordination of weight allocation, atomic operations, and incremental synchronization, it realizes the inventory consistency management under high throughput. Preferably, the inventory change records are batch written to the blockchain by time window to reduce the write load of the blockchain.

[0055] The digital rights near-field real-time write-off method based on edge computing described below can be correspondingly referred to the digital rights near-field real-time write-off system based on edge computing described above.

[0056] Please refer to the attached Figure 2 , the digital rights near-field real-time write-off method based on edge computing, includes the following steps: Step 1, generate a write-off identifier bound by time and space; Step 2, dynamically optimize the right allocation strategy; Step 3, dynamically shard the blockchain consensus; Step 4, multi-node collaborative write-off and inventory synchronization.

[0057] The device in this embodiment can be used to execute the above method embodiment, and its principle and technical effects are similar, so they will not be elaborated here.

[0058] Please refer to the attachedFigure 3 , the present invention also provides a computer device 1, including: a processor 11 and a memory 12. The memory 12 stores a computer program executable by the processor. When the computer program is executed by the processor, the above method is executed.

[0059] The present invention also provides a storage medium 13. A computer program is stored on the storage medium 13. When the computer program is run by the processor 11, the above method is executed.

[0060] Among them, the storage medium 13 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (abbreviated as SRAM), electrically erasable programmable read-only memory (abbreviated as EEPROM), erasable programmable read-only memory (abbreviated as EPROM), programmable read-only memory (abbreviated as PROM), read-only memory (abbreviated as ROM), magnetic memory, flash memory, magnetic disk or optical disk.

[0061] The program code in the storage medium is divided into the following modules according to functions. The storage addresses and execution logics of each module are as follows: Space-time coding generation module: Stores code segments for calculating the cancellation identifier (CID), including grid mapping, time window division, and hash algorithm implementation.

[0062] Dynamic weight optimization module: Stores the code logic of federated learning aggregation and quantum annealing solution, including LSTM local training, differential noise injection, and QUBO model construction.

[0063] Blockchain sharding control module: Stores code for calculating the dynamic sharding radius, electing the primary node, and lightweight BFT consensus protocol.

[0064] Cancellation execution and synchronization module: Stores weight distribution calculation, two-phase commit protocol, and reverse entropy synchronization algorithm.

[0065] Exception handling module: Stores exception handling logics such as timeout rollback, version vector repair, and majority data overwrite.

[0066] Interaction logic between program and hardware Processor instruction set adaptation: The quantum annealing optimization code segment is optimized for processors supporting quantum instruction extensions (such as QCE) to accelerate the solution of the Hamiltonian.

[0067] The SM3 / SM9 algorithm is implemented by calling a hardware cryptographic engine (such as the ARM TrustZone CryptoCell).

[0068] Edge device collaboration: The program obtains the local LSTM model parameters from the edge node through Remote Direct Memory Access (RDMA).

[0069] When synchronizing the inventory status, it is directly written to the FPGA-accelerated NVMe storage controller to reduce the CPU load.

[0070] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A digital rights near-field real-time verification system based on edge computing, characterized in that, Including: A spatio-temporal coding module, configured to generate a unique verification identifier according to the user's geographical location and time, and transmit the identifier to a dynamic weight module; A dynamic weight module, connected to the spatio-temporal coding module, generating an interest distribution strategy based on a federated learning model and an optimization algorithm, and outputting the optimized strategy to a blockchain sharding module; A blockchain sharding module, connected to the dynamic weight module, dynamically dividing a consensus network according to real-time verification density, performing distributed verification within the shard and generating a verification authorization instruction; A verification execution module, connected to the blockchain sharding module, allocating verification tasks according to the authorization instruction and synchronously updating the inventory status of multiple nodes.

2. The digital rights near-field real-time verification system based on edge computing according to claim 1, wherein The spatio-temporal coding module performs the following steps: Map the user coordinates to a standard grid to generate a spatio-temporal vector , where: ； Among them, is the grid resolution, is the time window length; Performing non-linear coding and hash operation on the spatio-temporal vector to generate a verification identifier CID.

3. The near-field real-time verification system for digital rights based on edge computing according to claim 1, wherein The dynamic weight module includes: A federated learning unit, aggregating the weight increments of edge nodes and injecting differential privacy noise, with the update formula: ; Wherein: ; A quantum annealing unit, constructing an optimization model including interest distribution constraints and solving it.

4. The digital rights near-field real-time verification system based on edge computing according to claim 1, wherein The optimization model is a quadratic unconstrained binary optimization model, and its Hamiltonian is: ； Wherein: is the penalty coefficient, .

5. The digital rights near-field real-time verification and cancellation system based on edge computing according to claim 1, characterized in that, The sharding radius of the blockchain sharding module satisfies: ； Among them, is the minimum slice radius, is the adjustment constant; is the cancellation density per unit area.

6. The digital rights near-field real-time verification system based on edge computing according to claim 1, wherein The consensus process of the blockchain sharding module includes: Electing a primary node based on the spatio-temporal identification distance; Signing the verification request using the SM9 algorithm, and the signature data includes the hash digest of the weight matrix.

7. The near-field real-time verification system for digital rights based on edge computing according to claim 1, characterized in that, The verification execution module allocates the verification quantity through the following formula: ； Wherein: is the total amount of rights requests, is the amount allocated to nodes.

8. The near-field real-time verification system for digital rights based on edge computing according to claim 6, characterized in that The generation formula of the SM9 signature is: ； Wherein: is a private key compliant with the GM / T 0044 standard.

9. A method for near-field real-time verification of digital rights based on edge computing, according to the system for near-field real-time verification of digital rights based on edge computing described in any one of claims 1-8, characterized in that, Including the following steps: Step 1, generating a verification identifier bound to space and time; Step 2, dynamically optimizing the interest distribution strategy; Step 3, dynamically sharding blockchain consensus; Step 4, multi-node collaborative verification and inventory synchronization.

10. A computer-readable storage medium storing a computer program, characterized in that, When the program is executed by a processor, it implements the steps of the method according to claim 9.

Images