A watermark-based universal multimedia interface protocol ownership detection method and device

Abstract

The application discloses a watermark-based general multimedia interface protocol right ownership detection method and device, and an interface holder of a general multimedia interface protocol holds to-be-verified interface configuration parameters and performs full-process right ownership detection and verification operation, the method comprises the following steps: extracting a joint watermark from the to-be-verified GPMI interface configuration parameters through secure multi-party computation; recovering eigenvalues of a joint identity matrix based on the extracted joint watermark; calculating a ciphertext Euclidean distance between the eigenvalues and a ciphertext feature vector provided by a verification party by using homomorphic encryption technology; obtaining the decrypted Euclidean distance through a double trapdoor homomorphic encryption outsourcing decryption mechanism, and determining the interface intellectual property right ownership according to the comparison result of the Euclidean distance and a preset threshold value. In the premise of not changing the interface function logic and transmission performance, the interface configuration characteristics and the ownership information are bound as a ciphertext watermark and embedded into the interface protocol module, so that the automatic declaration, verification and traceable right ownership of the interface implementation are realized.

Description

Technical Field

[0001] This invention relates to the fields of multimedia interface technology and digital ownership management, and in particular to a watermark-based method and apparatus for detecting ownership of a universal multimedia interface protocol. Background Technology

[0002] With the rapid development of information technology and multimedia applications, the capabilities for acquiring, transmitting, and displaying audio and video data are constantly improving, driving the continuous evolution of next-generation high-performance multimedia interface standards. In devices such as smart TVs, set-top boxes, monitors, and in-vehicle entertainment systems, the demand for high-bandwidth, low-latency transmission is constantly increasing, prompting interface protocol standards to continuously move towards integration and intelligence.

[0003] The General Purpose Multimedia Interface (GPMI), as a highly integrated and efficient audio and video interface standard, is gradually becoming a key channel for connection and data exchange between various terminal devices. Through systematic optimization of the physical and protocol layers, GPMI integrates ultra-high-definition video, multi-channel audio, control signals, and power management functions, forming a high-performance architecture of single-cable multi-channel aggregation. It supports features such as reversible insertion, hot-swapping, link adaptation, and fast wake-up, significantly improving interface flexibility and user experience, and providing system integrators with a lower development threshold and greater compatibility.

[0004] However, the high integration and openness of GPMI technology also bring significant intellectual property and implementation ownership risks. The protocol design and implementation process involves key technical details such as high-speed communication, electromagnetic compatibility, protocol parsing, hardware adaptation, and timing control. These achievements are often the result of significant resource investment by manufacturers or R&D institutions. Once these technical details are illegally copied or protocol parameters are reused without authorization, the economic value of the technical solution will be severely damaged. Especially in the current interface development and deployment model, which is primarily modular and automated, traditional methods such as patent registration, document signing, and code encryption are insufficient to cover the dynamic behavior during interface operation, easily creating a "declarable but difficult-to-verify" blind spot in ownership protection.

[0005] Furthermore, the GPMI interface standard's industrial ecosystem often involves collaborative development among multiple chip manufacturers, system integrators, and terminal manufacturers. Ambiguous definitions of interface configuration ownership, unclear division of responsibilities at the parameter level, and a lack of ownership verification during operation can easily lead to disputes over implementation rights. More seriously, there is currently a lack of mechanisms to verify interface technology ownership without disclosing implementation details. Once a dispute arises, the interface holder struggles to technically prove the legitimacy of their ownership, resulting in low efficiency, difficulty in obtaining evidence, and high costs in protecting their rights. Summary of the Invention

[0006] The purpose of this invention is to provide a watermark-based method and apparatus for detecting the ownership of a general multimedia interface protocol. Without changing the interface's functional logic and transmission performance, the method binds the interface configuration features and ownership information into a secret watermark and embeds it into the interface protocol module, thereby enabling automatic declaration, verification, and traceable confirmation of the ownership of the interface.

[0007] To address the aforementioned technical problems, a first aspect of this invention provides a watermark-based method for detecting ownership of a universal multimedia interface protocol. The interface holder, based on the universal multimedia interface protocol, holds the configuration parameters of the interface to be verified and performs a full-process ownership detection and verification operation. The detection method includes the following steps:

[0008] Extract the joint watermark from the GPMI interface configuration parameters to be verified through secure multi-party computation;

[0009] The eigenvalues ​​of the joint identity matrix are recovered based on the extracted joint watermark;

[0010] Using homomorphic encryption technology, the dense-state Euclidean distance between the feature value and the dense-state feature vector provided by the verifier is calculated;

[0011] The decrypted Euclidean distance is obtained through an outsourced decryption mechanism using double trapdoor homomorphic encryption. The ownership of the interface intellectual property rights is determined based on the comparison result between the Euclidean distance and a preset threshold.

[0012] Furthermore, the extraction of the joint watermark from the GPMI interface configuration parameters to be verified through secure multi-party computation includes:

[0013] The interface holder receives the second share of the watermark embedding matrix sent by the verification initiator. At the same time, the interface holder splits the interface configuration parameters to be verified into two components, the first parameter share and the second parameter share, which are equal to the original parameters. The first parameter share is then sent to the verification initiator.

[0014] The verification initiator calls multiple pre-generated verification triples, each consisting of three elements and satisfying a multiplicative secret sharing relationship. Based on their respective matrix shares and parameter shares, they perform vector multiplication operations row by row to jointly calculate the intermediate value of the product of the watermark embedding matrix and the interface configuration parameters. During the joint calculation process, the interface holder and the verification initiator do not exchange the original matrix shares and the original parameter shares.

[0015] The intermediate value of the product is input into a step function, which is defined as outputting a first state value when the input value is greater than or equal to zero and outputting a second state value when it is less than zero, thereby generating a joint watermark that represents the device's identity features.

[0016] Furthermore, the feature values ​​of the recovered joint identity matrix based on the extracted joint watermark include:

[0017] The binary sequence of the joint watermark is arranged from the most significant bit to the least significant bit. Each watermark value is multiplied by its corresponding bit weight and then summed to obtain a decimal numerical result.

[0018] The numerical results are used as the main feature values ​​of the joint identity matrix, which is generated by mapping the identity identifiers and attribute sets of the device components through a tree structure.

[0019] The trace eigenvalues ​​of the joint identity matrix are derived from the principal eigenvalues, and the trace eigenvalues ​​are mathematical eigenvalues ​​of the sum of the diagonal elements of the joint identity matrix.

[0020] The trace feature value is output as a feature value parameter for ownership verification.

[0021] Furthermore, the numerical result is used as the principal feature value of the joint identity matrix, which is generated by mapping the identity identifiers and attribute sets of the device-side components through a tree structure, including:

[0022] Obtain the unique identifiers and attribute sets of all participating device components, the attribute sets including device type, manufacturer code, and interface protocol version number;

[0023] Based on the hierarchical relationship between the identity identifiers and attribute sets, a tree-shaped data structure is constructed: the root node is set as the system authority identifier, the second-level nodes are the device type classifications, the third-level nodes are attached with specific device identity identifiers, and the leaf nodes are bound to attribute key-value pairs;

[0024] The tree-like data structure is mapped to a numerical matrix, a unique matrix index is assigned to each node, and the parent-child node relationship is transformed into the association weight value at the intersection of the matrix rows and columns, generating a joint identity matrix that reflects the device identity association.

[0025] The numerical result is used as the core eigenvalue of the joint identity matrix, which represents the scaling factor of the joint identity matrix in the direction of the largest eigenvector.

[0026] Furthermore, the step of using homomorphic encryption to calculate the encrypted Euclidean distance between the feature value and the encrypted feature vector provided by the verifier includes:

[0027] In the homomorphic encryption state, the trace feature value to be verified is subtracted from the encrypted feature vector provided by the verifier according to the element position to obtain the encryption difference vector;

[0028] Perform a homomorphic self-multiplication operation on each element of the encrypted difference vector to generate an encrypted square vector;

[0029] The encrypted square vector is processed by a preset number of vector rotation and accumulation operations. Each rotation and accumulation operation cyclically shifts the vector element by one position and adds it element by element to the vector before the shift, finally generating an encrypted distance vector in which only the first element is non-zero.

[0030] The first element of the encrypted distance vector is extracted as the encrypted Euclidean distance representing the joint identity similarity.

[0031] Furthermore, the method of obtaining the decrypted Euclidean distance through the outsourced decryption mechanism of double trapdoor homomorphic encryption includes:

[0032] The encrypted Euclidean distance is sent to the first decryptor and the second decryptor. The first decryptor holds the first homomorphic encryption partial private key, and the second decryptor holds the second homomorphic encryption partial private key. The partial private key is generated by randomly sampling and splitting the system master private key.

[0033] Receive a first intermediate decryption result generated by the first decryptor performing a partial decryption operation on the Euclidean distance of the encrypted state using a portion of its private key, and receive a second intermediate decryption result generated by the second decryptor performing a partial decryption operation using a portion of its private key;

[0034] By aggregating the first and second intermediate decryption results, a full decryption operation is performed to obtain a plaintext vector containing Euclidean distance, wherein only the first element of the plaintext vector is a non-zero value;

[0035] The first element of the plaintext vector is extracted as the actual Euclidean distance value.

[0036] Further, the interface holder aggregates the first and second intermediate decryption results and performs a full decryption operation to obtain a plaintext vector containing Euclidean distance, including:

[0037] Perform an arithmetic summation operation on the received first and second intermediate decryption results to generate an aggregated intermediate decryption value;

[0038] Extract a preset component from the original ciphertext of the dense-state Euclidean distance, wherein the preset component is a fixed data block generated during the encryption process based on the homomorphic encryption ring structure;

[0039] The preset component is subtracted from the aggregated decryption intermediate value to obtain the decrypted plaintext vector, in which all elements except the first element are zero.

[0040] Verify whether the numerical range of the first element is within the valid Euclidean distance interval. If valid, output the plaintext vector as the final decryption result.

[0041] Further, determining the ownership of interface intellectual property rights based on the comparison result of the Euclidean distance and the preset threshold includes:

[0042] The actual Euclidean distance value is compared with a preset intellectual property verification threshold value, which is a preset value based on the interface security level.

[0043] When the actual Euclidean distance value is less than or equal to the intellectual property verification threshold, an ownership confirmation certificate containing a verification timestamp and device identity information is generated, and it is determined that the interface configuration belongs entirely to the verification initiator.

[0044] When the actual Euclidean distance value is greater than the intellectual property verification threshold, a request for ownership tracing is initiated to an authoritative institution to carry out a multi-party collaborative process for solidifying infringement evidence.

[0045] Furthermore, the generation of the ownership confirmation credential containing the verification timestamp and device identity information includes:

[0046] The device identity identifier of the verification initiator, the actual Euclidean distance value, and the interface configuration parameter summary are combined into the core data of the ownership claim.

[0047] Request an authoritative institution to attach a tamper-proof verification label and a time stamp to the core data. The tamper-proof verification label is generated using an asymmetric cryptographic algorithm and can be publicly verified. The time stamp is issued by a trusted time source.

[0048] The core data, including the tamper-proof verification tag and time stamp, is encapsulated into a structured electronic certificate. The structured electronic certificate contains a machine-readable ownership declaration field and a verification metadata field.

[0049] Accordingly, a second aspect of the present invention provides a watermark-based universal multimedia interface protocol ownership detection device, which detects the ownership of a universal multimedia interface implementation using the above-described watermark-based universal multimedia interface protocol ownership detection method, including:

[0050] The watermark extraction module is used to extract a joint watermark from the configuration parameters of the GPMI interface to be verified through secure multi-party computation.

[0051] The feature extraction module is used to recover the feature values ​​of the joint identity matrix based on the extracted joint watermark;

[0052] The distance calculation module is used to calculate the dense Euclidean distance between the feature value and the dense feature vector provided by the verifier using homomorphic encryption technology.

[0053] The ownership determination module is used to obtain the decrypted Euclidean distance through the outsourced decryption mechanism of double trapdoor homomorphic encryption, and determine the ownership of the interface intellectual property rights based on the comparison result of the Euclidean distance and the preset threshold.

[0054] Accordingly, a third aspect of the present invention provides an electronic device, including: at least one processor; and a memory connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to cause the at least one processor to perform the above-described watermark-based universal multimedia interface protocol ownership detection method.

[0055] Accordingly, a fourth aspect of the present invention provides a computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the above-described watermark-based general multimedia interface protocol ownership detection method.

[0056] The above-described technical solutions of the embodiments of the present invention have the following beneficial technical effects:

[0057] 1. The attributes and identity information of device components are merged into a joint identity. A watermark is generated through feature value transformation and embedded into the interface protocol configuration as an encrypted credential for ownership verification. During ownership confirmation, the mathematical mapping relationship between the joint identity and the watermark can be used to recover the identity information from the interface configuration parameters to determine ownership, realizing ownership declaration based on identity. At the same time, the watermark embedding regularization term is introduced to achieve ownership binding with minimal impact on interface configuration performance, providing high security and high authentication efficiency for intellectual property verification, and significantly improving the authority and credibility of the ownership confirmation mechanism.

[0058] 2. Secure extraction of watermark information is achieved through secure two-party computation, ensuring the privacy and data security of participating parties; fully homomorphic encryption technology is used to complete the encrypted matching of device identities, effectively reducing the risk of identity leakage; and a double trapdoor fully homomorphic encryption algorithm is used to support outsourced decryption by the trusted center in offline mode, ensuring the security and trustworthiness of the data of both parties in the entire rights confirmation process, and providing a more robust and tamper-resistant intellectual property protection mechanism for the configuration of the joint interface protocol. Attached Figure Description

[0059] Figure 1 This is a flowchart of a watermark-based universal multimedia interface protocol ownership detection method provided in an embodiment of the present invention.

[0060] Figure 2 This is a block diagram of a watermark-based universal multimedia interface protocol ownership detection device module provided in an embodiment of the present invention.

[0061] Figure label:

[0062] 1. Watermark extraction module; 2. Feature value extraction module; 3. Distance calculation module; 4. Ownership determination module. Detailed Implementation

[0063] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments and the accompanying drawings. It should be understood that these descriptions are merely exemplary and not intended to limit the scope of the invention. Furthermore, descriptions of well-known structures and techniques are omitted in the following description to avoid unnecessarily obscuring the concept of the invention.

[0064] Please refer to Figure 1 The first aspect of this invention provides a watermark-based method for detecting the ownership of a universal multimedia interface protocol. The interface holder, based on the universal multimedia interface protocol, holds the configuration parameters of the interface to be verified and performs a full-process ownership detection and verification operation. The detection method includes the following steps:

[0065] S100 extracts a joint watermark from the GPMI interface configuration parameters to be verified through secure multi-party computation.

[0066] The interface holder extracts a joint watermark from the configuration parameters of the GPMI interface to be verified via a secure multi-party computation (MPC) protocol. The core of this process involves the verification initiator splitting the watermark embedding matrix into two additive secret shared shares (X1, X2) and sending X2 to the interface holder; simultaneously, the interface holder splits the interface configuration parameter θ into two shares (θ1, θ2) and sends θ1 to the verification initiator. Based on pre-generated Beaver triple vectors (satisfying a multiplicative secret sharing relationship), both parties jointly calculate the intermediate value of the product of the watermark embedding matrix XX and the interface configuration parameter θ through row-by-row vector multiplication without exchanging the original shares. This process leverages the blinding property of triples to ensure that neither party can steal the complete watermark matrix or configuration parameters. Finally, the interface holder inputs the product result into a step function (outputting 0 or 1) to generate a binary form of the joint watermark. Through mathematical partitioning and collaborative computation, watermark extraction is completed in a zero-knowledge environment, protecting the confidentiality of the verification party's watermark matrix and preventing the leakage of interface configuration parameters.

[0067] S200, recovering the eigenvalues ​​of the joint identity matrix based on the extracted joint watermark.

[0068] The interface holder, based on the extracted joint watermark (binary sequence), multiplies each watermark value by its bit weight from high to low bits and sums the results to obtain a decimal value. This value is directly used as the principal feature value of the joint identity matrix, which is generated by mapping the device component's identity identifier (such as device ID) and attribute set (such as manufacturer code, protocol version) through a tree structure: the root node is the authoritative institution identifier, the second-level nodes are classified according to device type, the third-level nodes are attached with specific device IDs, and the leaf nodes are bound to attribute key-value pairs. The association weights of parent and child nodes are transformed into the values ​​of the intersection points of the matrix rows and columns, forming a square matrix reflecting the association of device identities. The principal feature value characterizes the scaling strength of this matrix in the direction of the largest eigenvector and is the unique mathematical expression of the device group identity. The interface holder further derives the trace feature value (the sum of the elements on the diagonal of the matrix) as the core parameter for ownership verification. This aggregates discrete device identities into quantifiable mathematical features, providing unforgeable input for dense-state similarity calculation.

[0069] S300 uses homomorphic encryption technology to calculate the dense Euclidean distance between the feature value and the dense feature vector provided by the verifier.

[0070] The interface holder performs ownership similarity verification under a fully homomorphic encryption (CKKS algorithm) environment. First, it performs a closed-state subtraction operation on the trace feature value and the closed-state feature vector provided by the verifier, element by element, to generate an encrypted difference vector. Then, it performs homomorphic self-multiplication on each element of the difference vector to calculate the square, eliminating symbolic interference while amplifying feature differences. Next, it performs T-1 cyclic vector shift operations to perform rotational accumulation, adding the result element-wise to the original vector after each shift. Through iterative calculation, the sum of squares is gradually accumulated to the first element position, ultimately forming an encrypted vector where only the first element is non-zero. Extracting this first element yields the closed-state Euclidean distance representing the joint identity similarity. By fully utilizing the rotational homomorphic properties of the CKKS algorithm, the vector inner product operation is completed in a fully encrypted state, ensuring that sensitive data from both parties remains encrypted at all times.

[0071] S400 obtains the decrypted Euclidean distance through an outsourced decryption mechanism of double trapdoor homomorphic encryption, and determines the ownership of interface intellectual property rights based on the comparison result of the Euclidean distance and a preset threshold.

[0072] The interface holder initiates a double trapdoor decryption mechanism: the encrypted Euclidean distance is sent to the first and second decryption parties holding partial private keys, and each generates partially decrypted ciphertext; the interface holder performs arithmetic summation on the two to generate an aggregate value, and extracts the fixed component inherent to the encrypted ring structure from the original ciphertext, and obtains the plaintext vector by subtracting the correction operation of this component from the aggregate value; finally, the actual Euclidean distance value of the first element is compared with a preset threshold to generate a structured electronic certificate with an authoritative asymmetric signature and timestamp, determining the interface configuration ownership verification initiator; if the actual Euclidean distance value is greater than the preset threshold, the infringement tracing process is triggered. The private key splitting outsourcing mechanism not only avoids the risk of single-point decryption, but also strengthens the legal effect through authoritative endorsement, forming a closed-loop verifiable ownership protection system.

[0073] The above solution binds interface configuration features and ownership information into a secure watermark embedded in the interface protocol module without changing the interface functional logic and transmission performance. This enables automatic declaration, verification, and traceable ownership confirmation of the interface implementation rights. It is particularly suitable for GPMI interface systems with multi-layer protocol collaboration and cross-module implementation, and can widely support the security, reliability, and anti-tampering capabilities of interface ownership for device components in multi-party collaborative environments.

[0074] Furthermore, in S100, a joint watermark is extracted from the GPMI interface configuration parameters to be verified through secure multi-party computation, including:

[0075] S110, receive the second share of the watermark embedding matrix sent by the verification initiator, and at the same time, the interface holder splits the interface configuration parameters to be verified into two components, the first parameter share and the second parameter share, which are equal to the original parameters, and sends the first parameter share to the verification initiator.

[0076] When an interface holder (such as a device manufacturer) needs to claim ownership of an interface configuration, the verification initiator (such as a technology licensor) first sends the second share of the watermark embedding matrix to the interface holder. At this point, the interface holder splits the interface configuration parameters to be verified into two additive secret-shared components: the first parameter share is kept locally, while the second parameter share is sent to the verification initiator. This ensures that neither party can independently obtain the complete original interface configuration parameters or the watermark embedding matrix, fundamentally avoiding the risk of leakage of sensitive technical parameters during the verification process. Through distributed secret-sharing technology, without the need for third-party intervention, the access requirements for interface configuration data for ownership verification are met while strictly adhering to the technical confidentiality principles in multi-party collaboration within the industry ecosystem, effectively solving the problem of "unclear division of parameter-level responsibilities" pointed out in the background technology.

[0077] S120, the verification initiator calls multiple pre-generated verification triples. Each verification triple consists of three elements and satisfies the multiplicative secret sharing relationship. Based on their respective matrix shares and parameter shares, they perform vector multiplication operations row by row to jointly calculate the intermediate value of the product of the watermark embedding matrix and the interface configuration parameters. During the joint calculation, the interface holder and the verification initiator do not exchange the original matrix shares and the original parameter shares.

[0078] The above steps achieve secure watermark extraction through a cryptographic protocol. The interface holder and the verification initiator collaborate to perform the product calculation of the watermark embedding matrix and interface configuration parameters based on pre-generated multiple sets of Beaver triples. Specifically, both parties use their respective matrix and parameter shares to perform bitwise multiplication of vector elements row by row, masking the actual calculated value with redundant data in the triples. During this process, all interactive data is secretly shared, and the original matrix and configuration parameters remain encrypted or segmented, preventing either party from deducing the original information held by the other. This joint computation mechanism based on secure multi-party computation perfectly adapts to the multi-layered protocol collaboration characteristics of the GPMI interface—it can embed watermarks at various levels of the protocol stack (such as physical layer modulation parameters and protocol layer timing control parameters) while ensuring that key technical details are not exposed during cross-module implementation. Its innovation lies in transforming linear algebra operations into a cryptographic protocol, providing provably secure mathematical guarantees for ownership verification in dynamic interface behavior.

[0079] S130, input the intermediate value of the product into the step function. The step function is defined as outputting a first state value when the input value is greater than or equal to zero and outputting a second state value when it is less than zero, thereby generating a joint watermark that represents the device's identity characteristics.

[0080] The above steps complete the final reconstruction and ownership confirmation preparation of the watermark information. The intermediate value of the product obtained in step S120 (i.e., the linear combination result of the watermark embedding matrix and configuration parameters) is input into a step function for binarization: when the input value is greater than or equal to zero, it outputs "1", and when it is less than zero, it outputs "0", thereby generating a joint watermark sequence in binary form. This nonlinear conversion process maps the interface features in floating-point form to a discrete bit stream, significantly improving the robustness of the watermark under transmission noise or configuration disturbances, such as resisting timing offsets caused by electromagnetic compatibility; in addition, the generated binary watermark directly corresponds to the device identity features, providing standardized input for subsequent ownership determination. As can be seen from the whole text, this step function, as the last link in the watermark recovery link, forms a closed-loop verification logic with the dense-state Euclidean distance calculation, enabling the scheme to tolerate reasonable performance fluctuations in interface configuration (such as link adaptive adjustment) and accurately capture malicious tampering behavior, ultimately achieving a balance between "interface function fidelity" and "ownership binding strength".

[0081] The watermark extraction process described above does not require decryption of interface configuration parameters or the watermark matrix, thus protecting the technical privacy of chip manufacturers, integrators, and other participants. Furthermore, it ensures the non-repudiation of ownership claims through a mathematical mapping relationship (step function output → eigenvalue → joint identity). Especially for GPMI interfaces in complex electromagnetic environments such as smart TVs and automotive systems, this method can stably extract watermarks even during dynamic interface reconfiguration (e.g., hot-plug triggering link adaptation), significantly outperforming traditional static signature verification mechanisms.

[0082] Furthermore, the eigenvalues ​​of the extracted joint identity matrix based on the joint watermark recovery in S200 include:

[0083] S210: The binary sequence of the joint watermark is ordered from high bit to low bit. Each watermark value is multiplied by its corresponding bit weight and then summed to obtain a decimal result.

[0084] The above steps transform watermark information into mathematical features. The binary sequence of the joint watermark is expanded according to bit weights (higher bits correspond to higher weights), and a decimal value is generated through weighted summation. This is essentially the reverse process of binary encoding, restoring the discrete bit stream to continuous values, providing input for subsequent matrix operations. A mapping relationship between the watermark and mathematical space is established: each bit in the watermark sequence does not exist independently but constitutes a holistic feature through bit weight design. For example, the weight of the highest bit is much greater than that of the lowest bit, ensuring that even if lower bit flips occur due to interface transmission noise (such as electromagnetic compatibility interference), the recovered value still retains the main features. This fault-tolerant mechanism significantly improves the usability of the watermark in complex environments, especially adapting to the dynamic link conditions faced by the GPMI interface in scenarios such as smart TVs and in-vehicle systems.

[0085] S220, the numerical result is used as the main feature value of the joint identity matrix, which is generated by mapping the identity identifiers and attribute sets of the device components through a tree structure.

[0086] The joint identity matrix is ​​not arbitrarily generated, but rather constructed by mapping the identity identifiers (IDs) and attribute sets (ATs) of device components through a tree structure. The tree structure inherently possesses hierarchy: the root node represents the global characteristics of the device cluster, while the leaf nodes correspond to the independent attributes of individual devices. This mapping aggregates scattered device identity information into a unified mathematical expression, making the principal eigenvalue a unique representation of the cluster's identity. The principal eigenvalue reflects the direction of the matrix's maximum variance and is highly sensitive to the common characteristics of the device group. When an unauthorized device attempts to infiltrate the cluster, its attributes cause abrupt changes in the matrix structure, resulting in a significant deviation of the principal eigenvalue, thus providing a basis for detecting ownership anomalies.

[0087] S230, derive the trace eigenvalues ​​of the joint identity matrix from the principal eigenvalues. The trace eigenvalues ​​are mathematical eigenvalues ​​of the sum of the diagonal elements of the joint identity matrix.

[0088] The trace, as the sum of the diagonal elements of a matrix, has several advantages. First, the trace equals the sum of all eigenvalues, meaning the principal eigenvalues ​​can be linearly related to the trace. Second, the trace remains unchanged by matrix rotations, scaling, and other operations. This stability is crucial for ownership verification: interface configuration during legitimate tuning (such as adaptive adjustment of protocol parameters) may cause minor perturbations in the matrix, but the trace eigenvalues ​​remain relatively constant; while malicious tampering (such as replacing device attributes) will cause drastic fluctuations in the trace eigenvalues. Furthermore, calculating the trace does not require the complete matrix, only the diagonal elements, which provides an efficiency advantage for subsequent encrypted verification; the overhead of calculating the trace under homomorphic encryption is far lower than full matrix operations.

[0089] S240 outputs trace feature values ​​as feature value parameters for ownership verification.

[0090] First, the trace feature value encapsulates the identity essence of the device cluster and has a fixed dimension (scalar form), which can be directly used for dense state similarity comparison. Second, as the regularization target for watermark embedding in step 3, it forms a two-way binding with the interface configuration parameters. During the embedding stage, the configuration parameters are forced to approximate the trace feature value, and during the extraction stage, the value is restored from the configuration parameters, forming a closed-loop verification logic. Finally, at the level of industry collaboration, the trace feature value does not disclose individual device information (tree mapping has unidirectionality), which meets the protection needs of sensitive data in multi-party collaboration and effectively solves the pain point of "difficulty in verifying ownership without disclosing details" in the background technology.

[0091] The above-described method progressively transforms the physical layer's binary watermark into algebraic features (trace values), ultimately serving as a legal basis for ownership claims. Each step—tree mapping, matrix generation, and feature extraction—is verifiable and relies on mature linear algebra theory, avoiding the "declarable but difficult to verify" flaw of traditional solutions. Especially in ultra-high-definition video transmission scenarios, when the interface triggers protocol reconfiguration due to bandwidth fluctuations, it ensures that the trace value remains stable within the legal optimization range, while significantly deviating from unauthorized copying behavior, providing chip manufacturers and system integrators with irrefutable evidence of technology ownership.

[0092] Furthermore, in S220, the numerical result is used as the main feature value of the joint identity matrix. The joint identity matrix is ​​generated by mapping the identity identifiers and attribute sets of the device components through a tree structure, including:

[0093] S221, obtain the unique identifiers and attribute sets of all participating device-side components. The attribute set includes device type, manufacturer code, and interface protocol version number.

[0094] By collecting unique identifiers (such as chip serial numbers) and structured attribute sets (device type distinguishing physical forms like smart TVs / vehicle terminals, manufacturer code identifying the technology owner, and interface protocol version number locking in technical specifications) from all device components, a raw material library for ownership verification was constructed. Device type determines the basic framework of interface configuration (e.g., ultra-high-definition transmission requires specific timing control), manufacturer code relates to patent pool licensing relationships, and protocol version number corresponds to the iterative trajectory of implementation details. These three types of attributes form complementary dimensions, ensuring that the subsequently generated joint identity matrix can both represent the commonalities of technical solutions (e.g., standard configurations for the same device type) and capture the specificities of the implementer (e.g., manufacturer-specific optimization parameters), thus avoiding the risk of "ambiguous definition of interface configuration ownership" in the background technology from the source.

[0095] S222, construct a tree-like data structure based on the hierarchical relationship between the identity identifier and the attribute set: set the root node as the system authority identifier, the second-level nodes as the device type classification, the third-level nodes as the specific device identity identifier, and the leaf nodes as the attribute key-value pairs.

[0096] The hierarchical organization of identity logic is achieved through a tree structure. The root node is anchored to the system's authoritative body (TA), giving the matrix a globally trusted root. Second-level nodes are categorized by device type (e.g., display terminal / audio processing module), mapping to functional partitions in the industry ecosystem. Third-level nodes carry specific device IDs, identifying physical entities. Leaf nodes bind attribute key-value pairs (e.g., <manufacturer, 0x05A3><protocol version, V1.2.7>), carrying technical details. First, the hierarchical relationship directly reflects the topology of the device cluster (e.g., the master-slave relationship between the TV and speakers in a home theater system), allowing the matrix to naturally embed system-level ownership information. Second, the addition or deletion of tree nodes only affects the local structure, supporting dynamic expansion during hot-swapping of devices—when a new device is added, only the leaf nodes under the corresponding type branch need to be expanded, without reconstructing the entire matrix, perfectly adapting to the "plug-and-play" characteristics of the GPMI interface.

[0097] S223 maps the tree data structure into a numerical matrix, assigns a unique matrix index to each node, and transforms the parent-child node relationship into the association weight value at the intersection of the matrix rows and columns, generating a joint identity matrix that reflects the device identity association.

[0098] Each node is assigned a unique matrix index (e.g., the root node corresponds to row 0 / column 0), and parent-child relationships are transformed into matrix intersection weights: if there is a parent-child link between node A and B, then the matrix element M[A][B] is assigned the path weight; the weights of non-directly connected nodes are set to zero. The generated joint identity matrix is ​​essentially a mathematical abstraction of device relationships: diagonal elements reinforce individual identities (e.g., manufacturer-specific attributes), and off-diagonal elements encode inter-device collaborative relationships (e.g., protocol version compatibility constraints). The sparsity of the matrix and the weight distribution naturally resist forgery attacks. If an attacker tampers with the attributes of a single device, it will disrupt the consistency of local weights; if a device node is cloned, it will be detected due to index conflicts. This mechanism mathematically achieves the "attribution verification during operation" pursued by the background technology.

[0099] S224 uses the numerical results as the core eigenvalues ​​of the joint identity matrix, which represent the scaling factor of the joint identity matrix in the direction of the largest eigenvector.

[0100] The principal eigenvalue is extremely sensitive to changes in the overall matrix structure, while remaining robust to other perturbations. Specifically, when the interface configuration is legally optimized (e.g., fine-tuning protocol parameters), the matrix fluctuates only in the directions of secondary eigenvectors, with minimal changes in the principal eigenvalue. However, when ownership infringement occurs (e.g., unauthorized copying of configuration), a sudden change in the matrix structure leads to a significant shift in the principal eigenvalue. More importantly, this eigenvalue forms a closed loop with the watermark embedding. The watermark loss function forces the interface configuration parameters to converge towards this eigenvalue, while ownership verification extracts this value from the configuration, forming a mathematically driven chain of evidence for ownership confirmation.

[0101] Furthermore, S300 utilizes homomorphic encryption technology to calculate the encrypted Euclidean distance between the feature values ​​and the encrypted feature vectors provided by the verifier, including:

[0102] S310, under homomorphic encryption, the trace feature value to be verified is subtracted from the encrypted feature vector provided by the verifier according to the element position to obtain the encrypted difference vector.

[0103] The trace feature value to be verified and the encrypted feature vector provided by the verifier (encrypted and signed by an authoritative institution TA) are homomorphically subtracted at element position to generate an encrypted difference vector. This achieves end-to-end encrypted protection: the subtraction operation is directly applied to the ciphertext through a homomorphic encryption algorithm, without the need to decrypt the original feature value or feature vector, completely eliminating the risk of leakage of sensitive identity information during ownership verification. Each element corresponds to a certain dimension difference in the device identity matrix (such as manufacturer attribute deviation, protocol version offset), providing a basis for subsequent comprehensive judgment. Especially in multi-party collaboration scenarios, the above mechanism ensures that the chip manufacturer's proprietary technical parameters (such as electromagnetic compatibility compensation coefficient) are always in an encrypted state, fundamentally solving the "data confidentiality dilemma in cross-module implementation" in the background technology.

[0104] S320 performs a homomorphic self-multiplication operation on each element in the encrypted difference vector to generate an encrypted square vector.

[0105] A homomorphic self-multiplication operation (i.e., squaring) is performed on each element of the encrypted difference vector to generate an encrypted squared vector. The square function amplifies significant differences; when an attribute is altered due to infringement (such as unauthorized protocol version changes), the square value at the corresponding position will increase sharply. The convex function property of the square operation ensures that parameter fluctuations within the legal range (such as link adaptive fine-tuning triggered by hot-plugging) only cause gradual changes. The self-multiplication process is completed entirely in encrypted state. Relying on the homomorphic nature of homomorphic encryption, it avoids decryption overhead while inheriting the security strength of the encryption scheme. This implementation provides feasibility for real-time systems such as ultra-high-definition video transmission, achieving millisecond-level ownership verification response even with 10-gigabit bandwidth.

[0106] S330 processes the encrypted squared vector through a preset number of vector rotation and accumulation operations. Each rotation and accumulation operation cyclically shifts the vector element by one position and adds it to the vector before the shift element by element, ultimately generating an encrypted distance vector with only the first element being non-zero.

[0107] Utilizing the unique vector rotation operation of CKKS homomorphic encryption, the encrypted squared vector is iteratively accumulated: each rotation cyclically shifts the vector elements by one position and adds them element-wise to the state before the shift. After a preset number of rotations (usually the vector length minus one), the squared values ​​of all elements are aggregated to the first element, generating an encrypted distance vector where only the first element is non-zero. Conventional schemes require decrypting each element individually before summing, while rotation accumulation directly achieves cross-element aggregation in the ciphertext space, reducing computational complexity from linear to constant. When the interface protocol stack contains multiple layers of parameters (physical layer modulation factor, protocol layer timing control, etc.), a full comparison of hundreds of dimension feature vectors can be completed at once, perfectly supporting the "integrated interface ownership verification" emphasized in the background technology.

[0108] S340, extract the first element of the encrypted distance vector as the encrypted Euclidean distance representing the joint identity similarity.

[0109] The first element of the encrypted distance vector is the encrypted Euclidean distance (ED), which represents the similarity of the joint identities. Essentially, it is the sum of squares of the differences between the trace feature value and the verifier's feature vector across all dimensions. This compresses multidimensional authentication into a single-value judgment: firstly, mathematical proof shows that a smaller ED value indicates a better match between the interface configuration and the verifier's identity; secondly, the single-value form is highly suitable for subsequent threshold comparison mechanisms. Crucially, this distance value remains encrypted throughout the entire process, only being decrypted collaboratively by multiple parties during the final verification stage (AS and the verification initiator each hold a fragment of the private key), ensuring that even if the system is partially compromised, attackers cannot tamper with or forge the distance calculation result. This end-to-end security guarantee enables this method to effectively resist malicious behaviors such as man-in-the-middle attacks and replay attacks in security-sensitive scenarios such as in-vehicle entertainment systems.

[0110] Furthermore, the outsourced decryption mechanism in S400, which uses double trapdoor homomorphic encryption, obtains the decrypted Euclidean distance, including:

[0111] S411, the encrypted Euclidean distance is sent to the first decryptor and the second decryptor. The first decryptor holds the private key of the first homomorphic encryption part, and the second decryptor holds the private key of the second homomorphic encryption part. The private key is generated by randomly sampling and splitting the system master private key.

[0112] During the initialization phase, the system's master private key (i.e., the homomorphic encryption private key) is randomly sampled and split into two parts by an authoritative body (TA), which are then securely distributed to the first decryption party (the interface control unit AS) and the second decryption party (the verification initiator). This splitting is based on the mathematical principle of secret sharing, ensuring that no single decryption party can independently recover the complete private key. When the interface holder needs to verify ownership, the encrypted Euclidean distance ciphertext is sent to both parties simultaneously. This fundamentally avoids the risk of single-point private key leakage and conforms to the "least privilege" security principle in multi-party collaborative scenarios, providing a trusted foundation for subsequent decryption.

[0113] S412, receive the first intermediate decryption result generated by the first decryptor performing a partial decryption operation on the Euclidean distance of the encrypted state using a portion of its private key, and receive the second intermediate decryption result generated by the second decryptor performing a partial decryption operation using a portion of its private key.

[0114] Upon receiving the ciphertext, the first decryption party (AS) and the second decryption party (verification initiator) each perform partial decryption operations using a portion of their private keys. Specifically, each party performs local computations on the ciphertext using a portion of its private key, generating an intermediate result, according to the decryption rules of the homomorphic encryption algorithm. This process does not require interaction or sharing of private key fragments between the parties, effectively protecting the locality of the partial private key. Partial decryption breaks down the complete decryption operation into two linearly related sub-steps, each of which only generates incomplete intermediate ciphertext and cannot directly expose the plaintext information. This design ensures the verifiability of the decryption process while preventing either party from stealing or tampering with the final result.

[0115] S413, aggregate the first and second decryption intermediate results, perform full decryption operation to obtain a plaintext vector containing Euclidean distance, in which only the first element is a non-zero value.

[0116] The interface holder, acting as the coordinator of the decryption process, collects the intermediate results returned by both parties and performs aggregation operations. The core of this aggregation operation is the complete decryption function, which reconstructs the plaintext vector through linear combination and noise cancellation mechanisms. The vector's unique structure stems from the ingenious dense-state Euclidean distance calculation design in the patent. Through homomorphic rotation and accumulation operations, the similarity of the multidimensional vector is compressed to the first element, with the remaining positions cleared to zero. This structure not only simplifies the parsing logic but also significantly reduces the complexity of subsequent processing, ensuring the efficiency and accuracy of the decryption output.

[0117] S414 extracts the first element of the plaintext vector as the actual Euclidean distance value.

[0118] The interface holder extracts the first element from the aggregated plaintext vector, which is the actual Euclidean distance. This value is directly used for ownership determination: by comparing it with a preset threshold, an objective ruling on intellectual property ownership is achieved. The watermark embedding stage encodes the joint identity features into feature values, while the ownership verification stage recovers the quantified discrepancies between the suspected configuration and the declared identity. This mathematical mapping relationship provides strong robustness; even if the interface configuration is slightly tampered with or subject to noise interference, as long as the watermark features do not exceed the fault tolerance threshold, the ownership determination can still be accurately triggered. More importantly, the outsourced decryption mechanism ensures that the generation of the actual Euclidean distance value does not leak the original identity matrix or the plaintext of the watermark throughout the entire process, fundamentally eliminating the risk of information leakage during the verification process.

[0119] First, by employing double trapdoor splitting and distributed partial decryption, the single point of failure problem in homomorphic encryption private key management is solved, meeting the security requirements of multi-party checks and balances in industrial scenarios. Second, the calculation and verification of the encrypted Euclidean distance are performed entirely within the encrypted domain, ensuring that the interface holder, AS, and verification initiator cannot obtain each other's sensitive data (such as joint identity or feature vectors), achieving a closed loop of privacy protection. Finally, by combining the regularization constraints and smoothing loss function of watermark embedding, the solution endows the watermark with extremely strong resistance to attacks while maintaining interface transmission performance (such as low latency and high bandwidth). For infringement behaviors such as parameter tampering, model theft, or protocol cloning, rapid evidence collection can be achieved through statistically significant deviations in the actual Euclidean distance values, significantly improving the efficiency of rights protection and judicial credibility.

[0120] Furthermore, the interface holder in S413 aggregates the first and second intermediate decryption results and performs a full decryption operation to obtain a plaintext vector containing Euclidean distance, including:

[0121] S413a: Perform an arithmetic summation operation on the received first and second decryption intermediate results to generate an aggregated decryption intermediate value.

[0122] After receiving partial decryption results from the first decryption party (interface control unit AS) and the second decryption party (verification initiator), the interface holder performs an arithmetic summation operation. Utilizing the mathematical properties of homomorphic encryption algorithms, it combines the two partially decrypted data processed by different key fragments. Each intermediate result contains only partial decryption information and cannot independently reconstruct the plaintext, but their summation reconstructs the key structure of the complete decryption logic. This operation involves only basic arithmetic operations, requiring no complex cryptographic protocol interactions, significantly reducing the system's computational burden. More importantly, the summed result remains in a protected intermediate state, ensuring that the sensitive Euclidean distance value is not leaked to any participant before final decryption.

[0123] S413b extracts a preset component from the original ciphertext with the dense state Euclidean distance. The preset component is a fixed data block generated during the encryption process based on the homomorphic encryption ring structure.

[0124] Extracting the predefined component is a core step in homomorphic encryption. This component is not generated ad hocly, but rather deterministically produced during the initial encryption of Euclidean distance data, based on the mathematical structure of the encryption algorithm (such as polynomial rings and noise models). Its value is entirely determined by the system parameters selected during the encryption phase (including modulus size, noise distribution characteristics, etc.). The interface holder can directly read this component from the inherent structure of the ciphertext to be decrypted, without additional calculations or external assistance. This design embeds the key noise reduction factor required for decryption within the ciphertext itself, avoiding reliance on the key holder during decryption and ensuring the self-sufficiency and consistency of the decryption process.

[0125] S413c subtracts a preset component from the aggregated decryption intermediate value to obtain the decrypted plaintext vector, in which all elements except the first element are zero.

[0126] By subtracting a preset component from the aggregated intermediate value, precise removal of encryption noise is achieved. Mathematically, this is the reverse of the encryption process: encryption introduces noise with a specific structure to protect the plaintext, while decryption removes this noise through a preset subtraction operation. The decrypted output is a special vector form where only the first element is non-zero; this structure is not randomly generated but stems from a technical design in the previous steps that compresses multidimensional similarity information to the first and second elements of the vector through homomorphic operations. The generation of zero-valued elements actively eliminates redundant interference information, simplifying the parsing process and ensuring the clarity and reliability of the decryption result, effectively improving the effectiveness of the output signal.

[0127] S413d verifies whether the numerical range of the first element is within the valid Euclidean distance interval. If valid, it outputs the plaintext vector as the final decryption result.

[0128] Verifying the range of the first element's value after decryption is a crucial step in preventing security threats. While Euclidean distance is inherently non-negative and theoretically infinitely large, in practical interface ownership verification scenarios, a valid match must fall within a reasonable range defined by the watermark strength and system fault tolerance. By determining whether the value falls within this preset valid range (e.g., close to zero or not exceeding a specific technical threshold), the interface holder can efficiently identify and eliminate three types of anomalies: invalid values ​​due to decryption errors, distorted results caused by malicious forgery of ciphertext by attackers, and meaningless matches resulting from watermark corruption. This verification is performed directly in plaintext, requiring no additional cryptographic operations, and can guarantee the validity and credibility of the ownership confirmation conclusion at extremely low cost.

[0129] Furthermore, the S400 determines the ownership of interface intellectual property rights based on the comparison result of Euclidean distance and a preset threshold, including:

[0130] S421 compares the actual Euclidean distance value with a preset intellectual property verification threshold value, which is a preset value based on the interface security level.

[0131] The core logic of intellectual property ownership determination lies in comparing the actual Euclidean distance with a preset threshold. This threshold is not a fixed value but is dynamically set according to the security level of the interface: a stricter threshold is used in high-security scenarios (such as medical device interfaces) to reduce the false positive rate, while the threshold is appropriately relaxed in ordinary scenarios (such as home audio-visual interfaces) to improve compatibility. The distance value essentially reflects the similarity of watermark features; the smaller the distance, the closer the identity features extracted from the interface configuration are to the claimant. This determination mechanism based on mathematical similarity transforms subjective ownership disputes into objective quantitative comparisons, significantly improving the credibility of the verification results. The threshold setting integrates watermark embedding strength, interface protocol tolerance, and industry standard requirements, ensuring accurate ownership determination while protecting interface functionality.

[0132] S422, when the actual Euclidean distance value is less than or equal to the intellectual property verification threshold, an ownership confirmation certificate containing verification timestamp and device identity information is generated, and it is determined that the interface configuration belongs entirely to the verification initiator.

[0133] When the Euclidean distance does not exceed a threshold, a legally valid ownership confirmation credential is automatically generated. This credential contains two key elements: first, a trusted timestamp accurate to the millisecond level, issued by an authoritative time service, proving the absolute point in time when the verification occurred; and second, the digital identity information of the devices participating in the verification, linked across devices through a pre-registered federated identity tree. Credential generation represents legal ownership confirmation, and the verification initiator immediately receives the exclusive rights statement configured by this interface. This process is fully automated, requiring no manual intervention. It avoids the time-consuming nature of traditional notarization processes and uses cryptographic technology to ensure the credential is unforgeable and non-repudiable, providing immediate and effective electronic evidence for subsequent technology licensing or infringement litigation.

[0134] S423: When the actual Euclidean distance value is greater than the intellectual property verification threshold, a request for ownership tracing is initiated to an authoritative institution to carry out a multi-party collaborative process for solidifying infringement evidence.

[0135] If the Euclidean distance exceeds the security threshold, an ownership tracing mechanism is automatically triggered. This mechanism includes three layers of protection: First, a dispute request is submitted to an authoritative institution, activating pre-set distributed audit nodes to cross-verify the entire verification process logs, ensuring no operational tampering. Second, a multi-party collaborative evidence solidification process is initiated, encapsulating key information such as interface configuration parameters, watermark extraction records, and decryption intermediate data into an indivisible evidence package via a secure hash chain. Finally, based on blockchain technology, the evidence package is anchored to a public ledger, forming a timestamped, judicial-grade evidence. This not only achieves real-time preservation of infringement evidence but also prevents single-point malicious acts through a multi-institutional witnessing mechanism, constructing a complete evidence loop for subsequent legal proceedings.

[0136] Furthermore, the generation of the ownership confirmation credential containing the verification timestamp and device identity information in S422 includes:

[0137] S422a combines the device identity identifier of the verification initiator, the actual Euclidean distance value, and the interface configuration parameter summary into the core data of the ownership claim.

[0138] The core data for the ownership claim is integrated from the device identifier of the verification initiator, the actual Euclidean distance value, and the interface configuration parameter digest. The device identifier originates from a previously generated joint identity tree (such as ID and attribute set), ensuring uniqueness. The Euclidean distance value is a quantification of identity similarity obtained after decryption. The interface configuration parameter digest generates a fixed-length feature fingerprint from key parameters of the current interface configuration (such as modulation factor, control factor, etc.) using a hash function. These three elements combine to form an inseparable chain of verification evidence: the device identifier identifies the ownership subject, the Euclidean distance proves the technical attribution matching degree, and the configuration digest anchors the specific interface instance. This binding mechanism technically solidifies the integrity and relevance of the ownership claim, preventing isolated tampering of verification elements.

[0139] S422b requests an authoritative body to attach a tamper-proof verification label and a time stamp to the core data. The tamper-proof verification label is generated using an asymmetric cryptographic algorithm and can be publicly verified. The time stamp is issued by a trusted time source.

[0140] Core data must be submitted to an authoritative institution requesting additional security credentials. The authoritative institution uses an asymmetric cryptographic algorithm (such as a signature scheme based on a multiplicative cyclic group G) to generate a tamper-proof verification tag for the core data: the authoritative institution encrypts the data hash value with the system signature private key SKTASKTA, generating a publicly verifiable digital signature. This tag ensures that any tampering with the core data will cause signature verification to fail, achieving data integrity and non-repudiation. Simultaneously, the authoritative institution introduces a timestamp issued by a trusted time source (such as the RFC 3161 timestamp protocol) to accurately record the moment the verification operation occurred. The timestamp and the tamper-proof tag work together to resist backtracking attacks (preventing the forgery of ownership claims after the fact) and provide legally valid chronological evidence for dispute resolution.

[0141] S422c encapsulates core data with added tamper-proof verification tags and timestamps into a structured electronic certificate. The structured electronic certificate contains a machine-readable ownership declaration field and a verification metadata field.

[0142] Structured electronic credentials include an ownership declaration field and a verification metadata field. The ownership declaration field stores the device identity, ED value, configuration digest, and verification conclusion (success / failure) in a machine-readable format (e.g., JSON-LD), supporting automated ownership verification. The verification metadata field embeds information such as the public key index of the tamper-proof tag, the validity period of the timestamp, and the TA certificate chain, allowing the verifier to directly call public parameters to verify the authenticity of the signature and timestamp. This transforms the technical verification result into legally recognized electronic evidence: the ownership declaration field directly carries the technical ownership confirmation conclusion, while the metadata field provides lightweight, independent verification capabilities, enabling credential authenticity verification without relying on the original interface system, thus solving the blind spot problem of "declarable but difficult to verify" in traditional ownership protection.

[0143] The following is a complete embodiment to further explain and illustrate the above content:

[0144] Step 1) Initialize the GPMI interface system:

[0145] Initialization includes authoritative institutions Interface control unit Protocol Coordination Module as well as GPMI interface system for individual device components; authoritative institutions Initialize two orders as prime numbers The multiplicative cyclic group with generators is and ,in The generator is Authoritative institutions from Select the system signature private key The public key for verifying signatures in the computing system is ; Choose a collision-resistant, secure hash function And make public system parameters ; Generate two pairs of homomorphic encryption key pairs , and , ;Will Randomly split into two parts of the private key , ;Please set common parameters Public key , Broadcast to interface control unit and each device component Will be passed The calculated partial private key , Each broadcast securely to the interface control unit. Each device-side component Interface control unit Initialize the target module in the GPMI interface protocol stack For the local interface protocol module, and will Global initial iteration rounds and the global maximum number of iterations Broadcast to each device component Each device-side component The interface configuration dataset held is ,in, , , Indicates inclusion The configuration set for the behavior configuration of the interface, which contains the first... The configurations and their corresponding function category labels are as follows: , , .

[0146] Specifically, fully homomorphic encryption uses Algorithm, public key , and private key , The generation method is as follows:

[0147] ;

[0148] ;

[0149] ;

[0150] ;

[0151] ;

[0152] ;

[0153] ;

[0154] ;

[0155] ;

[0156] ;

[0157] in, Denotes a base, and , Represents the modulus. This indicates the depth of homomorphic multiplication. Indicates safety parameters, This represents the modulo operation. This represents a cycloidic polynomial of the second power. Represents positive integers. Represents integers, Represent real numbers, Indicates from The weight of Hamming randomly selected from the middle is equal to of dimensional vector, Indicates from the ring Extraction 3D polynomial vector, Each coefficient is taken from the variance. The discrete Gaussian distribution.

[0158] Among them, prime numbers Randomly select from the set of prime numbers with 1024 digits.

[0159] Specifically, partial private key , The calculation formulas are as follows:

[0160] ;

[0161] ;

[0162] ;

[0163] ;

[0164] ;

[0165] ;

[0166] in, The Bernoulli distribution with a representative parameter of 0.5 is used. Represented by Bernoulli distribution Randomly sampled values, For multiple passages through Bernoulli distribution Sample composition Dimensional vector.

[0167] Step 2) Negotiating the copyright of the joint interface for device-side components:

[0168] Based on device-side components Each person's identity marker With attribute collection Generate a joint identity tree ; Joint Identity Tree After mapping, a Joint identity matrix Then, using the joint identity matrix Solving for generating joint identities 。; Homomorphic public key Encrypted Federation Identity and eigenvectors ; for encrypted joint identity and eigenvectors Perform the signature separately; Design a one-way function Obtain the watermark embedding matrix ; Sending information to device components via secure channel Send watermark embedding matrix eigenvalues Secret Joint Identity and its signature and signature timestamp ,Towards Send dense-state feature vectors and its signature and signature timestamp Each device component will As a commitment to identity made public, and will Convert to the joint watermark of the interface to be embedded .

[0169] Furthermore, solve the joint identity problem. The generation method is as follows:

[0170] ;

[0171] in As a random factor, To and Random matrices of the same dimension and The elements follow a normal distribution.

[0172] Specifically, solve for the watermark embedding matrix. The generation method employs a matrix polynomial solution approach to calculate the joint identity matrix. Eigenvalues ​​and eigenvectors:

[0173] ;

[0174] in For a joint identity matrix Eigenvalues This is the corresponding feature vector.

[0175] By generating joint identity information through trusted institutions, undeniable evidence is provided for confirming intellectual property rights. Combined with digital watermarking technology, a mapping relationship is established between joint identity and interface copyright watermarks. In the event of a property dispute, the interface holder can extract the watermark from the suspicious interface configuration and restore the joint identity information. By comparing and verifying the ownership interface configuration, a more authoritative and credible proof of ownership is provided.

[0176] Furthermore, solve the joint watermark. The generation method is as follows:

[0177] .

[0178] Step 3) The device-side components inject the joint interface implementation ownership into the interface protocol module:

[0179] Each device-side component Configure the dataset through your own interface. For the local interface protocol module Conduct the first This configuration was optimized and implemented using the public key. Interface configuration parameters Perform fully homomorphic encryption, then configure the encrypted local interface parameters. Upload to interface control unit ,in Represents the encryption operation in a fully homomorphic encryption algorithm; Interface control unit calculate Encrypted average value of local parameters of each device component ; Will The device-side components use the private key to distribute the data. Decrypt.

[0180] In this embodiment, each device-side component By configuring the parameters in the local interface protocol module Optimize the interface to embed a joint identity watermark without changing the interface protocol's functional behavior.

[0181] In the In this configuration optimization, the fidelity loss item for interface functionality is defined. This is used to measure the difference between the interface protocol behavior under the current configuration and the standard functional reference. This difference can be modeled by the deviation of interface link stability, transmission delay, bit error rate, or protocol response curve. Loss Term It can be expressed by the following formula:

[0182] ;

[0183] in, Indicates the configuration parameters of the current interface. The resulting protocol execution response, This represents an ideal or standard interface response. It represents an arbitrary differentiable response deviation measurement function.

[0184] To bind ownership information to the interface while maintaining its functionality and performance, this invention introduces a watermark embedding regular expression. This item is used to map the joint identity information into an embeddable parameter form, forcing the bootstrap interface protocol parameters to be close to the embedded watermark:

[0185] ;

[0186] in, For device-side components A subset of adjustable configuration parameters (such as specific modulation factors and control factors in the interface protocol module). For embedding weight adjustment coefficients. Watermark loss term. The smooth-L1 loss is defined as follows:

[0187] ;

[0188] in, Representing the sigmoid function:

[0189] .

[0190] After optimization, the device-side components use the system public key to configure the interface parameters for the current round. Perform fully homomorphic encryption; the ciphertext format is:

[0191] ;

[0192] in, From Random values ​​sampled from the middle, , From Noise from random sampling.

[0193] Through the aforementioned joint optimization mechanism, this invention effectively achieves coordinated processing of watermark embedding and functional optimization without significantly affecting the performance of interface protocol behavior. This enhances the anti-interference and verifiability of the interface implementation of ownership watermarks, while protecting the privacy of interface data of device-side components.

[0194] Step 4) The interface holder and the device component extract the joint watermark from the joint interface protocol configuration:

[0195] Verification of initiator and interface holder pre-generation Group Triple vector ,in And satisfy The verification initiator embeds the watermark into the matrix. Divided into two additive secret shared shares and and will Send to the interface holder; the interface holder will configure the parameters. Split into and and will Send to the verification initiator; both parties utilize , , , Triple vector Joint computing and The product result.

[0196] In this embodiment, using , , , Triple vector Joint computing and The product result is calculated using Algorithm 1:

[0197]

[0198] in The step function is calculated as follows:

[0199] .

[0200] Step 5) Ownership of the interface holder's encrypted computation joint interface implementation:

[0201] The interface holder obtains the joint watermark from the extraction. Restore binding in the interface configuration parameters of the joint identity matrix eigenvalues ; Interface holding direction Request dense state eigenvalues It then verifies the validity of the encrypted feature value. Once valid, it utilizes the properties of homomorphic encryption to calculate the encrypted joint identity bound to the configuration. The interface holder calculates the encrypted joint identity extracted from the interface configuration. With the publicly disclosed encrypted joint identity of the verification initiator Dense Euclidean distance .

[0202] In this example, the encrypted joint identity extracted from the interface configuration is calculated. With the publicly disclosed encrypted joint identity of the verification initiator The dense-state Euclidean distance is calculated as follows:

[0203] ;

[0204] in, Representing vectors with vector Subtract the elements at corresponding positions. Representing vectors The One element. After obtaining the result of the subtraction, multiply it by itself:

[0205] ;

[0206] At this point, by combining the rotation operation of the CKKS algorithm, it is possible to achieve [the desired result] in dense state. The calculation. Specifically, the rotation operation can shift the first element of the CKKS ciphertext vector to the last position:

[0207] ;

[0208] set up Then rotate the result of the self-multiplication once and then combine it with... Adding them together, we get:

[0209] ;

[0210] Repeated rotation and addition operations Next, compare the results with... Long ciphertext vector Multiplication:

[0211] ;

[0212] Thus, the dense state is obtained and The similarity.

[0213] Step 6) The interface holder obtains the ownership verification result of the joint interface implementation:

[0214] The interface holder sends the dense Euclidean distance to and the verification initiator; The initiator and the verification provider each use a portion of their private keys. and Decryption yields partial decrypted ciphertext. and Interface holder collects and And completely decrypted and The Euclidean distance, take The first item and the preset intellectual property verification threshold Compare, if If the interface configuration is deemed to belong to the verification initiator, the intellectual property verification is successful. The interface configuration ownership verification failed.

[0215] In this example, the interface control unit The initiator and the verification provider each use a portion of their private keys. and Encryption distance Partial decryption is performed; the partial decryption formula is as follows:

[0216] ;

[0217] in, and They are respectively the encryption distance Part 1 and Part 2.

[0218] In this embodiment, the interface holder controls the interface control unit. Decryption distance And verify the initiator's partial decryption results Perform full decryption; the full decryption formula is:

[0219] ;

[0220] in, For ciphertext Part of it.

[0221] Accordingly, please refer to Figure 2 A second aspect of this invention provides a watermark-based universal multimedia interface protocol ownership detection device, which detects the ownership of a universal multimedia interface implementation using the aforementioned watermark-based universal multimedia interface protocol ownership detection method, including:

[0222] Watermark extraction module 1 is used to extract a joint watermark from the GPMI interface configuration parameters to be verified through secure multi-party computation;

[0223] Feature extraction module 2, which is used to recover the feature values ​​of the joint identity matrix based on the extracted joint watermark;

[0224] Distance calculation module 3 is used to calculate the dense Euclidean distance between the feature value and the dense feature vector provided by the verifier using homomorphic encryption technology;

[0225] The ownership determination module 4 is used to obtain the decrypted Euclidean distance through the outsourced decryption mechanism of double trapdoor homomorphic encryption, and determine the ownership of the interface intellectual property rights based on the comparison result of the Euclidean distance and the preset threshold.

[0226] Accordingly, a third aspect of the present invention provides an electronic device, including: at least one processor and a memory connected to the at least one processor. The memory stores instructions executable by the at least one processor, which, when executed by the at least one processor, cause the at least one processor to perform the aforementioned watermark-based universal multimedia interface protocol ownership detection method.

[0227] Accordingly, a fourth aspect of the present invention provides a computer-readable storage medium having computer instructions stored thereon, which, when executed by a processor, implement the above-described watermark-based general multimedia interface protocol ownership detection method.

[0228] The embodiments of the present invention aim to protect a watermark-based method and apparatus for detecting the ownership of a universal multimedia interface protocol, which has the following effects:

[0229] 1. The attributes and identity information of device components are merged into a joint identity. A watermark is generated through feature value transformation and embedded into the interface protocol configuration as an encrypted credential for ownership verification. During ownership confirmation, the mathematical mapping relationship between the joint identity and the watermark can be used to recover the identity information from the interface configuration parameters to determine ownership, realizing ownership declaration based on identity. At the same time, the watermark embedding regularization term is introduced to achieve ownership binding with minimal impact on interface configuration performance, providing high security and high authentication efficiency for intellectual property verification, and significantly improving the authority and credibility of the ownership confirmation mechanism.

[0230] 2. Secure extraction of watermark information is achieved through secure two-party computation, ensuring the privacy and data security of participating parties; fully homomorphic encryption technology is used to complete the encrypted matching of device identities, effectively reducing the risk of identity leakage; and a double trapdoor fully homomorphic encryption algorithm is used to support outsourced decryption by the trusted center in offline mode, ensuring the security and trustworthiness of the data of both parties in the entire rights confirmation process, and providing a more robust and tamper-resistant intellectual property protection mechanism for the configuration of the joint interface protocol.

[0231] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0232] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0233] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0234] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0235] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit it. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that modifications or equivalent substitutions can still be made to the specific implementation of the present invention. Any modifications or equivalent substitutions that do not depart from the spirit and scope of the present invention should be covered within the protection scope of the claims of the present invention.

Claims

1. A watermark-based method for detecting ownership of a universal multimedia interface protocol, characterized in that, The interface holder, based on the general multimedia interface protocol, holds the configuration parameters of the interface to be verified and performs a full-process ownership detection and verification operation. The detection method includes the following steps: Extract the joint watermark from the GPMI interface configuration parameters to be verified through secure multi-party computation; The eigenvalues ​​of the joint identity matrix are recovered based on the extracted joint watermark; Using homomorphic encryption technology, the dense-state Euclidean distance between the feature value and the dense-state feature vector provided by the verifier is calculated; The decrypted Euclidean distance is obtained through an outsourced decryption mechanism using double trapdoor homomorphic encryption. The ownership of the interface intellectual property rights is determined based on the comparison result between the Euclidean distance and a preset threshold. The extraction of the joint watermark from the GPMI interface configuration parameters to be verified through secure multi-party computation includes: The interface holder receives the second share of the watermark embedding matrix sent by the verification initiator. At the same time, the interface holder splits the interface configuration parameters to be verified into two components, the first parameter share and the second parameter share, which are equal to the original parameters. The first parameter share is then sent to the verification initiator. The verification initiator calls multiple pre-generated verification triples, each consisting of three elements and satisfying a multiplicative secret sharing relationship. Based on their respective matrix shares and parameter shares, they perform vector multiplication operations row by row to jointly calculate the intermediate value of the product of the watermark embedding matrix and the interface configuration parameters. During the joint calculation process, the interface holder and the verification initiator do not exchange the original matrix shares and the original parameter shares. The intermediate value of the product is input into a step function, which is defined as outputting a first state value when the input value is greater than or equal to zero and outputting a second state value when it is less than zero, thereby generating a joint watermark that represents the device's identity features. The feature values ​​of the extracted joint identity matrix recovered from the joint watermark include: The binary sequence of the joint watermark is arranged from the most significant bit to the least significant bit. Each watermark value is multiplied by its corresponding bit weight and then summed to obtain a decimal numerical result. The numerical results are used as the main feature values ​​of the joint identity matrix, which is generated by mapping the identity identifiers and attribute sets of the device components through a tree structure. The trace eigenvalues ​​of the joint identity matrix are derived from the principal eigenvalues, and the trace eigenvalues ​​are mathematical eigenvalues ​​of the sum of the diagonal elements of the joint identity matrix. The trace feature value is output as a feature value parameter for ownership verification.

2. The watermark-based universal multimedia interface protocol ownership detection method according to claim 1, characterized in that, The numerical result is used as the main feature value of the joint identity matrix, which is generated by mapping the identity identifiers and attribute sets of device components through a tree structure, including: Obtain the unique identifiers and attribute sets of all participating device components, the attribute sets including device type, manufacturer code, and interface protocol version number; Based on the hierarchical relationship between the identity identifiers and attribute sets, a tree-shaped data structure is constructed: the root node is set as the system authority identifier, the second-level nodes are the device type classifications, the third-level nodes are attached with specific device identity identifiers, and the leaf nodes are bound to attribute key-value pairs; The tree-like data structure is mapped to a numerical matrix, a unique matrix index is assigned to each node, and the parent-child node relationship is transformed into the association weight value at the intersection of the matrix rows and columns, generating a joint identity matrix that reflects the device identity association. The numerical result is used as the core eigenvalue of the joint identity matrix, which represents the scaling factor of the joint identity matrix in the direction of the largest eigenvector.

3. The watermark-based universal multimedia interface protocol ownership detection method according to claim 1, characterized in that, The step of using homomorphic encryption to calculate the encrypted Euclidean distance between the feature value and the encrypted feature vector provided by the verifier includes: In the homomorphic encryption state, the trace feature value to be verified is subtracted from the encrypted feature vector provided by the verifier according to the element position to obtain the encryption difference vector; Perform a homomorphic self-multiplication operation on each element of the encrypted difference vector to generate an encrypted square vector; The encrypted square vector is processed by a preset number of vector rotation and accumulation operations. Each rotation and accumulation operation cyclically shifts the vector element by one position and adds it element by element to the vector before the shift, finally generating an encrypted distance vector in which only the first element is non-zero. The first element of the encrypted distance vector is extracted as the encrypted Euclidean distance representing the joint identity similarity.

4. The watermark-based universal multimedia interface protocol ownership detection method according to claim 3, characterized in that, The method of obtaining the decrypted Euclidean distance through the outsourced decryption mechanism of double trapdoor homomorphic encryption includes: The encrypted Euclidean distance is sent to the first decryptor and the second decryptor. The first decryptor holds the first homomorphic encryption partial private key, and the second decryptor holds the second homomorphic encryption partial private key. The partial private key is generated by randomly sampling and splitting the system master private key. Receive a first intermediate decryption result generated by the first decryptor performing a partial decryption operation on the Euclidean distance of the encrypted state using a portion of its private key, and receive a second intermediate decryption result generated by the second decryptor performing a partial decryption operation using a portion of its private key; By aggregating the first and second intermediate decryption results, a full decryption operation is performed to obtain a plaintext vector containing Euclidean distance, wherein only the first element of the plaintext vector is a non-zero value; The first element of the plaintext vector is extracted as the actual Euclidean distance value.

5. The watermark-based universal multimedia interface protocol ownership detection method according to claim 4, characterized in that, The interface holder aggregates the first and second intermediate decryption results and performs a full decryption operation to obtain a plaintext vector containing Euclidean distance, including: Perform an arithmetic summation operation on the received first and second intermediate decryption results to generate an aggregated intermediate decryption value; Extract a preset component from the original ciphertext of the dense-state Euclidean distance, wherein the preset component is a fixed data block generated during the encryption process based on the homomorphic encryption ring structure; The preset component is subtracted from the aggregated decryption intermediate value to obtain the decrypted plaintext vector, in which all elements except the first element are zero. Verify whether the numerical range of the first element is within the valid Euclidean distance interval. If valid, output the plaintext vector as the final decryption result.

6. The watermark-based universal multimedia interface protocol ownership detection method according to claim 1, characterized in that, The step of determining the ownership of interface intellectual property rights based on the comparison result of the Euclidean distance and the preset threshold includes: The actual Euclidean distance value is compared with a preset intellectual property verification threshold value, which is a preset value based on the interface security level. When the actual Euclidean distance value is less than or equal to the intellectual property verification threshold, an ownership confirmation certificate containing a verification timestamp and device identity information is generated, and it is determined that the interface configuration belongs entirely to the verification initiator. When the actual Euclidean distance value is greater than the intellectual property verification threshold, a request for ownership tracing is initiated to an authoritative institution to carry out a multi-party collaborative process for solidifying infringement evidence.

7. The watermark-based universal multimedia interface protocol ownership detection method according to claim 6, characterized in that, The generation of the ownership confirmation credential, which includes a verification timestamp and device identity information, includes: The device identity identifier of the verification initiator, the actual Euclidean distance value, and the interface configuration parameter summary are combined into the core data of the ownership claim. Request an authoritative institution to attach a tamper-proof verification label and a time stamp to the core data. The tamper-proof verification label is generated using an asymmetric cryptographic algorithm and can be publicly verified. The time stamp is issued by a trusted time source. The core data, including the tamper-proof verification tag and time stamp, is encapsulated into a structured electronic certificate. The structured electronic certificate contains a machine-readable ownership declaration field and a verification metadata field.

8. A watermark-based universal multimedia interface protocol ownership detection device, characterized in that, The method for detecting the ownership of a general multimedia interface implementation using the watermark-based general multimedia interface protocol ownership detection method as described in any one of claims 1-7 includes: The watermark extraction module is used to extract a joint watermark from the configuration parameters of the GPMI interface to be verified through secure multi-party computation. The feature extraction module is used to recover the feature values ​​of the joint identity matrix based on the extracted joint watermark; The distance calculation module is used to calculate the dense Euclidean distance between the feature value and the dense feature vector provided by the verifier using homomorphic encryption technology. The ownership determination module is used to obtain the decrypted Euclidean distance through the outsourced decryption mechanism of double trapdoor homomorphic encryption, and determine the ownership of the interface intellectual property rights based on the comparison result of the Euclidean distance and the preset threshold.

Images