A cloud archive destruction and data provenance method and system

Abstract

The application relates to the technical field of data security, in particular to a cloud file destruction and data tracing method and system, which comprises the following steps: a distributed hash table algorithm is used to analyze a physical storage unit set of a cloud file to be destroyed; a chaotic sequence multiple overwriting operation is used to forcibly clean the physical layer; a storage unit state flip matrix is collected in real time, and a Shannon entropy value is calculated; a cryptography destruction certificate is generated after the Shannon entropy value breaks through a safe destruction baseline; and finally, the certificate is encapsulated into a terminal block of a full life cycle tracing chain. In the application, the chaotic overwriting based on the entropy value detection is executed by analyzing the physical storage unit set, and the residual magnetic characteristics of the data in the physical layer are forcibly destroyed, so that the hidden danger of false deletion in the cloud environment is solved, the unalterable destruction evidence is established in combination with a chain locking mechanism, and the safe closed loop of the data full life cycle is realized.

Description

Technical Field

[0001] This invention relates to the field of data security technology, and in particular to a method and system for cloud-based file destruction and data traceability. Background Technology

[0002] Data security technology involves the full lifecycle management of cloud data, aiming to prevent unauthorized access and tampering, and ensure the compliant removal of data. Traditional cloud file destruction and data traceability methods involve the system removing a file pointer from the file allocation table or marking storage space as available when a user issues a deletion request. The system logs record the operation time and operator ID as traceability information, while the physical data content is typically overwritten randomly by subsequent new data writes.

[0003] In traditional data security technologies, cloud-based file destruction typically involves only logical deletion or simple overwriting based on standard instructions, without delving into the physical layer to verify the actual data erasure status. This results in a high risk of data residue. Furthermore, existing traceability mechanisms often rely on centralized log records, lacking strong cryptographic evidence linked to the destruction action. This makes it difficult to provide an immutable and complete lifecycle closed-loop proof in a multi-tenant environment. When faced with storage media reuse or cross-domain migration scenarios, it cannot effectively resist malicious recovery or data residue restoration attacks by internal personnel. Summary of the Invention

[0004] The purpose of this invention is to address the shortcomings of existing technologies by proposing a cloud-based file destruction and data traceability method and system.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a cloud-based file destruction and data traceability method, comprising the following steps:

[0006] S1: Receive cloud archives to be destroyed, extract archive metadata, construct a full lifecycle traceability chain, use a distributed hash table algorithm to locate the cloud archives to be destroyed in the cloud storage cluster, obtain the cloud storage logical address, and resolve the cloud storage logical address into a set of physical storage units through the underlying driver mapping table.

[0007] S2: In response to the file destruction command, lock the physical storage unit set, generate an overwrite operation, drive the magnetic head or voltage controller to perform a forced write action on the physical storage unit set, collect the electromagnetic flip signal of the underlying storage medium, and construct a storage unit state flip matrix.

[0008] S3: Perform discrete cosine transform on the state flip matrix of the storage unit, extract high-frequency noise features, calculate Shannon entropy value, obtain noise benchmark, construct secure destruction baseline, compare Shannon entropy value with secure destruction baseline, if Shannon entropy value is greater than secure destruction baseline, generate cryptographic destruction proof, otherwise trigger secondary deep cleaning logic.

[0009] S4: Use an asymmetric encryption algorithm to perform digital signature processing on the cryptographic destruction proof, encapsulate the final block, link the final block to the end node of the full lifecycle traceability chain through a hash pointer, execute a logical deletion instruction, clear the file index metadata and access control key, and output a destruction report.

[0010] As a further aspect of the present invention, step S1 specifically comprises:

[0011] S11: Receive the cloud archive to be destroyed, parse the standard structure of the file header and the custom storage attribute tag, extract the archive metadata, verify the signature, initialize the basic genesis node of the full life cycle traceability chain according to the standard timestamp sequence and operation type feature field, and establish the metadata index topology structure.

[0012] The archive metadata includes the initial creation time, historical modification frequency, owner identification sequence, and corresponding security control level;

[0013] S12: Obtain the globally unique resource locator and shard hash sequence extracted from the metadata index topology, use them as hash digest input parameters and substitute them into the preset distributed hash table algorithm to perform consistent hash ring space routing calculation, traverse the virtual node routing mapping table clockwise to match the nearest available storage node, locate shard information, and aggregate to generate a logical address distribution set.

[0014] The logical address distribution set includes the cloud storage logical addresses of the cloud files to be destroyed distributed in a cloud storage cluster with a multi-availability zone architecture;

[0015] S13: Obtain the set of logical address distributions, call the virtual file system interface of the storage layer, query the underlying driver mapping table maintained in the storage controller, convert the three-dimensional logical block addresses one by one into physical layer absolute addressing parameters containing cylinder number, head number and sector number, resolve the cloud storage logical address into the real hardware storage space location, and aggregate to form the set of physical storage units.

[0016] The physical storage unit set encompasses all data shards.

[0017] As a further aspect of the present invention, step S2 specifically comprises:

[0018] S21: In response to the file destruction command, send the highest priority hardware latch signal to the input / output scheduling queue of the storage layer to lock the physical storage unit set, block all concurrent read and write requests, call the chaotic dynamic system equation to generate a multi-dimensional overwrite data stream with nonlinear pseudo-random characteristics, and construct a working instruction set;

[0019] The set of working instructions serves the overwrite operation;

[0020] S22: Obtain the working instruction set, send the underlying erase / write control message to the hard disk microcontroller or flash memory conversion layer through the storage bus, drive the mechanical disk head to perform random magnetization flip or control the solid-state drive voltage controller to apply erase voltage, perform a forced write action on the physical storage unit set in the locked state, and obtain the physical overwrite confirmation state.

[0021] S23: Obtain the physical overwrite confirmation status, synchronously activate the high-frequency digital-to-analog converter bypass monitoring circuit deployed on the storage node motherboard, collect the microscopic electromagnetic reversal signal generated by the underlying storage medium during the overwrite process of the physical storage unit set, quantize and align the continuously sampled analog signal features with the spatiotemporal dimension, and establish the storage unit state reversal matrix.

[0022] As a further aspect of the present invention, step S3 specifically comprises:

[0023] S31: Obtain the state flip matrix of the storage unit, map the discrete time-domain simulated sampling values ​​to a two-dimensional spatial array, perform two-dimensional discrete cosine transform on the pixel-level features of each row and column in the matrix, convert the data from the spatial domain to the frequency domain, separate the low-frequency basic background energy distribution, extract high-frequency noise features, and establish a frequency domain feature spectrum model.

[0024] The high-frequency noise characteristics are determined by the disordered abrupt changes in the physical state of the medium;

[0025] S32: Obtain the frequency domain feature spectrum model, use the probability density function to statistically analyze the energy distribution law and random fluctuation range of the high-frequency noise features in different frequency bands, calculate the Shannon entropy value of the current state system according to the uncertainty measurement formula in information theory, obtain the environmental noise benchmark by combining the noise-free spectrum parameters of the factory default storage medium, and construct the safe destruction baseline according to the threshold of the safety standard white paper.

[0026] S33: Obtain the Shannon entropy value and the secure destruction baseline, execute the numerical comparison logic based on floating-point precision. If the Shannon entropy value is greater than the secure destruction baseline, it proves that the original data has been completely covered by the physical chaotic state. Based on this result, call the elliptic curve digital signature algorithm to generate the cryptographic destruction proof. Otherwise, send an abnormal alarm signal to the storage controller and trigger the secondary deep cleaning logic.

[0027] The cryptographic destruction proof includes the device fingerprint;

[0028] The secondary deep cleaning logic includes alternating hot writing and cold power-on operations.

[0029] As a further aspect of the present invention, step S4 specifically comprises:

[0030] S41: Obtain the cryptographic destruction proof, call the pre-distributed asymmetric encryption system private key parameters, use the asymmetric encryption algorithm to perform digital signature processing on the hash digest of the proof content, package and aggregate the signed proof file, the network protocol address of the destruction execution node and the timestamp, perform encapsulation operation according to the underlying data structure standard of the blockchain ledger, and generate the final block;

[0031] The terminator block has a fixed storage capacity;

[0032] S42: Obtain the final block, calculate the secure hash algorithm digest value of the block header, construct a hash pointer pointing to the previous traceability node, link the final block to the end node of the full life cycle traceability chain in a chain-like encrypted reference manner, complete the full-link data closed-loop record of the file from creation to destruction, and generate on-chain solidified credentials.

[0033] S43: Obtain the on-chain solidified credential, submit a logical deletion instruction containing a forced execution flag to the cloud database management system, clear the file index metadata in the system registry and directory tree, simultaneously revoke and destroy all associated temporary and long-term access control keys, integrate the front-end destruction instruction and the back-end on-chain credential, and output a destruction report.

[0034] The destruction report contains a complete operation log.

[0035] As a further aspect of the present invention, the execution process of the consistent hashing ring space routing calculation specifically includes:

[0036] Obtain the globally unique resource locator and the shard hash sequence, extract the preset hash function algorithm feature parameters, convert the locator string into a fixed-length binary data stream, substitute it into the hash function for bit operations and non-linear obfuscation processing, map it to a logical ring numerical space composed of 2^32 virtual slots, and obtain the initial resource mapping hash point.

[0037] Obtain the initial resource mapping hash point, read the storage node configuration information table, perform multiple hash calculations on the network socket address and hardware serial number of all available physical nodes according to the same hash function, generate virtual node mapping coordinates, synchronously project the coordinates into the same logical ring numerical space, and establish a node spatial distribution topology.

[0038] The virtual node mapping coordinates are multiple dispersed coordinates generated for a single physical node;

[0039] Obtain the node spatial distribution topology, take the initial resource mapping hash point as the starting cursor, perform a one-way scan traversal along the positive clockwise trajectory of the logical ring numerical space with increasing numerical values, identify and lock the first hit virtual node mapping coordinates, confirm its corresponding real physical storage node by reverse lookup of the routing table, and establish a routing binding relationship.

[0040] The routing binding relationship refers to the direct mapping between resource locators and actual physical hardware addresses.

[0041] As a further aspect of the present invention, the construction process of the working instruction set specifically includes:

[0042] The spatial capacity and current operating temperature parameters of the physical storage unit set are obtained, the initial fractal variables and time evolution step size of the chaotic dynamic equation are set, and a nonlinear calculation model coupled with the logistic mapping function and the Lorentz attractor is adopted to perform iterative calculation, generate a floating-point number sequence, and establish an initial chaotic data source.

[0043] The floating-point sequence has multidimensional pseudo-random characteristics and high initial value sensitivity;

[0044] The initial chaotic data source is obtained, and double-precision bitwise operations are performed to truncate each floating-point value in the sequence. The unordered bit stream of the mantissa is extracted, and the data is reassembled and packaged according to the page erase alignment rules of the storage medium. The weight parameters of the multiple overwrite instructions are calculated. The calculation formula is:

[0045] ;

[0046] in, Represents the weight parameter of multiple overwrite instructions. Represents the temperature compensation coefficient. This represents the spatial capacity of the physical storage unit set. Represents the baseline space capacity. This represents the current operating temperature parameter. Represents the reference operating temperature parameter. Represents the time decay factor. Represents the number of iterations in the chaotic sequence, generating a standardized overwrite data packet;

[0047] Obtain the standardized overwrite data packet and the weight parameters of the multiple overwrite instructions, dynamically adjust the sending frequency and concurrent writing depth of different data packets in the underlying device queue according to the weight parameters, combine them to form an overwrite strategy, and output the working instruction set according to the strategy.

[0048] The coverage strategy is a multiple alternation scheme targeting different storage sectors;

[0049] The set of working instructions has spatiotemporal disorder characteristics.

[0050] As a further aspect of the present invention, the calculation process of the Shannon entropy specifically includes:

[0051] The frequency domain characteristic spectrum model is obtained, and the entire high-frequency analysis bandwidth is divided into a continuous and non-overlapping sub-band sequence according to the preset discrete frequency band division rules. The instantaneous power spectral density values ​​in each sub-band are extracted using a bandpass filter bank, and the absolute energy integral is calculated.

[0052] The absolute energy integral corresponds to each of the sub-frequency bands;

[0053] The absolute energy integral is obtained, the total global radiation energy within the high-frequency analysis bandwidth is statistically analyzed, the ratio of the absolute energy integral of a single sub-band to the total global radiation energy is calculated, and after normalization, a discrete probability distribution vector is obtained to establish a system microstate probability distribution model.

[0054] The discrete probability distribution vector reflects the probability distribution pattern of noise features occurring in each sub-band.

[0055] Obtain the discrete probability distribution vector in the probability distribution model of the system's microstate. For each probability element in the vector, calculate its product with its base-2 logarithm. Then, sum the product results of all dimensions and take the negative value to obtain the Shannon entropy value.

[0056] The Shannon entropy value is determined by the absolute degree of disorder in the physical magnetic state of the underlying storage medium.

[0057] As a further aspect of the present invention, the encapsulation process of the termination block specifically includes:

[0058] Obtain the signed proof document, call the secure hash algorithm to perform a secondary mixed hash operation on the main content of the document and the signature-related information to generate a transaction certificate hash value, use it as the digital fingerprint of the signed proof document, and establish the block transaction data payload;

[0059] The hash value of the transaction certificate is a globally unique identifier with a fixed length of 256 bits.

[0060] Obtain the network protocol address and timestamp of the destruction execution node, and combine it with the protocol version number of the current blockchain network, the proof-of-work parameters of the consensus mechanism, and the header hash of the previous verified block, and construct the block header metadata set according to the node organization logic of the standard Merkle tree.

[0061] The transaction data payload of the block and the metadata set of the block header are obtained. The two are concatenated and encoded at the byte level through the underlying serialization interface. Cross-node verification cyclic redundancy code of the network transport layer is applied to lock all data segments in an immutable distributed ledger storage structure and encapsulate them to form a final block.

[0062] The final block is broadcast to all nodes in the network for consensus verification.

[0063] A cloud-based file destruction and data traceability system, the system being used to implement the aforementioned cloud-based file destruction and data traceability method, the system comprising:

[0064] The archive location and parsing module is used to receive cloud archives to be destroyed, extract archive metadata, build a full life cycle traceability chain, use a distributed hash table algorithm to locate the cloud archives to be destroyed in the cloud storage cluster, obtain the cloud storage logical address, and resolve the cloud storage logical address into a set of physical storage units through the underlying driver mapping table.

[0065] The state overwrite acquisition module is used to respond to the file destruction command, lock the physical storage unit set, generate an overwrite operation, drive the magnetic head or voltage controller to perform a forced write action on the physical storage unit set, acquire the electromagnetic flip signal of the underlying storage medium, and construct the storage unit state flip matrix.

[0066] The feature extraction and determination module is used to perform discrete cosine transform on the state flip matrix of the storage unit, extract high-frequency noise features, calculate Shannon entropy value, obtain noise benchmark, construct secure destruction baseline, compare Shannon entropy value with secure destruction baseline, and generate cryptographic destruction proof if Shannon entropy value is greater than secure destruction baseline; otherwise, trigger secondary deep cleaning logic.

[0067] The closed-loop destruction and cleanup module is used to perform digital signature processing on the cryptographic destruction proof using an asymmetric encryption algorithm, encapsulate the final block, link the final block to the end node of the full lifecycle traceability chain through a hash pointer, execute logical deletion instructions, clear the file index metadata and access control keys, and output a destruction report.

[0068] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0069] In this invention, by constructing a full lifecycle traceability chain and analyzing the set of physical storage units, precise positioning of cloud data from the logical layer to the physical layer is achieved. By utilizing chaotic sequence multiple overwrite operations and entropy detection of the storage unit state flip matrix, the residual magnetic characteristics of the data are forcibly destroyed at the physical layer, ensuring the physical irrecoverability of the data. This effectively solves the hidden danger of false deletion caused by virtualization layer isolation in the cloud environment. Combined with cryptographic destruction proof and chain locking of the final block, it provides legally valid immutable mathematical evidence for the destruction of archives, and thoroughly establishes a secure closed loop for the entire lifecycle of data from generation to complete annihilation. Attached Figure Description

[0070] Figure 1 This is a flowchart illustrating the overall process of the cloud-based file destruction and data traceability method of the present invention.

[0071] Figure 2 This is a detailed flowchart of the cloud file location and logical address parsing process of this invention;

[0072] Figure 3 This is a detailed flowchart of the physical locking and overwriting process of the storage unit in this invention;

[0073] Figure 4 This is a detailed flowchart of the high-frequency noise feature extraction and security determination process of the present invention;

[0074] Figure 5 This is a detailed flowchart of the destruction proof on-chain and file logic clearing for this invention. Detailed Implementation

[0075] To make the objectives, technical solutions, and advantages of this invention clearer, the software-based technical solution is described in detail below with reference to system architecture diagrams and embodiments. It should be understood that the specific embodiments described herein are only for explaining the technical solutions of this invention and do not constitute a limitation on the scope of protection.

[0076] In the description of this invention, the system architecture relationships or data processing flows indicated by terms such as "layer," "module," "interface," "data flow," "client," and "server" are all defined based on the architecture diagram or flowchart corresponding to the embodiments. This way of describing is only used to clearly illustrate the logical relationships between the elements in the technical solution, and not to limit the physical deployment form. The term "multiple" includes two or more technical units, including but not limited to multiple data nodes, processing threads, service instances, or functional components and other scalable elements. The specific number is determined according to the actual business scenario and needs to be specifically specified.

[0077] Please see Figure 1 and Figure 2 This invention provides a technical solution: a method for cloud-based file destruction and data traceability, comprising the following steps:

[0078] S1: Receive cloud archives to be destroyed, extract archive metadata, build a full lifecycle traceability chain, use a distributed hash table algorithm to locate the cloud archives to be destroyed in the cloud storage cluster, obtain the cloud storage logical address, and resolve the cloud storage logical address into a set of physical storage units through the underlying driver mapping table.

[0079] S11: Receive cloud files to be destroyed, parse the standard structure of the file header and custom storage attribute tags, extract file metadata, verify signatures, initialize the basic genesis node of the full lifecycle traceability chain based on standard timestamp sequences and operation type feature fields, and establish metadata index topology structure.

[0080] Archive metadata includes initial creation time, historical modification frequency, owner identification sequence, and corresponding security control level;

[0081] S12: Obtain the globally unique resource locator and shard hash sequence extracted from the metadata index topology, use them as hash digest input parameters and substitute them into the preset distributed hash table algorithm to perform consistent hash ring space routing calculation, traverse the virtual node routing mapping table clockwise to match the nearest available storage node, locate shard information, and aggregate to generate a logical address distribution set.

[0082] The logical address distribution set contains the cloud storage logical addresses of the cloud files to be destroyed, distributed across a multi-availability zone cloud storage cluster.

[0083] The execution process of consistent hashing ring space routing calculation specifically includes:

[0084] Obtain the globally unique resource locator and the shard hash sequence, extract the preset hash function algorithm feature parameters, convert the locator string into a fixed-length binary data stream, substitute it into the hash function for bit operations and non-linear obfuscation processing, map it to a logical ring numerical space composed of 2^32 virtual slots, and obtain the initial resource mapping hash point.

[0085] Obtain the initial resource mapping hash point, read the storage node configuration information table, perform multiple hash calculations on the network socket address and hardware serial number of all available physical nodes according to the same hash function, generate virtual node mapping coordinates, synchronously project the coordinates into the same logical ring numerical space, and establish the node spatial distribution topology.

[0086] The virtual node mapping coordinates are multiple scattered coordinates generated corresponding to a single physical node;

[0087] Obtain the node spatial distribution topology. Using the initial resource mapping hash point as the starting cursor, perform a one-way scan traversal along the positive clockwise trajectory of the logical ring numerical space with increasing values. Identify and lock the first hit virtual node mapping coordinates. Confirm its corresponding real physical storage node by reverse lookup of the routing table and establish a routing binding relationship.

[0088] Routing binding refers to the direct mapping between resource locators and actual physical hardware addresses;

[0089] S13: Obtain the set of logical address distributions, call the virtual file system interface of the storage underlying layer, query the underlying driver mapping table maintained in the storage controller, convert the three-dimensional logical block addresses one by one into physical layer absolute addressing parameters containing cylinder number, head number and sector number, resolve the cloud storage logical address into the real hardware storage space location, and aggregate to form a set of physical storage units.

[0090] The physical storage unit set encompasses all data fragments.

[0091] The system receives a 1024-megabyte cloud archive data stream to be destroyed. It locates the contiguous storage area with an offset of 1024 bytes from the beginning of the data stream header, sequentially segments and reads the header binary stream, parsing out an 8-byte standard timestamp sequence, a 4-byte operation type feature field, and a 32-byte custom storage attribute tag. From the parsed tag, it extracts archive metadata, including a 10-byte initial creation time timestamp, a 4-byte historical modification frequency count, a 16-letter alternating alphanumeric owner identifier sequence, and a security control level represented by numbers from 1 to 5. After extracting the archive metadata, it extracts a 256-byte digital signature verification segment appended to the end of the file. The system calls a pre-defined public key parameter library to extract a 2048-bit asymmetric encryption public key bound to the owner's identity sequence. The 1024-megabyte original file payload data is divided into consecutive 64-byte blocks according to the 256-bit output standard of a secure hash algorithm. After 64 rounds of internal loop processing including bitwise operations and nonlinear functions, a 32-byte original data digest fingerprint is generated. The extracted digital signature verification segment is decrypted and restored using a large number modular exponentiation operation performed with the 2048-bit public key. The restored digest data is compared with the recalculated original data digest fingerprint. When all bits in both match one-to-one, the digital signature verification operation is completed. Based on the verified standard timestamp sequence and operation type feature field, the basic genesis node of the full lifecycle traceability chain is initialized in the local relational database. A block height identifier with number 0 is assigned to this genesis node. A metadata index topology structure is established with archive metadata as the primary key and multi-level resource identifiers as foreign keys, forming a node association network in a directed acyclic graph format.

[0092] The system extracts a 128-byte globally unique resource locator (GLOR) from the metadata index topology and a 16-byte hash sequence corresponding to each of the 64 distributed data fragments. These are used as input parameters for a hash digest and fed into a pre-defined distributed hash table algorithm for consistent hashing ring space routing calculation. The system extracts the first version of the pre-defined message digest algorithm's feature parameters, initializes five 32-bit link variables, and sequentially converts the 128-byte GLOR string into the corresponding binary data stream of the US Standard Code for Information Interchange (SICE). The data stream is padded with a high-order 1 and several zeros until the total length satisfies the condition that the modulo 512 equals 448. Finally, 64 bits of the original data length information are appended. The padded data stream is divided into multiple 512-bit data blocks, each of which is further divided into 16 32-bit words. By substituting the words into a hash function, a total of 80 bitwise operations including logical AND, logical OR, XOR, and circular left shift are performed, along with non-linear obfuscation. Five linked variables are updated, and finally, the five variables are concatenated and output as a 160-bit initial resource mapping hash point. The lower 32 bits are then extracted and mapped to a logical ring-shaped numerical space consisting of 2^32 virtual slots, yielding the corresponding decimal coordinate value of 1856739021. The storage node configuration information table is read to locate 50 physical nodes in the cloud storage cluster that are in an available state. The 32-bit network socket address and 48-bit hardware media access control sequence number of each node are read. The concatenated string of network socket address and hardware sequence number is subjected to multiple hash calculation according to the same message digest algorithm as above. Combined with the salted random number set in the node configuration, 100 virtual node mapping coordinates are generated for each physical node. The 32-bit hash truncation values ​​of a total of 5000 virtual nodes are synchronously projected into the same logical ring numerical space as above to establish the node spatial distribution topology.

[0093] The topology of the node's spatial distribution is obtained. Starting with the initial resource mapping hash point of 1856739021 as the cursor, a unidirectional clockwise scan is performed along the logical ring numerical space, increasing from 0 to 4294967295. The first virtual node mapping coordinate with a value greater than the starting cursor is identified and locked, with a locked value of 1856740000. The actual physical storage node corresponding to the virtual node's network protocol address 192.168.10.55 is confirmed by reverse lookup of the routing table, establishing a direct routing binding relationship between the resource locator and the actual physical hardware address. The virtual node routing mapping table is traversed clockwise to match the nearest available storage node, locating the storage node information for all 64 data shards. The logical unit number and logical block offset corresponding to all shards are aggregated to generate a logical address distribution set covering the distribution across three data centers in the multi-availability zone architecture.

[0094] The system retrieves the logical address distribution set, calls the virtual file system interface of the server operating system kernel-mode storage, and issues a query command with administrator privileges to traverse the underlying driver mapping table maintained in the storage controller. Each 48-bit three-dimensional logical block address in the set is converted into physical-level absolute addressing parameters. Given a logical block address value of 15000000, the system reads the hard drive firmware configuration parameters to determine the number of sectors per track as 63 and the number of heads as 16. It then performs division and modulo operations: dividing 15000000 by 1008 (the product of 63 and 16) yields a cylinder number of 14880; dividing 15000000 by 63 and taking the modulo of the quotient with 16 yields a head number of 15; and taking the modulo of 15000000 with 63 and adding 1 yields a sector number of 16. The above calculations resolve all logical addresses of cloud storage into hardware storage space locations containing actual cylinder, head, and sector parameters, converging them into a physical storage unit set encompassing all 64 data shards. Experimental data shows that in a test scenario involving concurrent resolution of 500 shard logical addresses, the latency of this hierarchical mapping addressing algorithm remains within 15 milliseconds, representing an 85% improvement in address location efficiency compared to the traditional file system's directory tree traversal addressing method.

[0095] Please see Figure 1 and Figure 3 S2: Respond to the file destruction command, lock the physical storage unit set, generate an overwrite operation, drive the magnetic head or voltage controller to perform a forced write action on the physical storage unit set, collect the electromagnetic reversal signal of the underlying storage medium, and construct the storage unit state reversal matrix.

[0096] S21: In response to the file destruction command, send the highest priority hardware latch signal to the input / output scheduling queue of the storage layer, lock the set of physical storage units, block all concurrent read and write requests, call the chaotic dynamic system equation to generate a multi-dimensional overwrite data stream with nonlinear pseudo-random characteristics, and construct the working instruction set.

[0097] The instruction set serves the overwrite operation;

[0098] The process of constructing the work instruction set specifically includes:

[0099] The system obtains the spatial capacity and current operating temperature parameters of the physical storage unit set, sets the initial fractal variables and time evolution step size of the chaotic dynamic equation, adopts a nonlinear calculation model coupled with the logistic mapping function and the Lorentz attractor, performs iterative calculations, generates a floating-point sequence, and establishes the initial chaotic data source.

[0100] Floating-point sequences exhibit multidimensional pseudo-random characteristics and high sensitivity to initial values;

[0101] The process involves acquiring an initial chaotic data source, performing double-precision bitwise operations to truncate each floating-point value in the sequence, extracting the unordered bit stream from the mantissa, reassembling the data and packaging it according to the page erase alignment rules of the storage medium, and calculating the weight parameters of the multiple overwrite instructions. The calculation formula is:

[0102] ;

[0103] in, Represents the weight parameter of multiple overwrite instructions. Represents the temperature compensation coefficient. Represents the spatial capacity of a collection of physical storage units. Represents the baseline space capacity. This represents the current operating temperature parameter. Represents the reference operating temperature parameter. Represents the time decay factor. Represents the number of iterations in the chaotic sequence, generating a standardized overwrite data packet;

[0104] Obtain standardized overwrite data packets and multiple overwrite instruction weight parameters, dynamically adjust the sending frequency and concurrent writing depth of different data packets in the underlying device queue based on the weight parameters, combine them to form an overwrite strategy, and output a set of working instructions based on the strategy.

[0105] The overlay strategy is a multiple-alternation scheme targeting different storage sectors;

[0106] The working instruction set has the characteristics of spatiotemporal disorder;

[0107] S22: Obtain the working instruction set, send the underlying erase / write control message to the hard disk microcontroller or flash memory conversion layer through the storage bus, drive the mechanical disk head to perform random magnetization flipping or control the solid-state drive voltage controller to apply erase voltage, perform a forced write action on the physical storage unit set in the locked state, and obtain the physical overwrite confirmation state.

[0108] S23: Obtain the physical overwrite confirmation status, synchronously activate the high-frequency digital-to-analog converter bypass monitoring circuit deployed on the storage node motherboard, collect the microscopic electromagnetic reversal signal generated by the underlying storage medium during the overwrite process of the physical storage unit set, quantize and align the characteristics of the continuously sampled analog signal with the spatiotemporal dimension, and establish the storage unit state reversal matrix.

[0109] In response to the forced file destruction command issued by the front-end console, a high-priority hardware latch signal with interrupt vector number 255 is directly injected into the input / output scheduling queue of the underlying storage layer. This locks the physical storage unit set obtained in step S1, modifies the access permission bitmask at the underlying driver level, and suspends and moves all concurrent read and write requests for this physical storage unit set into a blocking queue. The chaotic dynamics system equations deployed inside the controller are invoked to generate a multi-dimensional overwrite data stream with nonlinear pseudo-random characteristics to construct the working instruction set. The spatial capacity parameter of the physical storage unit set is obtained; the current set capacity is 512 gigabytes. The current operating temperature parameter is read from the temperature sensor bus built into the hard drive motherboard as 45 degrees Celsius. The initial fractal variable of the chaotic dynamics equations is set to 0.654321, the time evolution step is 0.01 seconds, and a nonlinear calculation model coupled with a logistic mapping function and a Lorentz attractor is adopted. The initial fractal variables are substituted into the nonlinear equation and iterative calculations are performed 1000 times. In each iteration, coordinate projection points on the three-dimensional coordinate system are extracted to generate an initial chaotic data source containing a sequence of 1000 double-precision floating-point numbers. For each value in this floating-point sequence, 64-bit double-precision bit operations are performed to truncate it, discarding the high-order exponent and sign bits, and extracting the 52-bit mantissa to construct a continuous unordered bit stream. This unordered bit stream is then filled, sliced, and packaged according to the 4-kilobyte page erase / write alignment rule of the flash memory chip.

[0110] Introducing physical space capacity, operating temperature, and time iteration parameters, we begin calculating the weight parameters of the multiple overwrite instructions. The weight parameters are calculated using the following formula:

[0111] ;

[0112] in, Represents the temperature compensation coefficient. Represents the spatial capacity of a collection of physical storage units. Represents the baseline space capacity. This represents the current operating temperature parameter. Represents the reference operating temperature parameter. Represents the time decay factor. This represents the number of iterations in the chaotic sequence.

[0113] Obtain the space capacity parameters of the above physical storage unit set. The value is 512, which is the baseline space capacity parameter. The preset value is 256, and this is the current operating temperature parameter. The value is 45, which is the reference operating temperature parameter. The default value is 25. Set the temperature compensation coefficient. The value is 1.5, the time decay factor. The value is 0.8, which is used to obtain the number of iterations of the chaotic sequence as set above. The value is 1000. The capacity parameter is... (512) and baseline capacity (256) Substituting the values ​​into the ratio term, the capacity ratio is calculated to be 2. The operating temperature... (45) and reference operating temperature (25) Substituting the ratio term, the temperature ratio is calculated to be 1.8. Dividing the capacity ratio (2) by the temperature ratio (1.8), the environmental carrying capacity coefficient is 1.111. The temperature compensation coefficient... Multiplying (1.5) by the environmental carrying capacity coefficient (1.111), the foundation overwriting momentum value is calculated to be 1.666. The number of iterations is then obtained. (1000) and calculate its natural logarithm as 6.907, then compare this logarithm with the time decay factor. Multiplying by (0.8) yields a chaos deep penetration factor of 5.525. Adding the base overwrite momentum value (1.666) to the chaos deep penetration factor (5.525) finally yields the weight parameters of the multiple overwrite instructions. The value is 7.191. This result indicates that the current physical load of the device is optimally coupled with the chaotic random number generation depth, and overwriting will be performed at a moderately high write frequency.

[0114] Table 1. Underlying Hardware Resource Parameters and Dynamic Calculation Parameters:

[0115] Parameter Chinese name Parameter collected values Space capacity of physical storage unit set 512 Baseline space capacity 256 Current operating temperature parameters 45 Reference operating temperature parameters 25 Chaotic sequence iteration count 1000 Temperature compensation coefficient 1.5 Time decay factor 0.8 Calculation result of multiple overwrite instruction weight parameters 7.191

[0116] Referring to Table 1, the dynamic adjustment calculation of the overwrite weight parameters can be effectively realized by monitoring the actual capacity and temperature values.

[0117] Based on the calculated weight parameter of 7.191, standardized overwrite data packets are generated. The frequency of data packet distribution in the underlying peripheral component interconnect standard bus device queue is dynamically adjusted to 500 times per second, and the concurrent write depth queue length is set to 32. This forms an overwrite strategy covering 5 random full-disk alternating writes. Based on the strategy, a working instruction set with spatiotemporally disordered characteristics is output. This working instruction set is obtained, and a low-level erase / write control message carrying opcode 0x04 is sent to the hard disk microcontroller or flash memory conversion layer via the storage bus. This controls the voltage controller inside the solid-state drive to apply a 20-volt erase voltage, breaking down the oxide layer of the floating gate transistor. A forced high-level write action is performed on the physical storage cell set in the locked state, overwriting the original charge state. After the microcontroller sends back a confirmation character indicating that the write is complete, a physical overwrite confirmation status is obtained. Simultaneously, the high-frequency digital-to-analog converter bypass monitoring circuit deployed on the storage node motherboard is activated to collect the microscopic electromagnetic reversal signals generated by the underlying storage medium during the overwrite process of the physical storage cell set. By using an analog-to-digital converter with a sampling frequency of 100 MHz, the voltage transient analog signal during the overwrite period is continuously captured. The characteristics of the continuously sampled analog signal are quantized with 8-bit precision, and the spatiotemporal dimensions are aligned according to the specification of 1024 points per sector, to establish a two-dimensional storage cell state flip matrix with dimensions of 1024 rows by 64 columns.

[0118] Please see Figure 1 and Figure 4 S3: Perform discrete cosine transform on the state flip matrix of the storage unit, extract high-frequency noise features, calculate the Shannon entropy value, obtain the noise baseline, construct the secure destruction baseline, compare the Shannon entropy value with the secure destruction baseline, if the Shannon entropy value is greater than the secure destruction baseline, generate a cryptographic destruction proof, otherwise trigger the second deep cleaning logic.

[0119] S31: Obtain the storage unit state flip matrix, map the discrete time-domain simulated sampling values ​​to a two-dimensional spatial array, perform two-dimensional discrete cosine transform on the pixel-level features of each row and column in the matrix, convert the data from the spatial domain to the frequency domain, separate the low-frequency basic background energy distribution, extract high-frequency noise features, and establish a frequency domain feature spectrum model.

[0120] High-frequency noise characteristics are determined by the disordered abrupt changes in the physical state of the medium;

[0121] S32: Obtain the frequency domain characteristic spectrum model, use the probability density function to statistically analyze the energy distribution law and random fluctuation range of high frequency noise characteristics in different frequency bands, calculate the Shannon entropy value of the current state system according to the uncertainty measurement formula in information theory, obtain the environmental noise benchmark by combining the noise-free spectrum parameters of the factory default storage medium, and construct a safe destruction baseline based on the threshold of the safety standard white paper.

[0122] The calculation process of Shannon entropy specifically includes:

[0123] The frequency domain characteristic spectrum model is obtained. According to the preset discrete frequency band division rules, the entire high-frequency analysis bandwidth is divided into a continuous and non-overlapping sub-frequency band sequence. The instantaneous power spectral density values ​​in each sub-frequency band are extracted using a bandpass filter bank, and the absolute energy integral is calculated.

[0124] The absolute energy integral corresponds to each sub-frequency band;

[0125] Obtain the absolute energy integral, statistically analyze the total global radiation energy within the high-frequency analysis bandwidth, calculate the ratio between the absolute energy integral of a single sub-band and the total global radiation energy, and obtain the discrete probability distribution vector after normalization to establish a probability distribution model of the system's microstate.

[0126] The discrete probability distribution vector reflects the probability distribution pattern of noise characteristics occurring in each sub-band.

[0127] Obtain the discrete probability distribution vector in the probability distribution model of the system's microstate. For each probability element in the vector, calculate its product with its base-2 logarithm. Then, sum the product results of all dimensions and take the negative value to obtain the Shannon entropy value.

[0128] Shannon entropy is quantified by the absolute degree of disorder in the physical magnetic state of the underlying storage medium;

[0129] S33: Obtain the Shannon entropy value and the secure destruction baseline, and execute the numerical comparison logic based on floating-point precision. If the Shannon entropy value is greater than the secure destruction baseline, it proves that the original data has been completely covered by the physical chaotic state. Based on this result, call the elliptic curve digital signature algorithm to generate a cryptographic destruction proof. Otherwise, send an abnormal alarm signal to the storage controller and trigger the secondary deep cleaning logic.

[0130] Cryptographic destruction proof includes the device fingerprint;

[0131] The secondary deep cleaning logic includes alternating operations of hot writing and cold power-on.

[0132] A 1024-row, 64-column two-dimensional storage cell state flip matrix containing microscopic voltage sampling data during the overwrite period is obtained. This matrix maps the discrete-time domain analog sampling values ​​to a two-dimensional spatial data array in memory. A two-dimensional discrete cosine transform (DCT) calculation is performed, using a pre-defined basis function cosine transform kernel matrix of the same size. When operating on each row of the spatial array, the vector of 1024 elements in each row is multiplied by a projection column vector generated from the discretization of cosine functions of different frequencies, producing 1024 one-dimensional horizontal frequency domain coefficients. Then, each column of the new matrix composed of these 1024 rows is extracted, and the 64-element vector of each column is multiplied by the corresponding vertical frequency domain cosine projection matrix to obtain the final two-dimensional DCT frequency domain coefficient matrix. Through this cascaded linear projection matrix multiplication operation, the data is completely transformed from the spatial domain composed of time and voltage amplitude to the frequency domain composed of spatially varying frequencies and energy density distribution. Traverse the generated frequency domain coefficient matrix, retain the low-frequency basic background energy distribution coefficients in the upper left corner where the sum of the horizontal and vertical index subscripts is less than or equal to 20, set them to zero or separate them for backup, and extract the coefficient values ​​of the remaining positions where the sum of the horizontal and vertical index subscripts is greater than 20. These coefficient values ​​characterize the high-frequency noise features caused by the disordered abrupt change in the physical state of the medium. Combine these features to construct a frequency domain feature spectrum model.

[0133] A frequency domain characteristic spectrum model is obtained. According to a preset discrete frequency band division rule, the entire high-frequency analysis bandwidth from 1 MHz to 50 MHz is divided into 10 continuous and non-overlapping sub-band sequences with a bandwidth of 5 MHz. Using a Butterworth bandpass filter bank in digital signal processing, the instantaneous power spectral density values ​​within these 10 sub-bands are extracted. The discrete power spectral density points within each sub-band are summed to calculate the absolute energy integrals corresponding to 10 different sub-bands. The absolute energy integral of a certain sub-band in the 1-5 MHz interval within the statistical period is calculated to be 15000. Summing the absolute energy integrals of the 10 sub-bands across the entire bandwidth yields a total global radiated energy of 185000. The absolute energy integral of a single sub-band is compared with the total global radiated energy. For example, 15000 is divided by 185000 to get 0.081. The above ratio calculation is performed sequentially for normalization to obtain a discrete probability distribution vector containing 10 probability elements, and a probability distribution model of the system's microstate is established.

[0134] The aforementioned Butterworth bandpass filter bank refers to an electronic filter with a maximum flat amplitude response. It has no ripple in the passband and decreases monotonically in the stopband, and is used to accurately separate signal components in a specific frequency range.

[0135] The discrete probability distribution vector generated from the system's microstate probability distribution model is obtained. This vector contains 10 probability values ​​between 0 and 1, with the sum of all probability values ​​being 1. For the first probability element 0.081 in the vector, its base-2 logarithm is calculated as -3.626. Multiplying 0.081 by its logarithm (-3.626) yields -0.293. The same product calculation is performed on the other nine dimensions. The sum of all ten product results is accumulated, yielding a sum of -2.854. Negating this sum gives the Shannon entropy value of 2.854, which quantifies the absolute disorder of the underlying storage medium's physical magnetic state. The noise-free spectral parameters of the factory-default solid-state drive storage medium in its unwritten state are extracted as the environmental noise benchmark. Combined with the minimum tolerance for random disturbances specified in the safety standard white paper, the environmental noise benchmark is increased by 30%, resulting in a safe destruction baseline of 2.500.

[0136] The calculated Shannon entropy value of 2.854 is compared with the constructed secure destruction baseline of 2.500, and a numerical comparison logic based on 64-bit double-precision floating-point precision is executed. It is determined that 2.854 is strictly greater than 2.500, satisfying the comparison condition. This result indicates that the randomness of electromagnetic noise generated in the physical storage unit array has significantly exceeded the security limit, proving that the original cloud archive data has been completely covered by a physical chaotic state and cannot be restored by hysteresis. Based on this decision, an encryption subroutine containing an elliptic curve cryptography algorithm digital certificate mechanism is invoked. Using the recognized secure standard curve parameters as the elliptic curve generator point, a 256-bit private key integer is randomly selected within the prime number field. The scalar multiplication operation between the base point and the private key is calculated to obtain the public key coordinates. The Shannon entropy calculation result, the corresponding timestamp, and the unique 48-bit media access control hardware address of the physical device are concatenated to form the message text to be signed. The elliptic curve digital signature algorithm is executed to obtain a cryptographic destruction proof containing two 256-bit integers representing the signature and the device fingerprint sequence. Experimental data shows that when the secure destruction baseline is set to 2.500, the accuracy of determining the randomness of overwrite using Shannon entropy reaches 99.9%, which is 45% higher than the traditional method that relies solely on the write success flag. If the Shannon entropy value calculated by the system is less than or equal to 2.500, a pulse width modulation (PWM) anomaly alarm signal is immediately sent to the system-on-a-chip (SoC) storage controller mounted on the main control board, simultaneously triggering a secondary deep cleaning logic. This cleaning logic sends a self-test protocol temperature and pressure control command to the SSD controller, forcibly increasing the power consumption of the driver chip to generate an 80-degree Celsius hot write state to cover the storage area once, followed by actively cutting off the power supply for 5 seconds and performing a cold power-on state to cover once. Hot write and cold power-on are executed alternately until the re-collected Shannon entropy value is greater than the secure destruction baseline.

[0137] Table 2 Measurement of Microscopic Frequency Domain Energy Distribution of Memory Cell Status:

[0138] Discrete frequency band distribution interval division Absolute energy integral measurement 1 to 5 15000 6 to 10 23000 11 to 15 38000 16 to 20 12000 21 to 50 97000

[0139] Table 2 presents the distribution range of each sub-band used to calculate the disorder level of the system and the corresponding microscopic frequency domain energy measurement results. By obtaining high-frequency noise characteristic data of specific frequency bands, the accurate acquisition and quantification of the parameters substituted into the information theory entropy value calculation formula are ensured.

[0140] Please see Figure 1 and Figure 5 S4: Use an asymmetric encryption algorithm to perform digital signature processing on the cryptographic destruction proof, encapsulate the final block, link the final block to the end node of the full lifecycle traceability chain through a hash pointer, execute the logical deletion instruction, clear the file index metadata and access control key, and output the destruction report;

[0141] S41: Obtain the cryptographic destruction proof, call the pre-distributed asymmetric encryption system private key parameters, use the asymmetric encryption algorithm to perform digital signature processing on the hash digest of the proof content, package and aggregate the signed proof file, the network protocol address of the destruction execution node and the timestamp, perform the encapsulation operation according to the underlying data structure standard of the blockchain ledger, and generate the final block;

[0142] Termination blocks have a fixed storage capacity;

[0143] The encapsulation process of a finalized block specifically includes:

[0144] Obtain the signed proof document, call the secure hash algorithm to perform a secondary mixed hash operation on the main content of the document and the signature-related information to generate a transaction certificate hash value, use it as the digital fingerprint of the signed proof document, and establish the block transaction data payload;

[0145] The transaction certificate hash value is a globally unique identifier with a fixed length of 256 bits;

[0146] Obtain the network protocol address and timestamp of the destruction execution node, and combine it with the protocol version number of the current blockchain network, the proof-of-work parameters of the consensus mechanism, and the header hash of the previous verified block, and construct the block header metadata set according to the node organization logic of the standard Merkle tree.

[0147] Obtain the block transaction data payload and the block header metadata set, perform byte-level concatenation and encoding conversion on the two through the underlying serialization interface, apply cross-node verification cyclic redundancy code of the network transport layer, lock all data segments in an immutable distributed ledger storage structure, and encapsulate them to form a final block;

[0148] The final block is broadcast to all nodes in the network for consensus verification;

[0149] S42: Obtain the final block, calculate the secure hash algorithm digest value of the block header, construct a hash pointer pointing to the previous traceability node, link the final block to the end node of the full life cycle traceability chain in a chain-like encrypted reference manner, complete the full-link data closed-loop record of the file from creation to destruction, and generate on-chain solidified credentials.

[0150] S43: Obtain on-chain solidified credentials, submit a logical deletion command containing a forced execution flag to the cloud database management system, clear file index metadata in the system registry and directory tree, synchronously revoke and destroy all associated temporary and long-term access control keys, integrate the front-end destruction command and the back-end on-chain credentials, and output a destruction report.

[0151] The destruction report includes a complete operation log.

[0152] Obtain the cryptographic destruction proof file generated in step S3, and call the 4096-bit asymmetric encryption system private key parameter, which is pre-distributed and protected by a hardware security module in the cloud key management microservice interface. Call the 256-bit version of the secure hash algorithm to perform hash calculation on the full text of the cryptographic destruction proof file, generating a 256-bit digest character sequence. Substitute the system private key parameter and the digest character sequence into the asymmetric encryption large number modular exponentiation framework to perform digital signature processing, generating a 512-byte signature ciphertext segment. Obtain the network protocol address of the server where the operation execution node is located, which is 192.168.100.55, and synchronously obtain the 13-bit timestamp value accurate to milliseconds synchronized by the National Time Service Center's network time protocol. Serialize, package, and aggregate the signature ciphertext segment, node network protocol address, and timestamp according to the key-value pair text protocol structure, and perform encapsulation operations according to the standard block data structure of the underlying blockchain ledger to generate a final block with a fixed storage capacity of 1 megabyte.

[0153] The detailed encapsulation process for this final block is executed as follows: The signed proof document is obtained, and a secure hash algorithm is used to concatenate the main text fragment of the document with the signature-related information, including the network protocol address and timestamp. A secondary hybrid hash operation, consisting of 64 rounds of internal bitwise operations and XOR logic processing, is performed on the merged long string to generate a fixed-length 256-bit (32-byte) transaction credential hash value. This hexadecimal hash string serves as the unique digital fingerprint of this data destruction event across the entire network, thus establishing the block transaction data payload, which includes the digital fingerprint, destruction execution command code, and node logs. The protocol version number of the current consortium blockchain network required for encapsulation is identified as version 4, the proof-of-work difficulty target parameter required by the consensus mechanism is set to a numerical scale prefixed with 10 consecutive zero bits, and the header 32-byte hash string of the previous block that has already passed network-wide consensus verification is retrieved. The aforementioned block transaction data payload is divided into multiple underlying transaction fragments. Following the node organization logic of a standard Merkle tree, the hash values ​​of every two adjacent transaction fragments are concatenated and then hashed again. This process is recursively repeated upwards until a unique Merkle root hash value representing the entire transaction data payload is generated. The version number, previous block hash, Merkle root hash, difficulty target, and timestamp are combined and sorted to construct an 80-byte block header metadata set.

[0154] The system acquires the block transaction data payload containing digital fingerprints and an 80-byte block header metadata set, and performs byte-level concatenation and fusion operations through the underlying application programming interface. A multinomial generation algorithm is used to generate a standard 32-bit network transport layer cross-node cyclic redundancy check (CRC) code for the concatenated byte stream. The generated 4-byte checksum is appended to the end of the block data. The data segment is locked using a digital signature and checksum, ensuring its integrity within the immutable distributed ledger storage structure, and formally encapsulated to form a final block. This final block is then broadcast peer-to-peer to all nodes within the local area network that have consensus accounting permissions, awaiting consensus verification from the remaining 20 nodes.

[0155] Retrieve the encapsulated and broadcast-waiting final block, extract its 80-byte block header metadata, and calculate the 256-bit hash value of the block header as the identity identifier of the final block. Retrieve the last updated node pointer position maintained in the in-memory database by the full lifecycle tracing chain initialized in step S1, and set the parent pointer field of the final block to the hash identifier of the current end node on the chain, thereby constructing a hash pointer pointing to the preceding tracing node. Through a chained encrypted reference method, strictly insert the final block into the very end of the full lifecycle tracing chain, and simultaneously update the end node mapping relationship in the global state tree, completely completing the closed-loop record of the entire data chain from initial creation, several state transitions, to its complete destruction. Synchronously write the packaged data, including encrypted proofs and node consensus states, into the read-only persistent storage on disk, generating an on-chain solidified credential that cannot be tampered with by a single entity.

[0156] The system extracts the on-chain solidified credential number and timestamp information successfully recorded at the blockchain's underlying layer. It then sends a Structured Query Language (SCL) command to the backend cloud-based relational database management system, submitting a logical delete command carrying a forced cascading delete execution flag. Upon receiving this command, the database engine scans the system-level registry and the underlying file system directory tree allocation table, matching all metadata entries bound to the physically destroyed file. The matched index rows are marked as obsolete in the database disk file for physical space reclamation, thoroughly clearing the associated file index metadata from the system registry and directory tree. The system calls the key management microservice interface to query the 256-bit Advanced Encryption Standard (AES) access control key bound to the file's globally unique identifier and all temporarily distributed application authorization session key caches. For the memory address space where these keys are stored, the system directly overwrites and writes them to a zero-byte array, simultaneously revoking and completely destroying all associated temporary and long-term access control keys. The destruction request command content and execution environment parameter information initially captured by the front-end system are integrated with the on-chain credential number and processing time in milliseconds fixed in the back-end ledger and encapsulated into a unified data format object. A report rendering engine then outputs a destruction report in portable document format, completing the entire data tracing and cleanup loop. Experimental data shows that in processing a large cluster containing 100,000 index records, due to the automated closed-loop operation of physical overwriting, metadata cleanup, and key zeroing, the total time for logical cleanup and on-chain consensus is controlled within 2.5 seconds, and the system's concurrent processing throughput is nearly doubled.

[0157] A cloud-based file destruction and data traceability system is provided, which is used to execute the aforementioned cloud-based file destruction and data traceability method. The system includes:

[0158] The file location and parsing module is used to receive cloud files to be destroyed, extract file metadata, build a full lifecycle traceability chain, use a distributed hash table algorithm to locate the cloud files to be destroyed in the cloud storage cluster, obtain the cloud storage logical address, and resolve the cloud storage logical address into a set of physical storage units through the underlying driver mapping table.

[0159] The state overwrite acquisition module is used to respond to file destruction commands, lock the physical storage unit set, generate overwrite operations, drive the magnetic head or voltage controller to perform forced write actions on the physical storage unit set, acquire electromagnetic flip signals of the underlying storage medium, and construct a storage unit state flip matrix.

[0160] The feature extraction and determination module is used to perform discrete cosine transform on the storage unit state flip matrix, extract high-frequency noise features, calculate Shannon entropy value, obtain noise baseline, construct secure destruction baseline, compare Shannon entropy value with secure destruction baseline, generate cryptographic destruction proof if Shannon entropy value is greater than secure destruction baseline, otherwise trigger secondary deep cleaning logic.

[0161] The closed-loop destruction and cleanup module is used to perform digital signature processing on cryptographic destruction proofs using asymmetric encryption algorithms, encapsulate the final block, link the final block to the end node of the full lifecycle traceability chain through a hash pointer, execute logical deletion instructions, clear file index metadata and access control keys, and output a destruction report.

[0162] The above embodiments illustrate preferred embodiments of the present invention. Any equivalent adjustments to the technical solution based on software engineering methods are within the scope of protection, including but not limited to: implementing algorithm logic using different programming languages, refactoring functional modules into services, adjusting data interaction protocols, and optimizing resource scheduling strategies. Any implementation scheme derived from reasonable modifications to the data processing flow, service call chain, or system architecture layer without departing from the core technology of the present invention should be considered within the protection scope defined by the technical solution of the present invention.

Claims

1. A method for cloud-based file destruction and data traceability, characterized in that, Includes the following steps: S1: Receive cloud archives to be destroyed, extract archive metadata, construct a full lifecycle traceability chain, use a distributed hash table algorithm to locate the cloud archives to be destroyed in the cloud storage cluster, obtain the cloud storage logical address, and resolve the cloud storage logical address into a set of physical storage units through the underlying driver mapping table. S2: In response to the file destruction command, lock the physical storage unit set, generate an overwrite operation, drive the magnetic head or voltage controller to perform a forced write action on the physical storage unit set, collect the electromagnetic flip signal of the underlying storage medium, and construct a storage unit state flip matrix. S3: Perform discrete cosine transform on the state flip matrix of the storage unit, extract high-frequency noise features, calculate Shannon entropy value, obtain noise benchmark, construct secure destruction baseline, compare Shannon entropy value with secure destruction baseline, if Shannon entropy value is greater than secure destruction baseline, generate cryptographic destruction proof, otherwise trigger secondary deep cleaning logic. S4: Use an asymmetric encryption algorithm to perform digital signature processing on the cryptographic destruction proof, encapsulate the final block, link the final block to the end node of the full lifecycle traceability chain through a hash pointer, execute a logical deletion instruction, clear the file index metadata and access control key, and output a destruction report.

2. The cloud-based file destruction and data traceability method according to claim 1, characterized in that, The specific steps of S1 are as follows: S11: Receive the cloud archive to be destroyed, parse the standard structure of the file header and the custom storage attribute tag, extract the archive metadata, verify the signature, initialize the basic genesis node of the full life cycle traceability chain according to the standard timestamp sequence and operation type feature field, and establish the metadata index topology structure. The archive metadata includes the initial creation time, historical modification frequency, owner identification sequence, and corresponding security control level; S12: Obtain the globally unique resource locator and shard hash sequence extracted from the metadata index topology, use them as hash digest input parameters and substitute them into the preset distributed hash table algorithm to perform consistent hash ring space routing calculation, traverse the virtual node routing mapping table clockwise to match the nearest available storage node, locate shard information, and aggregate to generate a logical address distribution set. The logical address distribution set includes the cloud storage logical addresses of the cloud files to be destroyed distributed in a cloud storage cluster with a multi-availability zone architecture; S13: Obtain the set of logical address distributions, call the virtual file system interface of the storage layer, query the underlying driver mapping table maintained in the storage controller, convert the three-dimensional logical block addresses one by one into physical layer absolute addressing parameters containing cylinder number, head number and sector number, resolve the cloud storage logical address into the real hardware storage space location, and aggregate to form the set of physical storage units. The physical storage unit set encompasses all data shards.

3. The cloud-based file destruction and data traceability method according to claim 1, characterized in that, The specific steps of S2 are as follows: S21: In response to the file destruction command, send the highest priority hardware latch signal to the input / output scheduling queue of the storage layer to lock the physical storage unit set, block all concurrent read and write requests, call the chaotic dynamic system equation to generate a multi-dimensional overwrite data stream with nonlinear pseudo-random characteristics, and construct a working instruction set; The set of working instructions serves the overwrite operation; S22: Obtain the working instruction set, send the underlying erase / write control message to the hard disk microcontroller or flash memory conversion layer through the storage bus, drive the mechanical disk head to perform random magnetization flip or control the solid-state drive voltage controller to apply erase voltage, perform a forced write action on the physical storage unit set in the locked state, and obtain the physical overwrite confirmation state. S23: Obtain the physical overwrite confirmation status, synchronously activate the high-frequency digital-to-analog converter bypass monitoring circuit deployed on the storage node motherboard, collect the microscopic electromagnetic reversal signal generated by the underlying storage medium during the overwrite process of the physical storage unit set, quantize and align the continuously sampled analog signal features with the spatiotemporal dimension, and establish the storage unit state reversal matrix.

4. The cloud-based file destruction and data traceability method according to claim 1, characterized in that, The specific steps of S3 are as follows: S31: Obtain the state flip matrix of the storage unit, map the discrete time-domain simulated sampling values ​​to a two-dimensional spatial array, perform two-dimensional discrete cosine transform on the pixel-level features of each row and column in the matrix, convert the data from the spatial domain to the frequency domain, separate the low-frequency basic background energy distribution, extract high-frequency noise features, and establish a frequency domain feature spectrum model. The high-frequency noise characteristics are determined by the disordered abrupt changes in the physical state of the medium; S32: Obtain the frequency domain feature spectrum model, use the probability density function to statistically analyze the energy distribution law and random fluctuation range of the high-frequency noise features in different frequency bands, calculate the Shannon entropy value of the current state system according to the uncertainty measurement formula in information theory, obtain the environmental noise benchmark by combining the noise-free spectrum parameters of the factory default storage medium, and construct the safe destruction baseline according to the threshold of the safety standard white paper. S33: Obtain the Shannon entropy value and the secure destruction baseline, execute the numerical comparison logic based on floating-point precision. If the Shannon entropy value is greater than the secure destruction baseline, it proves that the original data has been completely covered by the physical chaotic state. Based on this result, call the elliptic curve digital signature algorithm to generate the cryptographic destruction proof. Otherwise, send an abnormal alarm signal to the storage controller and trigger the secondary deep cleaning logic. The cryptographic destruction proof includes the device fingerprint; The secondary deep cleaning logic includes alternating hot writing and cold power-on operations.

5. The cloud-based file destruction and data traceability method according to claim 1, characterized in that, The specific steps of S4 are as follows: S41: Obtain the cryptographic destruction proof, call the pre-distributed asymmetric encryption system private key parameters, use the asymmetric encryption algorithm to perform digital signature processing on the hash digest of the proof content, package and aggregate the signed proof file, the network protocol address of the destruction execution node and the timestamp, perform encapsulation operation according to the underlying data structure standard of the blockchain ledger, and generate the final block; The terminator block has a fixed storage capacity; S42: Obtain the final block, calculate the secure hash algorithm digest value of the block header, construct a hash pointer pointing to the previous traceability node, link the final block to the end node of the full life cycle traceability chain in a chain-like encrypted reference manner, complete the full-link data closed-loop record of the file from creation to destruction, and generate on-chain solidified credentials. S43: Obtain the on-chain solidified credential, submit a logical deletion instruction containing a forced execution flag to the cloud database management system, clear the file index metadata in the system registry and directory tree, simultaneously revoke and destroy all associated temporary and long-term access control keys, integrate the front-end destruction instruction and the back-end on-chain credential, and output a destruction report. The destruction report contains a complete operation log.

6. The cloud-based file destruction and data traceability method according to claim 2, characterized in that, The execution process of the consistent hashing ring space routing calculation specifically includes: Obtain the globally unique resource locator and the shard hash sequence, extract the preset hash function algorithm feature parameters, convert the locator string into a fixed-length binary data stream, substitute it into the hash function for bit operations and non-linear obfuscation processing, map it to a logical ring numerical space composed of 2^32 virtual slots, and obtain the initial resource mapping hash point. Obtain the initial resource mapping hash point, read the storage node configuration information table, perform multiple hash calculations on the network socket address and hardware serial number of all available physical nodes according to the same hash function, generate virtual node mapping coordinates, synchronously project the coordinates into the same logical ring numerical space, and establish a node spatial distribution topology. The virtual node mapping coordinates are multiple dispersed coordinates generated for a single physical node; Obtain the node spatial distribution topology, take the initial resource mapping hash point as the starting cursor, perform a one-way scan traversal along the positive clockwise trajectory of the logical ring numerical space with increasing numerical values, identify and lock the first hit virtual node mapping coordinates, confirm its corresponding real physical storage node by reverse lookup of the routing table, and establish a routing binding relationship. The routing binding relationship refers to the direct mapping between resource locators and actual physical hardware addresses.

7. The cloud-based file destruction and data traceability method according to claim 3, characterized in that, The construction process of the working instruction set specifically includes: The spatial capacity and current operating temperature parameters of the physical storage unit set are obtained, the initial fractal variables and time evolution step size of the chaotic dynamic equation are set, and a nonlinear calculation model coupled with the logistic mapping function and the Lorentz attractor is adopted to perform iterative calculation, generate a floating-point number sequence, and establish an initial chaotic data source. The floating-point sequence has multidimensional pseudo-random characteristics and high initial value sensitivity; The initial chaotic data source is obtained, and double-precision bitwise operations are performed to truncate each floating-point value in the sequence. The unordered bit stream of the mantissa is extracted, and the data is reassembled and packaged according to the page erase alignment rules of the storage medium. The weight parameters of the multiple overwrite instructions are calculated. The calculation formula is: ； in, Represents the weight parameter of multiple overwrite instructions. Represents the temperature compensation coefficient. This represents the spatial capacity of the physical storage unit set. Represents the baseline space capacity. This represents the current operating temperature parameter. Represents the reference operating temperature parameter. Represents the time decay factor. Represents the number of iterations in the chaotic sequence, generating a standardized overwrite data packet; Obtain the standardized overwrite data packet and the weight parameters of the multiple overwrite instructions, dynamically adjust the sending frequency and concurrent writing depth of different data packets in the underlying device queue according to the weight parameters, combine them to form an overwrite strategy, and output the working instruction set according to the strategy. The coverage strategy is a multiple alternation scheme targeting different storage sectors; The set of working instructions has spatiotemporal disorder characteristics.

8. The cloud-based file destruction and data traceability method according to claim 4, characterized in that, The calculation process for the Shannon entropy specifically includes: The frequency domain characteristic spectrum model is obtained, and the entire high-frequency analysis bandwidth is divided into a continuous and non-overlapping sub-band sequence according to the preset discrete frequency band division rules. The instantaneous power spectral density values ​​in each sub-band are extracted using a bandpass filter bank, and the absolute energy integral is calculated. The absolute energy integral corresponds to each of the sub-frequency bands; The absolute energy integral is obtained, the total global radiation energy within the high-frequency analysis bandwidth is statistically analyzed, the ratio of the absolute energy integral of a single sub-band to the total global radiation energy is calculated, and after normalization, a discrete probability distribution vector is obtained to establish a system microstate probability distribution model. The discrete probability distribution vector reflects the probability distribution pattern of noise features occurring in each sub-band. Obtain the discrete probability distribution vector in the probability distribution model of the system's microstate. For each probability element in the vector, calculate its product with its base-2 logarithm. Then, sum the product results of all dimensions and take the negative value to obtain the Shannon entropy value. The Shannon entropy value is determined by the absolute degree of disorder in the physical magnetic state of the underlying storage medium.

9. The cloud-based file destruction and data traceability method according to claim 5, characterized in that, The encapsulation process of the terminator block specifically includes: Obtain the signed proof document, call the secure hash algorithm to perform a secondary mixed hash operation on the main content of the document and the signature-related information to generate a transaction certificate hash value, use it as the digital fingerprint of the signed proof document, and establish the block transaction data payload; The hash value of the transaction certificate is a globally unique identifier with a fixed length of 256 bits. Obtain the network protocol address and timestamp of the destruction execution node, and combine it with the protocol version number of the current blockchain network, the proof-of-work parameters of the consensus mechanism, and the header hash of the previous verified block, and construct the block header metadata set according to the node organization logic of the standard Merkle tree. The transaction data payload of the block and the metadata set of the block header are obtained. The two are concatenated and encoded at the byte level through the underlying serialization interface. Cross-node verification cyclic redundancy code of the network transport layer is applied to lock all data segments in an immutable distributed ledger storage structure and encapsulate them to form a final block. The final block is broadcast to all nodes in the network for consensus verification.

10. A cloud-based archive destruction and data traceability system, characterized in that, The system is used to implement the cloud-based file destruction and data traceability method according to any one of claims 1-9, the system comprising: The archive location and parsing module is used to receive cloud archives to be destroyed, extract archive metadata, build a full life cycle traceability chain, use a distributed hash table algorithm to locate the cloud archives to be destroyed in the cloud storage cluster, obtain the cloud storage logical address, and resolve the cloud storage logical address into a set of physical storage units through the underlying driver mapping table. The state overwrite acquisition module is used to respond to the file destruction command, lock the physical storage unit set, generate an overwrite operation, drive the magnetic head or voltage controller to perform a forced write action on the physical storage unit set, acquire the electromagnetic flip signal of the underlying storage medium, and construct the storage unit state flip matrix. The feature extraction and determination module is used to perform discrete cosine transform on the state flip matrix of the storage unit, extract high-frequency noise features, calculate Shannon entropy value, obtain noise benchmark, construct secure destruction baseline, compare Shannon entropy value with secure destruction baseline, and generate cryptographic destruction proof if Shannon entropy value is greater than secure destruction baseline; otherwise, trigger secondary deep cleaning logic. The closed-loop destruction and cleanup module is used to perform digital signature processing on the cryptographic destruction proof using an asymmetric encryption algorithm, encapsulate the final block, link the final block to the end node of the full lifecycle traceability chain through a hash pointer, execute logical deletion instructions, clear the file index metadata and access control keys, and output a destruction report.

Images