A core digital asset theft prevention and dynamic tracing system and method based on confidential computing and business strong binding

Abstract

This invention discloses a core digital asset anti-theft and dynamic traceability system and method based on confidential computing and strong business binding, belonging to the fields of information security and digital rights protection technology. It is used to prevent high-privilege internal personnel from stealing massive amounts of digital assets and supports extremely fast dynamic traceability. The system includes a client, a business hub, an encrypted storage layer, and a TEE confidential computing node. The method includes generating a key, encrypting the data to obtain ciphertext, shipping the goods according to the delivery request, decrypting the data through the TEE confidential computing node, and returning the data to the client. This invention achieves storage layer encryption through envelope encryption, controls key release through a two-factor verification of payment vouchers and TEE remote proof, achieves millisecond-level blind watermark embedding through a combination of offline frequency domain pre-computation and online slot injection, and achieves zero key residue through memory shredding, thus realizing the physical separation of technical management rights and data reading rights.

Description

Technical Field

[0001] This invention discloses a core digital asset anti-theft and dynamic traceability system and method based on confidential computing and strong business binding, belonging to the field of information security and digital copyright protection technology. Background Technology

[0002] In the fields of satellite remote sensing, military industry, and high-end commercial data transactions, massive amounts of high-precision images are often stored directly in enterprise object storage systems (such as MinIO). Existing data theft prevention systems primarily guard against external attacks, but face a serious risk of "insider theft" by privileged internal users (such as CTOs and system administrators). System administrators with root privileges on the underlying server are not bound by application-layer access control (ACLs): existing ACLs are ineffective against system administrators with root privileges on the underlying server. Administrators can directly copy storage disks or steal core encryption keys or plaintext data through memory dumps when applications decrypt data.

[0003] The bottleneck of real-time traceability for large-volume digital assets: To prevent internal personnel from taking screenshots or leaking data piecemeal, tamper-proof frequency domain blind watermarks (such as DWT / DCT distributed wavelet transform) need to be injected during data distribution. However, remote sensing images are huge (often reaching GB level). Performing full frequency domain transformation and watermark injection in real time when a user requests to download the data would lead to severe memory overflow (OOM) and unacceptable response latency, seriously hindering the commercial flow of core data. Summary of the Invention

[0004] The purpose of this invention is to provide a core digital asset anti-theft and dynamic traceability system and method based on confidential computing and strong business binding, in order to solve the problems in the prior art, such as high-privilege operation and maintenance personnel being able to physically steal plaintext data, data distribution authorization being disconnected from finance leading to forged requests, and serious performance bottlenecks in real-time frequency domain watermark injection of large-volume digital assets.

[0005] A core digital asset anti-theft and dynamic traceability system based on confidential computing and strong business binding includes a client request module, a business central server, an encrypted storage layer, a TEE confidential computing node, and a high-privilege area; The client request module includes a request construction unit and a data receiving buffer; The business central server includes an access gateway submodule, an order verification module, a routing scheduling unit, a tiled gateway submodule, and a financial reconciliation program; The encrypted storage layer includes an object storage cluster and a key management system; the key management system includes a root master key storage module, an envelope encryption / decryption engine, and a remote authentication server. The TEE confidential computing node includes a normal execution environment and a protected trusted memory region; the protected trusted memory region includes a local proof agent process, a session key negotiation process, a decryption execution engine, a frequency domain watermarking engine, and a memory shredding daemon process; the frequency domain watermarking engine includes a slot parsing subunit and an inverse transform operation subunit; High-privilege zones refer to the underlying server root privilege domains possessed by system administrators, database administrators, and storage maintenance personnel.

[0006] The client establishes a two-way communication connection with the business hub via the HTTPS / TLS protocol; The business hub establishes bidirectional communication connections with the third-party payment gateway (order status query), object storage cluster (reading encrypted data and envelopes), key management system (triggering key release), and TEE confidential computing node (distributing tasks and receiving data) through the internal network. TEE confidential computing nodes establish data read connections with the object storage cluster via the S3 protocol and establish key release connections with the key management system via remote proof protocol and encrypted session channels. The memory shredding daemon directly manipulates the physical memory address bus through CPU privileged instructions, without relying on operating system kernel services.

[0007] A method for preventing theft and dynamically tracing core digital assets based on confidential computing and strong business integration, comprising: S1. Generate a one-time data key, perform discrete wavelet transform on the original asset and reserve watermark payload slots to generate intermediate image data, use the data key to perform authentication encryption on the intermediate image data to obtain ciphertext data and store it in the object storage cluster, call the key management system to use the root master key to perform asymmetric encryption on the data key to generate ciphertext envelopes and store them in the envelope storage bucket of the object storage cluster. S2. The business hub receives the client's pickup request. The order verification module verifies the signature of the payment receipt with the third-party payment gateway. The routing and scheduling unit routes the encrypted data and encrypted envelope to the TEE confidential computing node. The key management system initiates remote authentication to the TEE confidential computing node to verify the platform's identity and code metric. After the verification is successful, the key management system unpacks the envelope and transmits the plaintext data key to the protected trusted memory area of ​​the TEE confidential computing node through an encrypted session. S3 and TEE confidential computing nodes use plaintext data keys to decrypt ciphertext intermediate files within a protected trusted memory area, extract request context information to generate a traceability bit sequence, embed the traceability bit sequence in a reserved slot in the frequency domain, perform discrete wavelet inverse transform, and output spatial domain image data with dynamic blind watermark. S4. Stream the watermarked image data to the client. After delivery, the memory shredding daemon performs deterministic overwriting on the memory pages containing plaintext data keys and plaintext image data in the protected trusted memory area, and generates audit logs for financial reconciliation.

[0008] S1 includes S1.1, which calls a cryptographically secure random number generator to generate a one-time data key for the digital asset file to be added to the database. The key length is 256 bits, and an asset identifier is generated for the current asset. S1 includes S1.2, where the system performs discrete wavelet transform on the original image data, decomposes the image into multiple frequency domain sub-bands, and selects the high-frequency diagonal detail sub-band with the lowest visual sensitivity as the watermark embedding domain. In the current sub-band coefficient matrix, using the hash value of the asset identifier as a random seed, N coefficient positions are randomly selected uniformly, and their coordinate set is recorded. For each selected coefficient, a quantization index modulation strategy is used to adjust it to the preset quantization grid center point to reserve watermark load slots; The system encapsulates the coordinate set, quantization step size, random seed, and embedded subband identifier into a metadata header, which together with the transformed coefficients constitutes the intermediate image data.

[0009] S1 includes S1.3, using the data key to perform authentication encryption operation on the intermediate image data and its metadata header to obtain ciphertext data, which is stored in the ciphertext data storage bucket of the object storage cluster, and the storage path is associated with the asset identifier record; S1 includes S1.4, the envelope encryption interface for the business hub to call the key management system; The key management system's envelope encryption and decryption engine loads the root master key from the hardware security module, and uses the master key's public key to perform asymmetric encryption on the data key and additional authentication data to generate a ciphertext envelope; The key management system writes the encrypted envelopes into the envelope storage bucket of the object storage cluster and establishes a one-to-one correspondence index with the asset identifier.

[0010] S2 includes S2.1, where the client holds the paid order number and sends a pickup request message carrying the order number, asset identifier, and user identity certificate to the business center via the HTTPS protocol; S2 includes S2.2, where the order verification module of the business hub initiates a synchronous query of the order status to the third-party payment gateway based on the order number, and obtains a payment receipt containing the payment gateway's private key signature; The order verification module uses a pre-installed payment gateway public key certificate to verify the signature's validity and check the consistency between the order amount and the asset's listed price. Once verification is successful, an internal payment voucher will be generated. S2 includes S2.3, where the routing and scheduling unit queries and obtains the access path and encrypted envelope data of the encrypted intermediate file based on the asset identifier, combines the payment voucher and user identity information to encapsulate it into an internal task descriptor, selects the target TEE confidential computing node based on the load balancing algorithm, and pushes it through the internal remote procedure call protocol.

[0011] S2 includes S2.4, after the TEE confidential computing node receives the task, a two-way remote proof is initiated between the local proof agent and the remote proof server of the key management system; The key management system generates random challenges and sends them to the TEE confidential computing node. The TEE confidential computing node calls CPU hardware instructions to generate a hardware signature report containing platform identity information and code metric values ​​in the trusted memory area. The certificate chain of the remote verification server's verification report signature can be traced back to the hardware manufacturer's root certificate, confirming that the platform runs on genuine TEE hardware, and comparing the code metric with the system's preset whitelist; Once verification is successful, a disassembly license is granted to the envelope encryption / decryption engine. S2 includes S2.5. After the remote verification is passed, the envelope encryption and decryption engine calls the hardware security module to perform asymmetric decryption on the ciphertext envelope using the master control key private key to recover the plaintext data key. The key management system transmits the plaintext data key to the TEE confidential computing node through an encrypted session channel established based on TEE local proof. The TEE confidential computing node then decrypts the key and stores it in a protected trusted memory area.

[0012] S3 includes S3.1, where the decryption execution engine of the TEE confidential computing node performs authentication and decryption on the ciphertext intermediate file using the received plaintext data key in a protected trusted memory area, restoring the intermediate image data and its metadata header. S3 includes S3.2, the slot parsing subunit extracts the order number, user identity identifier, request source IP address, operation timestamp and asset identifier from the task descriptor, concatenates them in a fixed order, and generates a traceability bit sequence using a hash message authentication code algorithm with a key. The key is a temporary key generated in the protected memory area when the TEE confidential computing node starts. S3 includes S3.3, where the frequency domain watermarking engine extracts the slot coordinate set and quantization step size parameter from the metadata header, modifies each selected frequency domain coefficient using quantization index modulation based on the traceability bit value, and embeds the bit into the reserved slot.

[0013] S3 includes S3.4. After all coefficients are modified, the inverse transform operation subunit performs discrete wavelet inverse transform to reconstruct spatial image data with full-amplitude dynamic frequency domain blind watermark from the modified sub-band coefficients. S3 includes S3.5, where when order verification fails and the request is marked as preview, the business hub will route the request to the tiled gateway submodule; The tiled gateway submodule reads the corresponding low-resolution tiles from the object storage cluster based on the viewport parameters, embeds a lightweight implicit watermark containing the user's identity in the least important pixel, and outputs it after overlaying a visible copyright notice.

[0014] S4 includes S4.1, the frequency domain watermarking engine of the TEE confidential computing node transmits watermarked spatial image data back to the business center in blocks through an encrypted channel, and the business center synchronously forwards it to the client in a streaming manner. A chunked transmission mode is adopted to avoid generating plaintext temporary files on the persistent media of the TEE confidential computing node; S4 includes S4.2, at the moment when the data stream output is completed or abnormally interrupted, the memory shredding daemon process obtains the physical address and length of the memory region storing the plaintext data key, the decrypted intermediate frequency domain data and the reconstructed spatial domain image data, and the secure memory overwrite execution unit directly operates the physical memory to perform the overwrite in the triple overwrite mode. After completion, the processor memory barrier instruction and the compiler optimization barrier instruction are executed to prevent the overwrite operation from being optimized by out-of-order reordering. S4 includes S4.3, whereby the business hub generates a structured audit log after successful asset delivery, containing asset identifier, order number, user identity identifier, delivery timestamp, and TEE remote proof summary, which is synchronously stored in the tamper-proof audit database. The financial reconciliation process periodically compares the order numbers in the audit logs with the settlement payment records of the third-party payment gateway. If a delivery is found to have no corresponding payment record, a security alarm is triggered and the relevant asset permissions are locked.

[0015] Compared to existing technologies, this invention offers the following advantages: It combines envelope encryption with TEE confidential computation, encapsulating the core decryption action within a hardware-level encrypted CPU-trusted execution environment. Even if a system administrator removes memory modules or performs memory packet sniffing, they only obtain hardware-level garbled data, achieving a physical separation of technical management rights and data access rights. It deeply binds the business pipeline to KMS decryption authorization, preventing internal R&D or IT personnel from obtaining data without authorization by writing backdoors or bypassing normal business processes. It overcomes the performance bottleneck of real-time traceability for large-volume assets: the innovative "offline pre-computation + online slot injection" dynamic-static separation mechanism compresses the frequency domain watermarking process, which originally took several minutes, to milliseconds, perfectly supporting high-concurrency online commercial monetization of massive amounts of remote sensing images. Attached Figure Description

[0016] Figure 1 This is a system architecture diagram of the present invention; Figure 2 It is a timing diagram of key interaction and memory shredding based on envelope encryption (DEK / CMK) and TEE confidential computing nodes; Figure 3 It is an algorithm flow diagram of offline pre-computation frequency domain transformation and online ultra-fast watermark injection for large-volume images. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention are described clearly and completely below. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0018] A core digital asset anti-theft and dynamic traceability system based on confidential computing and strong business binding includes a client request module, a business central server, an encrypted storage layer, a TEE confidential computing node, and a high-privilege area; The client request module includes a request construction unit and a data receiving buffer; The business central server includes an access gateway submodule, an order verification module, a routing scheduling unit, a tiled gateway submodule, and a financial reconciliation program; The encrypted storage layer includes an object storage cluster and a key management system; the key management system includes a root master key storage module, an envelope encryption / decryption engine, and a remote authentication server. The TEE confidential computing node includes a normal execution environment and a protected trusted memory region; the protected trusted memory region includes a local proof agent process, a session key negotiation process, a decryption execution engine, a frequency domain watermarking engine, and a memory shredding daemon process; the frequency domain watermarking engine includes a slot parsing subunit and an inverse transform operation subunit; High-privilege zones refer to the underlying server root privilege domains possessed by system administrators, database administrators, and storage maintenance personnel.

[0019] The client establishes a two-way communication connection with the business hub via the HTTPS / TLS protocol; The business hub establishes bidirectional communication connections with the third-party payment gateway (order status query), object storage cluster (reading encrypted data and envelopes), key management system (triggering key release), and TEE confidential computing node (distributing tasks and receiving data) through the internal network. TEE confidential computing nodes establish data read connections with the object storage cluster via the S3 protocol and establish key release connections with the key management system via remote proof protocol and encrypted session channels. The memory shredding daemon directly manipulates the physical memory address bus through CPU privileged instructions, without relying on operating system kernel services.

[0020] A method for preventing theft and dynamically tracing core digital assets based on confidential computing and strong business integration, comprising: S1. Generate a one-time data key, perform discrete wavelet transform on the original asset and reserve watermark payload slots to generate intermediate image data, use the data key to perform authentication encryption on the intermediate image data to obtain ciphertext data and store it in the object storage cluster, call the key management system to use the root master key to perform asymmetric encryption on the data key to generate ciphertext envelopes and store them in the envelope storage bucket of the object storage cluster. S2. The business hub receives the client's pickup request. The order verification module verifies the signature of the payment receipt with the third-party payment gateway. The routing and scheduling unit routes the encrypted data and encrypted envelope to the TEE confidential computing node. The key management system initiates remote authentication to the TEE confidential computing node to verify the platform's identity and code metric. After the verification is successful, the key management system unpacks the envelope and transmits the plaintext data key to the protected trusted memory area of ​​the TEE confidential computing node through an encrypted session. S3 and TEE confidential computing nodes use plaintext data keys to decrypt ciphertext intermediate files within a protected trusted memory area, extract request context information to generate a traceability bit sequence, embed the traceability bit sequence in a reserved slot in the frequency domain, perform discrete wavelet inverse transform, and output spatial domain image data with dynamic blind watermark. S4. Stream the watermarked image data to the client. After delivery, the memory shredding daemon performs deterministic overwriting on the memory pages containing plaintext data keys and plaintext image data in the protected trusted memory area, and generates audit logs for financial reconciliation.

[0021] S1 includes S1.1, which calls a cryptographically secure random number generator to generate a one-time data key for the digital asset file to be added to the database. The key length is 256 bits, and an asset identifier is generated for the current asset. S1 includes S1.2, where the system performs discrete wavelet transform on the original image data, decomposes the image into multiple frequency domain sub-bands, and selects the high-frequency diagonal detail sub-band with the lowest visual sensitivity as the watermark embedding domain. In the current sub-band coefficient matrix, using the hash value of the asset identifier as a random seed, N coefficient positions are randomly selected uniformly, and their coordinate set is recorded. For each selected coefficient, a quantization index modulation strategy is used to adjust it to the preset quantization grid center point to reserve watermark load slots; The system encapsulates the coordinate set, quantization step size, random seed, and embedded subband identifier into a metadata header, which together with the transformed coefficients constitutes the intermediate image data.

[0022] S1 includes S1.3, using the data key to perform authentication encryption operation on the intermediate image data and its metadata header to obtain ciphertext data, which is stored in the ciphertext data storage bucket of the object storage cluster, and the storage path is associated with the asset identifier record; S1 includes S1.4, the envelope encryption interface for the business hub to call the key management system; The key management system's envelope encryption and decryption engine loads the root master key from the hardware security module, and uses the master key's public key to perform asymmetric encryption on the data key and additional authentication data to generate a ciphertext envelope; The key management system writes the encrypted envelopes into the envelope storage bucket of the object storage cluster and establishes a one-to-one correspondence index with the asset identifier.

[0023] S2 includes S2.1, where the client holds the paid order number and sends a pickup request message carrying the order number, asset identifier, and user identity certificate to the business center via the HTTPS protocol; S2 includes S2.2, where the order verification module of the business hub initiates a synchronous query of the order status to the third-party payment gateway based on the order number, and obtains a payment receipt containing the payment gateway's private key signature; The order verification module uses a pre-installed payment gateway public key certificate to verify the signature's validity and check the consistency between the order amount and the asset's listed price. Once verification is successful, an internal payment voucher will be generated. S2 includes S2.3, where the routing and scheduling unit queries and obtains the access path and encrypted envelope data of the encrypted intermediate file based on the asset identifier, combines the payment voucher and user identity information to encapsulate it into an internal task descriptor, selects the target TEE confidential computing node based on the load balancing algorithm, and pushes it through the internal remote procedure call protocol.

[0024] S2 includes S2.4, after the TEE confidential computing node receives the task, a two-way remote proof is initiated between the local proof agent and the remote proof server of the key management system; The key management system generates random challenges and sends them to the TEE confidential computing node. The TEE confidential computing node calls CPU hardware instructions to generate a hardware signature report containing platform identity information and code metric values ​​in the trusted memory area. The certificate chain of the remote verification server's verification report signature can be traced back to the hardware manufacturer's root certificate, confirming that the platform runs on genuine TEE hardware, and comparing the code metric with the system's preset whitelist; Once verification is successful, a disassembly license is granted to the envelope encryption / decryption engine. S2 includes S2.5. After the remote verification is passed, the envelope encryption and decryption engine calls the hardware security module to perform asymmetric decryption on the ciphertext envelope using the master control key private key to recover the plaintext data key. The key management system transmits the plaintext data key to the TEE confidential computing node through an encrypted session channel established based on TEE local proof. The TEE confidential computing node then decrypts the key and stores it in a protected trusted memory area.

[0025] S3 includes S3.1, where the decryption execution engine of the TEE confidential computing node performs authentication and decryption on the ciphertext intermediate file using the received plaintext data key in a protected trusted memory area, restoring the intermediate image data and its metadata header. S3 includes S3.2, the slot parsing subunit extracts the order number, user identity identifier, request source IP address, operation timestamp and asset identifier from the task descriptor, concatenates them in a fixed order, and generates a traceability bit sequence using a hash message authentication code algorithm with a key. The key is a temporary key generated in the protected memory area when the TEE confidential computing node starts. S3 includes S3.3, where the frequency domain watermarking engine extracts the slot coordinate set and quantization step size parameter from the metadata header, modifies each selected frequency domain coefficient using quantization index modulation based on the traceability bit value, and embeds the bit into the reserved slot.

[0026] S3 includes S3.4. After all coefficients are modified, the inverse transform operation subunit performs discrete wavelet inverse transform to reconstruct spatial image data with full-amplitude dynamic frequency domain blind watermark from the modified sub-band coefficients. S3 includes S3.5, where when order verification fails and the request is marked as preview, the business hub will route the request to the tiled gateway submodule; The tiled gateway submodule reads the corresponding low-resolution tiles from the object storage cluster based on the viewport parameters, embeds a lightweight implicit watermark containing the user's identity in the least important pixel, and outputs it after overlaying a visible copyright notice.

[0027] S4 includes S4.1, the frequency domain watermarking engine of the TEE confidential computing node transmits watermarked spatial image data back to the business center in blocks through an encrypted channel, and the business center synchronously forwards it to the client in a streaming manner. A chunked transmission mode is adopted to avoid generating plaintext temporary files on the persistent media of the TEE confidential computing node; S4 includes S4.2, at the moment when the data stream output is completed or abnormally interrupted, the memory shredding daemon process obtains the physical address and length of the memory region storing the plaintext data key, the decrypted intermediate frequency domain data and the reconstructed spatial domain image data, and the secure memory overwrite execution unit directly operates the physical memory to perform the overwrite in the triple overwrite mode. After completion, the processor memory barrier instruction and the compiler optimization barrier instruction are executed to prevent the overwrite operation from being optimized by out-of-order reordering. S4 includes S4.3, whereby the business hub generates a structured audit log after successful asset delivery, containing asset identifier, order number, user identity identifier, delivery timestamp, and TEE remote proof summary, which is synchronously stored in the tamper-proof audit database. The financial reconciliation process periodically compares the order numbers in the audit logs with the settlement payment records of the third-party payment gateway. If a delivery is found to have no corresponding payment record, a security alarm is triggered and the relevant asset permissions are locked.

[0028] The following description, in conjunction with the accompanying drawings, further illustrates the system architecture of this invention. Figure 1 As shown, the system includes a client, a business hub (SpringBoot), an encrypted storage layer, a TEE confidential computing node (SGX / SEV), and a high-privilege area. The client includes paid requests (pickup / preview) and unpaid previews (tile-based degraded display). Paid requests are transmitted to the order verification stage in the business hub for payment receipt verification. The data is then output to the object storage MinIO ciphertext intermediate file in the encrypted storage layer, and then transmitted via pass-through receipt and ciphertext to the TEE secure memory of the TEE confidential computing node for decryption / watermark injection / memory shredding. Unpaid requests enter the third-party payment gateway (signature receipt) in the business hub, and then split into two branches: the first branch transmits the payment receipt to the order verification stage, and the second branch performs tile-based gateway (low resolution + LSB watermark). The encrypted storage layer's key management system CMK master key is transmitted to the RemoteAttestation (code hash verification) of the TEE confidential computing node via remote proof. The CMK decrypts the DEK→TEE and enters the TEE secure memory. High-privilege areas include system administrators (Root privileges). When performing physical copying of hard drives / memory packets, they can only obtain ciphertext / garbled text and cannot access the plaintext DEK.

[0029] This invention is based on the key exchange and memory shredding timing of envelope encryption (DEK / CMK) and TEE confidential computing nodes, as follows: Figure 2As shown, the business hub first verifies the payment receipt (order number + signature), then obtains the encrypted intermediate state and encrypted envelope, initiates proof into the KMS module, the second step initiates RemoteAttestation to verify the TEE code hash, the third step decrypts the encrypted envelope using CMK, then obtains the plaintext DEK, and inputs it into the TEEWorker (SGX / SEV); the TEEWorker returns a proof report to the second step and receives the output of the third step, performs a TLS encrypted tunnel, and transmits the plaintext DEK to the TEE memory, the fourth step injects the DEK decryption intermediate state file into the traceability sequence and generates a watermark, the fifth step outputs the data to the business hub, and after output, performs memset_s memory shredding (Zeroize). The object storage MinIO module pulls the encrypted intermediate state file to the fourth step.

[0030] The algorithm for offline pre-calculation of frequency domain transform and online ultra-fast watermark injection of large-volume images in this invention is as follows: Figure 3 As shown, the process includes an offline pre-computation stage (before asset storage, a time-consuming operation) and an online rapid injection stage (on request, millisecond-level). The offline pre-computation stage involves performing Discrete Wavelet Transform (DWT / DCT) frequency domain decomposition on the original remote sensing image (GB-level TIFF), pre-setting watermark payload slots (high-frequency / mid-frequency regions) in the frequency domain coefficient matrix, then performing DEK symmetric encryption to obtain a ciphertext intermediate file, performing object storage MinIO (ciphertext intermediate state persistence), generating an envelope, and using KMS to perform CMK encryption of the DEK to obtain a ciphertext envelope. The online rapid injection stage involves requesting the context (account / IP / timestamp / order number), performing hash calculation to obtain the traceability sequence, and then injecting it along with the retrieved ciphertext envelope into the TEE secure memory. The DEK decryption intermediate file is then performed, followed by traceability sequence injection to obtain frequency domain slots, performing Inverse Fast Transform (IDWT), outputting a data stream with a dynamic blind watermark, and then delivery. Finally, memset_s shreds the memory (DEK + plaintext are all zeroed out).

[0031] The following description, in conjunction with embodiments of the present invention, provides further details. Step one involves the encryption and pre-calculation of digital asset envelopes. After the remote sensing satellite image data is generated, the system generates a one-time random data key (DEK).

[0032] Perform Discrete Wavelet Transform (DWT) on the original image, reserve data slots in a specific high-frequency / mid-frequency coefficient matrix, and use the processed file as an intermediate file.

[0033] Use the DEK to perform symmetric encryption on the intermediate file and persist the ciphertext file to an object storage cluster (such as MinIO).

[0034] The system calls an external key management system (KMS), which uses the master key (CMK) controlled by the enterprise's top management to encrypt and encapsulate the DEK, generating a ciphertext envelope and binding it to the image asset.

[0035] Step two involves request verification and TEE authentication based on payment credentials. The client initiates a pickup request, and the business hub (such as a Spring Boot cluster) sends a status poll or callback verification to the third-party payment gateway to obtain a successful payment receipt containing the order number and cryptographic signature.

[0036] The business hub obtains the "ciphertext intermediate file" and the corresponding "ciphertext envelope" from the underlying storage layer, and routes them together with the payment receipt to the designated independent secure worker computing node through the internal high-speed communication bus.

[0037] Worker compute nodes run in a hardware-supported confidential computing environment (TEE). KMS initiates a Remote Attestation to the node to verify the integrity and hash consistency of its underlying operating system and watermarking program binary.

[0038] After successful verification, KMS uses CMK to decrypt the encrypted envelope within its own security boundary and sends the plaintext DEK to the encrypted memory area of ​​the TEE security node through the TLS encrypted tunnel.

[0039] Step three involves fast memory-level decryption and source watermark injection. The TEE compute node uses the received plaintext DEK in secure memory to perform decryption on the ciphertext intermediate file.

[0040] Extract the order context from the request (including the requesting account, IP, timestamp, and order number), and perform hash calculations to generate a tracing sequence.

[0041] The source sequence is rapidly injected into a pre-reserved frequency domain slot, and then an Inverse Fast Transform (IDWT) is performed to generate the final output data stream with a full-amplitude dynamic frequency domain blind watermark.

[0042] For requests from unpaid front-end users, the business hub intercepts and redirects them to the tiled gateway. The edge gateway extracts only the low-resolution tiles corresponding to the coordinates and overlays visible and extremely lightweight hidden watermarks in the spatial domain to complete the downgraded and leak-proof display.

[0043] Step four involves memory shredding and asset delivery. The TEE compute node will output the data stream with the watermark injected to the business hub or directly to the user terminal.

[0044] The moment the output operation completes or an interruption request is requested, the security daemon within the TEE node immediately invokes the underlying overwrite instruction to fill all memory stack addresses containing the plaintext DEK, plaintext image, and decryption intermediate state with zeros (ZeroizeMemory), completely destroying any trace of the operation. The business system records a complete export audit log and submits it to the finance department for comparison using the automated inflow and outflow reconciliation mechanism.

[0045] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A core digital asset anti-theft and dynamic traceability system based on confidential computing and strong business binding, characterized in that, This includes the client request module, the business hub server, the encrypted storage layer, the TEE confidential computing node, and the high-privilege area; The client request module includes a request construction unit and a data receiving buffer; The business central server includes an access gateway submodule, an order verification module, a routing scheduling unit, a tiled gateway submodule, and a financial reconciliation program; The encrypted storage layer includes an object storage cluster and a key management system; the key management system includes a root master key storage module, an envelope encryption / decryption engine, and a remote authentication server. The TEE confidential computing node includes a normal execution environment and a protected trusted memory region; the protected trusted memory region includes a local proof agent process, a session key negotiation process, a decryption execution engine, a frequency domain watermarking engine, and a memory shredding daemon process; the frequency domain watermarking engine includes a slot parsing subunit and an inverse transform operation subunit; High-privilege zones refer to the underlying server root privilege domains possessed by system administrators, database administrators, and storage maintenance personnel.

2. The core digital asset anti-theft and dynamic traceability system based on confidential computing and strong business binding as described in claim 1, characterized in that, The client establishes a two-way communication connection with the business hub via the HTTPS / TLS protocol; The business hub establishes bidirectional communication connections with the third-party payment gateway (order status query), object storage cluster (reading encrypted data and envelopes), key management system (triggering key release), and TEE confidential computing node (distributing tasks and receiving data) through the internal network. TEE confidential computing nodes establish data read connections with the object storage cluster via the S3 protocol and establish key release connections with the key management system via remote proof protocol and encrypted session channels. The memory shredding daemon directly manipulates the physical memory address bus through CPU privileged instructions, without relying on operating system kernel services.

3. A method for preventing theft and dynamically tracing core digital assets based on confidential computing and strong business integration, using the core digital asset anti-theft and dynamic traceability system based on confidential computing and strong business integration as described in claim 1, characterized in that... include: S1. Generate a one-time data key, perform discrete wavelet transform on the original asset and reserve watermark payload slots to generate intermediate image data, use the data key to perform authentication encryption on the intermediate image data to obtain ciphertext data and store it in the object storage cluster, call the key management system to use the root master key to perform asymmetric encryption on the data key to generate ciphertext envelopes and store them in the envelope storage bucket of the object storage cluster. S2. The business hub receives the client's pickup request. The order verification module verifies the signature of the payment receipt with the third-party payment gateway. The routing and scheduling unit routes the encrypted data and encrypted envelope to the TEE confidential computing node. The key management system initiates remote authentication to the TEE confidential computing node to verify the platform's identity and code metric. After the verification is successful, the key management system unpacks the envelope and transmits the plaintext data key to the protected trusted memory area of ​​the TEE confidential computing node through an encrypted session. S3 and TEE confidential computing nodes use plaintext data keys to decrypt ciphertext intermediate files within a protected trusted memory area, extract request context information to generate a traceability bit sequence, embed the traceability bit sequence in a reserved slot in the frequency domain, perform discrete wavelet inverse transform, and output spatial domain image data with dynamic blind watermark. S4. Stream the watermarked image data to the client. After delivery, the memory shredding daemon performs deterministic overwriting on the memory pages containing plaintext data keys and plaintext image data in the protected trusted memory area, and generates audit logs for financial reconciliation.

4. The method for preventing theft and dynamically tracing core digital assets based on confidential computing and strong business binding as described in claim 3, characterized in that, S1 includes S1.1, which calls a cryptographically secure random number generator to generate a one-time data key for the digital asset file to be added to the database. The key length is 256 bits, and an asset identifier is generated for the current asset. S1 includes S1.2, where the system performs discrete wavelet transform on the original image data, decomposes the image into multiple frequency domain sub-bands, and selects the high-frequency diagonal detail sub-band with the lowest visual sensitivity as the watermark embedding domain. In the current sub-band coefficient matrix, using the hash value of the asset identifier as a random seed, N coefficient positions are randomly selected uniformly, and their coordinate set is recorded. For each selected coefficient, a quantization index modulation strategy is used to adjust it to the preset quantization grid center point to reserve watermark load slots; The system encapsulates the coordinate set, quantization step size, random seed, and embedded subband identifier into a metadata header, which together with the transformed coefficients constitutes the intermediate image data.

5. The method for preventing theft and dynamically tracing core digital assets based on confidential computing and strong business binding as described in claim 4, characterized in that, S1 includes S1.3, using the data key to perform authentication encryption operation on the intermediate image data and its metadata header to obtain ciphertext data, which is stored in the ciphertext data storage bucket of the object storage cluster, and the storage path is associated with the asset identifier record; S1 includes S1.4, the envelope encryption interface for the business hub to call the key management system; The key management system's envelope encryption and decryption engine loads the root master key from the hardware security module, and uses the master key's public key to perform asymmetric encryption on the data key and additional authentication data to generate a ciphertext envelope; The key management system writes the encrypted envelopes into the envelope storage bucket of the object storage cluster and establishes a one-to-one correspondence index with the asset identifier.

6. The method for preventing theft and dynamically tracing core digital assets based on confidential computing and strong business binding as described in claim 5, characterized in that, S2 includes S2.1, where the client holds the paid order number and sends a pickup request message carrying the order number, asset identifier, and user identity certificate to the business center via the HTTPS protocol; S2 includes S2.2, where the order verification module of the business hub initiates a synchronous query of the order status to the third-party payment gateway based on the order number, and obtains a payment receipt containing the payment gateway's private key signature; The order verification module uses a pre-installed payment gateway public key certificate to verify the signature's validity and check the consistency between the order amount and the asset's listed price. Once verification is successful, an internal payment voucher will be generated. S2 includes S2.3, where the routing and scheduling unit queries and obtains the access path and encrypted envelope data of the encrypted intermediate file based on the asset identifier, combines the payment voucher and user identity information to encapsulate it into an internal task descriptor, selects the target TEE confidential computing node based on the load balancing algorithm, and pushes it through the internal remote procedure call protocol.

7. The method for preventing theft and dynamically tracing core digital assets based on confidential computing and strong business binding as described in claim 6, characterized in that, S2 includes S2.4, after the TEE confidential computing node receives the task, a two-way remote proof is initiated between the local proof agent and the remote proof server of the key management system; The key management system generates random challenges and sends them to the TEE confidential computing node. The TEE confidential computing node calls CPU hardware instructions to generate a hardware signature report containing platform identity information and code metric values ​​in the trusted memory area. The certificate chain of the remote verification server's verification report signature can be traced back to the hardware manufacturer's root certificate, confirming that the platform runs on genuine TEE hardware, and comparing the code metric with the system's preset whitelist; Once verification is successful, a disassembly permission is granted to the envelope encryption / decryption engine. S2 includes S2.

5. After the remote verification is passed, the envelope encryption and decryption engine calls the hardware security module to perform asymmetric decryption on the ciphertext envelope using the master control key private key to recover the plaintext data key; The key management system transmits the plaintext data key to the TEE confidential computing node through an encrypted session channel established based on TEE local proof. The TEE confidential computing node then decrypts the key and stores it in a protected trusted memory area.

8. The method for preventing theft and dynamically tracing core digital assets based on confidential computing and strong business binding as described in claim 7, characterized in that, S3 includes S3.1, where the decryption execution engine of the TEE confidential computing node performs authentication and decryption on the ciphertext intermediate file using the received plaintext data key in a protected trusted memory area, restoring the intermediate image data and its metadata header. S3 includes S3.2, the slot parsing subunit extracts the order number, user identity identifier, request source IP address, operation timestamp and asset identifier from the task descriptor, concatenates them in a fixed order, and generates a traceability bit sequence using a hash message authentication code algorithm with a key. The key is a temporary key generated in the protected memory area when the TEE confidential computing node starts. S3 includes S3.3, where the frequency domain watermarking engine extracts the slot coordinate set and quantization step size parameter from the metadata header, modifies each selected frequency domain coefficient using quantization index modulation based on the traceability bit value, and embeds the bit into the reserved slot.

9. A method for preventing theft and dynamically tracing core digital assets based on confidential computing and strong business binding, as described in claim 8, is characterized in that... S3 includes S3.

4. After all coefficients are modified, the inverse transform operation subunit performs discrete wavelet inverse transform to reconstruct spatial image data with full-amplitude dynamic frequency domain blind watermark from the modified sub-band coefficients. S3 includes S3.5, where when order verification fails and the request is marked as preview, the business hub will route the request to the tiled gateway submodule; The tiled gateway submodule reads the corresponding low-resolution tiles from the object storage cluster based on the viewport parameters, embeds a lightweight implicit watermark containing the user's identity in the least important pixel, and outputs it after overlaying a visible copyright notice.

10. A method for preventing theft and dynamically tracing core digital assets based on confidential computing and strong business binding, as described in claim 9, is characterized in that... S4 includes S4.1, the frequency domain watermarking engine of the TEE confidential computing node transmits watermarked spatial image data back to the business center in blocks through an encrypted channel, and the business center synchronously forwards it to the client in a streaming manner. A chunked transmission mode is adopted to avoid generating plaintext temporary files on the persistent media of the TEE confidential computing node; S4 includes S4.2, at the moment when the data stream output is completed or abnormally interrupted, the memory shredding daemon process obtains the physical address and length of the memory region storing the plaintext data key, the decrypted intermediate frequency domain data and the reconstructed spatial domain image data, and the secure memory overwrite execution unit directly operates the physical memory to perform the overwrite in the triple overwrite mode. After completion, the processor memory barrier instruction and the compiler optimization barrier instruction are executed to prevent the overwrite operation from being optimized by out-of-order reordering. S4 includes S4.3, whereby the business hub generates a structured audit log after successful asset delivery, containing asset identifier, order number, user identity identifier, delivery timestamp, and TEE remote proof summary, which is synchronously stored in the tamper-proof audit database. The financial reconciliation process periodically compares the order numbers in the audit logs with the settlement payment records of the third-party payment gateway. If a delivery is found to have no corresponding payment record, a security alarm is triggered and the relevant asset permissions are locked.

Images