Enterprise financial data sharing security processing method and system based on privacy computing

Abstract

The invention discloses an enterprise financial data sharing security processing method and system based on privacy computing, and relates to the technical field of data security and privacy protection, and the method comprises the following steps: S100, constructing a space-time mapping structure of watermark propagation according to the time sequence and node topology information of an enterprise financial data return link, and constructing a space-time mapping structure of watermark propagation; extracting an initial position of phase deviation, and generating a phase anchor point list; and S200, based on the phase anchor point list, identifying overlapping conflict areas between the paths before data aggregation, establishing a conflict window, and extracting phase drift trajectories of the paths to form trajectory records. According to the method, the watermark propagation space-time mapping and the hierarchical alignment index are constructed, so that the watermark rhythm guidance under path drifting is realized, and the data identification continuity is improved; and through combination of peak shifting embedding and phase locking adjustment mechanisms, a watermark structure is unified, signal interference is avoided, and the stability of a traceability chain is ensured.

Description

Technical Field

[0001] This invention relates to the field of data security and privacy protection technology, specifically to a method and system for securely processing shared enterprise financial data based on privacy computing. Background Technology

[0002] Privacy-based computation-driven secure sharing of enterprise financial data refers to a secure approach that enables participating parties to complete joint computations and share results without directly exposing the original data content when multiple enterprises or institutions need to analyze, compare, or collaboratively compute financial data. This method typically combines homomorphic encryption, secure multi-party computation, federated learning, and differential privacy technologies to ensure that sensitive financial data participates in computation while encrypted or anonymized, guaranteeing that the data remains invisible and undisclosed throughout the entire transmission, storage, and computation process. Building upon this, a distributed computing framework for big data processing, parallel task scheduling, and massive data sharding and aggregation mechanisms are introduced. This allows multi-source, heterogeneous, and high-dimensional enterprise financial data to be efficiently aggregated, jointly modeled, and batch analyzed under privacy-controlled conditions, thus balancing computational performance and security boundaries. Through the synergy of privacy-based computation and big data processing capabilities, this approach supports large-scale secure sharing and in-depth collaborative analysis of cross-enterprise financial data while meeting the stringent requirements of enterprises for trade secret protection, audit compliance, and regulatory traceability, achieving the sharing goals of data usability without visibility, controllable computation processes, and verifiable and traceable analysis results.

[0003] The existing technology has the following shortcomings: In existing technologies, dynamic traceability monitoring of enterprise financial data sharing processes typically involves embedding digital watermarks into the data stream to track the data source and flow. However, when financial data is transmitted through multiple nodes, paths, and loops within a multi-party privacy computing framework, existing digital watermark embedding mechanisms generally employ static positions or fixed frequency domain markings, making it difficult to adapt to dynamic changes in the data flow path. As data is frequently forwarded and re-encrypted between different computing nodes, the adaptive adjustments made by the system to maintain the stability of the identifier often cause phase drift in the watermark embedding position. When data from multiple paths is retransmitted and superimposed at the convergence node, the drifted watermark will generate phase superposition interference in the frequency domain, causing some identifier signals to cancel each other out or become distorted. Once such implicit interference occurs, it will directly cause a break in the data traceability chain, making it impossible to accurately extract the responsibility identifier, thereby causing the entire data responsibility tracking mechanism to fail. When facing data leakage or abnormal sharing events, the system will be unable to trace back to a specific node or entity, seriously affecting data security and trustworthiness in a privacy computing environment.

[0004] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0005] The purpose of this invention is to provide a secure processing method and system for sharing enterprise financial data based on privacy computing, so as to solve the problems in the background art mentioned above.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a secure processing method for sharing enterprise financial data based on privacy computing, comprising the following steps: S100 constructs a spatiotemporal mapping structure for watermark propagation based on the time sequence and node topology information of the enterprise financial data backhaul link, and extracts the starting position of phase offset to generate a list of phase anchor points. S200, based on the phase anchor point list, identifies overlapping and conflicting areas between paths before data aggregation, establishes conflict windows, and extracts the phase drift trajectory of each path to form a trajectory record. S300, based on phase drift trajectory recording, generates alignment guide texture within the watermark carrier, sets an independent start scale for each data return path, and constructs a hierarchical alignment index; S400 performs a frequency division and layered staggered peak embedding operation based on the layered alignment index, redistributing the watermark signal in the overlapping conflict area to the adjacent time window to generate a staggered peak embedding time spectrum. The S500, based on the staggered embedding time spectrum, injects an anti-phase sentinel watermark during data backhaul aggregation, performs echo phase-locked signal adjustment, and dynamically converges the phase through a rhythm scaling control mechanism to maintain the continuity and stability of the traceability link.

[0007] Preferably, step S100 includes: Based on the collaborative path of enterprise financial data in privacy computing scenarios, the data flow links participating in joint computing are organized at the node level, a data backhaul link topology covering the entire process is constructed, and the connection relationship and path weight of each node are marked. Based on the link topology, the time sequence information of enterprise financial data during transmission is extracted, and a time sequence chain with hierarchical relationship is constructed to mark the position of data segments between nodes on the global time axis. By combining the node topology and the time sequence chain, a spatiotemporal mapping structure for watermark propagation is established, and the watermark propagation trajectory is mapped to the spatial node position and the time node sequence to form a three-dimensional spatiotemporal mapping surface. Based on the spatiotemporal mapping structure, the propagation characteristics of the watermark in each path are analyzed, the starting position of the phase shift of the watermark signal is marked, a list of phase anchor points is generated, and the path identifier, offset node and corresponding time coordinate information are recorded.

[0008] Preferably, when generating the phase anchor point list, the starting offset position of the watermark signal is synchronously associated with the spatial coordinates in the node topology and the time nodes in the time sequence chain. The initial phase change point is determined by tracing back the path propagation process, and the anchor point correspondence is established based on the propagation direction information, so as to achieve accurate positioning of the phase offset position and unified identification of the anchor point data.

[0009] Preferably, step S200 includes: Based on the generated spatiotemporal mapping structure and phase anchor point list of watermark propagation, we conduct centralized analysis on multiple backhaul paths of enterprise financial data, identify the intersection nodes of each path in the data flow convergence process, and establish a node correspondence mapping. Based on the identified path intersection nodes, combined with the offset starting point of each path in the phase anchor point list, potential conflict sections between paths are identified, and the time boundary and spatial span of the overlapping sections are determined. Based on the identified potential conflict zones, a conflict window structure is constructed around the path combinations with intersecting relationships. Continuous time segments are established by combining the time dimension and path distribution, and the path participation relationship and overlap density level are marked. Based on the constructed conflict window, the phase drift trajectory of each path from the phase anchor point to the intersection node is extracted, and the path start time, node position, drift amplitude and change direction are recorded to form a trajectory record.

[0010] Preferably, when constructing the conflict window structure, the path is divided into continuous time segments on the time axis. By classifying the degree of overlap of watermark signals within the time segments, the overlap density level and interference direction are determined. A joint time and space mapping is established based on the differences in path distribution to limit the conflict range and interference boundary of watermark signals before data convergence.

[0011] Preferably, step S300 includes: Based on the phase evolution trend and jump node information of each data path in the phase drift trajectory record, a guiding reference structure is constructed in the watermark carrier to map the node information, time span and phase change amplitude of each path to a unified reference interval. By combining the phase drift trajectory data of each path, a path-specific guide texture is generated in the guide reference structure to ensure that the guide texture is consistent with the phase drift rhythm of the path and to avoid overlapping areas between textures. For each generated guide texture, an independent starting scale is set based on the anchor point time position of the path to determine the initial embedding time point of the watermark signal, and a time identifier signal and path number label are embedded in the carrier. After all paths have been set with independent starting scales, the guide textures and starting scales of each path are combined vertically to construct a hierarchical alignment index, which is used to guide the differentiated embedding control of the watermark signal before convergence.

[0012] Preferably, the hierarchical structure of the bit index is divided into levels according to path priority and embedding order. Each level corresponds to a specific embedding channel. The paths start sequentially in time, are independently distributed in space, and run interleaved in the frequency domain, so that the watermark signal maintains a continuous rhythm before convergence and avoids embedding interference.

[0013] Preferably, step S400 includes: Based on the hierarchical structure, starting scale and guide texture distribution information in the hierarchical alignment index, high-density conflict areas with overlapping trends in the watermark embedding process are identified, and the conflict areas are hierarchically labeled. Around the identified conflict areas, analyze the time buffer segments in adjacent time segments that can be redistributed for watermarks, construct a time buffer structure containing multiple micro-time windows, and record the time distance and path compatibility between the buffer segments and the conflict areas. Based on the time buffer structure, a reallocation operation is performed on the watermark embedding points located in the conflict area, moving the embedding points to the corresponding buffer segment, and dynamically fine-tuning the path start scale to maintain rhythm consistency. After the embedding points are redistributed, the embedding time points, guiding texture segments and hierarchical information of each path are merged to generate a staggered embedding time spectrum that covers the watermark embedding behavior of the entire path, which is used to guide subsequent embedding control.

[0014] Preferably, step S500 includes: Based on the staggered embedding time spectrum and the convergence time of each data path, a unified control interval suitable as the starting point for watermark aggregation is identified, and the time boundary and rhythm buffer zone of the aggregation interval are determined. Within the convergence control range, guided by the path phase difference, an anti-phase sentinel watermark is injected, causing the watermark signal to exhibit a rhythmic convergence tendency in the time dimension, while avoiding interference with the original high-density watermark segment. After the sentinel watermark is injected, an echo lock signal is applied to the path watermark to make the path watermark signals resonate periodically in phase rhythm and form a phase alignment trend at the end of the convergence interval. After echo lock-in modulation, a rhythm stretching control mechanism is introduced to slightly stretch or compress the watermark rhythm at the end of the path, so that all path watermark signals can complete rhythm convergence and maintain phase consistency before the data convergence point.

[0015] A privacy-preserving computing-based enterprise financial data sharing and secure processing system includes a watermark spatiotemporal mapping construction module, a conflict detection and trajectory analysis module, an alignment-guided index generation module, a peak-shifting embedding scheduling module, and a phase-locked loop control and echo correction module. The watermark spatiotemporal mapping construction module constructs a spatiotemporal mapping structure for watermark propagation based on the time sequence and node topology information of the enterprise financial data backhaul link, and extracts the starting position of phase offset to generate a list of phase anchor points. The collision detection and trajectory analysis module, based on the phase anchor point list, identifies overlapping collision areas between paths before data aggregation, establishes collision windows, and extracts the phase drift trajectory of each path to form a trajectory record. The alignment guide index generation module generates alignment guide textures within the watermark carrier based on phase drift trajectory records, sets independent start-up scales for each data return path, and constructs a hierarchical alignment index. The staggered embedding scheduling module performs frequency division and hierarchical staggered embedding operation based on the hierarchical alignment index, redistributing the watermark signal in the overlapping conflict area to the adjacent time window to generate the staggered embedding time spectrum. The phase-locked loop (PLL) and echo correction module injects an anti-phase sentinel watermark during data backhaul aggregation based on the staggered embedding time spectrum, performs echo phase-locked loop signal adjustment, and dynamically converges the phase through a rhythm scaling control mechanism to maintain the continuity and stability of the traceability link.

[0016] The technical effects and advantages provided by the present invention in the above technical solution are as follows: This invention effectively addresses the path drift and phase misalignment issues arising from multi-path, multi-node financial data during privacy computation by constructing a spatiotemporal mapping structure and hierarchical alignment index for watermark propagation. By precisely extracting phase anchor points and drift trajectories, it dynamically guides the watermark embedding position and rhythm, ensuring the watermark signal maintains structural coherence and rhythmic stability throughout complex data backhaul links. This improves the tracking continuity of data identifiers and prevents identifier failure due to signal drift.

[0017] This invention employs a staggered embedding and echo-locked loop adjustment mechanism to proactively absorb rhythm differences and converge phases before the watermark signals converge, ensuring the uniformity of the watermark structure and the clarity of recognition at the final node. By dynamically adjusting the embedding time spectrum and rhythmic scaling operations, the watermark signals achieve a rhythmic coordination state during the data aggregation stage, effectively avoiding cancellation or interference problems caused by signal superposition, thereby maintaining the integrity of the data tracking chain and the stability of watermark recognition. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0019] Figure 1 This is a flowchart of the method for secure processing of enterprise financial data sharing based on privacy computing according to the present invention.

[0020] Figure 2 This is a schematic diagram of the modules of the enterprise financial data sharing and secure processing system based on privacy computing according to the present invention. Detailed Implementation

[0021] Exemplary embodiments will now be described more fully with reference to the accompanying drawings. However, these exemplary embodiments can be implemented in many forms and should not be construed as limited to the examples set forth herein; rather, they are provided so that the description of this disclosure will be more complete and fully convey the concept of the exemplary embodiments to those skilled in the art.

[0022] This invention provides, for example Figure 1 The privacy-preserving computation-based secure processing method for sharing enterprise financial data includes the following steps: S100 constructs a spatiotemporal mapping structure for watermark propagation based on the time sequence and node topology information of the enterprise financial data backhaul link, and extracts the starting position of phase offset to generate a list of phase anchor points. This step proposes a processing method that constructs a spatiotemporal mapping structure for watermark propagation based on the time sequence and node topology information of the enterprise financial data backhaul link, and extracts the starting position of the phase offset to generate a list of phase anchor points. This processing method mainly includes the following steps: Based on the actual collaborative paths of enterprise financial data in privacy-preserving computing scenarios, the data flow links participating in joint computing are organized at the node level, constructing a data backhaul link topology covering the entire process. Specifically, in the data exchange process among multiple participants, each node represents a data processing unit, which can be an initial data provider, a computational intermediate node, or a result receiver. By tracing the complete path of data from the source node, to multiple intermediate nodes, and finally to the return of the final result, the connection relationships of each jump node and its upstream and downstream dependencies are marked, forming a complete node topology graph. During the construction of the topology graph, the path weight of each connection relationship also needs to be marked. This weight can represent data jump complexity, data encryption layers, or bandwidth usage, reflecting the differences in transmission load brought by different paths to watermark propagation. On this basis, a clear link topology layout is formed, providing structural support for subsequent joint modeling of time and path.

[0023] Based on the link topology, the temporal sequence information of enterprise financial data during transmission is further extracted. The temporal information for each link should be archived based on the processing order of data at different nodes, transmission delays, and computation times. After obtaining the actual transmission delay data between nodes, a time series model is constructed sequentially to ensure that the position of the data segment processed by each node is accurately marked on the global time axis. Especially for data streams with multiple backhauls in complex paths, it is necessary to clarify their initial entry into the path, the length of time spent waiting for processing at intermediate nodes, and the dwell time during each jump. Through continuous analysis of the time series between nodes, a hierarchical temporal sequence chain can be constructed. This time chain will serve as the basis for analyzing the watermark phase change trend, providing dynamic temporal dimension input conditions for subsequently establishing the propagation mapping structure.

[0024] With the node topology and temporal sequence chain already constructed, a spatiotemporal mapping structure for watermark propagation is established by combining the information from both. The core of this mapping structure lies in mapping the dynamic trajectory of the watermark during data propagation to spatial node positions and temporal node sequences in a one-to-one correspondence. A three-dimensional spatiotemporal mapping surface is formed by superimposing the jump path of each data packet between nodes with its transmission time. This mapping surface describes the phase shift path of the watermark information during data processing caused by operations such as encryption / decryption, encoding conversion, and format deformation, thereby capturing the cumulative change trend of the watermark signal in each jump process. Especially for nodes where multiple paths intersect, the mutual influence and overlap between different paths should be considered when constructing the mapping structure to ensure that the watermark signal change trajectory corresponding to each path can be clearly represented in the overall structure. Through a complete spatiotemporal mapping structure, the watermark propagation trajectory becomes controllable and predictable, providing dual spatial and temporal reference coordinates for subsequent anchor point positioning.

[0025] Based on the constructed spatiotemporal mapping structure, the propagation characteristics of the watermark in each data path are analyzed, and the starting position of the phase shift of the watermark signal on each path is marked, generating a phase anchor point list accordingly. Specifically, by tracing back along the time dimension on each path, the position where the watermark first undergoes a phase change is identified, which is marked as the starting offset point of that path. This starting offset position usually occurs when the data undergoes its first encryption conversion or format adjustment operation across nodes, so the accurate position can be determined by combining the node characteristics and time nodes in the mapping structure. Subsequently, based on the starting offset point on each path, its corresponding time node, spatial location, node number, and other information are uniformly summarized to form an anchor point list. The anchor point item in each record should include the path identifier, the node where the offset occurred, the corresponding time coordinate, and propagation direction information. This anchor point list will serve as the basic data for subsequent path drift trajectory alignment and conflict prediction, supporting the construction of alignment guidance textures and the execution of staggered embedding strategies. By standardizing and organizing the anchor point list, the consistency and controllability of watermark identification in multi-path data in complex transmission environments are further improved.

[0026] S200, based on the phase anchor point list, identifies overlapping and conflicting areas between paths before data aggregation, establishes conflict windows, and extracts the phase drift trajectory of each path to form a trajectory record. This step proposes a processing method based on a phase anchor point list. Before data aggregation, it identifies overlapping and conflicting regions between paths, establishes conflict windows, and extracts the phase drift trajectories of each path to form trajectory records. This method builds upon the previously constructed spatiotemporal mapping structure for watermark propagation and the phase anchor point list, further enhancing the ability to identify the changing patterns of watermark signals during the transmission of multi-path data in privacy-preserving computing scenarios. To implement this processing, the technical solution includes the following steps: Based on the generated spatiotemporal mapping structure and phase anchor point list of watermark propagation, a centralized analysis is performed on multiple backhaul paths of enterprise financial data to identify the intersection nodes of each path during data flow convergence. Data path intersection refers to the area where two or more data paths are temporally close, spatially overlapping, or logically exhibiting a merging trend. These intersection nodes are typically located in intermediate processing units before the joint computing platform or result aggregation node, and are high-risk areas for watermark signal overlap or interference. In identifying intersection nodes, each data flow in the watermark propagation path needs to be compared, and its propagation endpoint needs to be checked for temporal and spatial overlap with other paths. Based on this, a node correspondence mapping is established. The goal of this step is to identify all possible path combinations where watermarks may overlap, laying the foundation for subsequent judgment of conflict zones.

[0027] Based on the identified path intersection nodes and the offset start point of each path in the phase anchor point list, it is determined whether there are potential conflict sections between paths. In this step, the propagation time and time offset amplitude between any two paths from their anchor points to the intersection node should be calculated using the time coordinates recorded in the anchor point list as a reference. If the difference in propagation time between two paths is within a set range and their anchor point offset directions are similar, then a conflict is considered possible. At this point, the data carrying density, watermark embedding frequency, and phase change trend within this time period should be further analyzed to determine the specific boundaries of the overlapping sections. Such overlapping sections are defined as potential conflict sections and are candidate intervals for constructing conflict windows. During the construction process, the path identification method must be unified to ensure that the time length, spatial span, and path sequence of all potential conflict sections correspond one-to-one, maintaining consistency in the processing flow.

[0028] After identifying all potential conflict zones, a corresponding conflict window structure is constructed for each group of intersecting paths. A conflict window refers to an area within a timeframe before data convergence where there is a risk of watermark overlap between paths. Constructing conflict windows requires considering both the temporal dimension and path distribution. Firstly, a unified time frame is established based on the propagation segment of each path from its anchor point to the intersection node, dividing it into multiple consecutive time segments and labeling the path participation relationships within each time segment. Secondly, in the spatial dimension, the data-carrying structure of each path is arranged and combined to map the possible overlapping areas and order of watermark signals within each time segment. Ultimately, each conflict window includes not only path combination information and time segmentation information, but also the predicted overlap density level and relative interference direction to describe the cross-over and overlap situation. This structure will provide a basis for conflict zone location for subsequent alignment guidance.

[0029] Based on the established conflict window structure, the phase drift trajectory of each path from the phase anchor point to the intersection node is further extracted to form a complete trajectory record. The trajectory record is used to depict the dynamic phase evolution of the watermark signal during data propagation. The extraction method needs to combine the continuous change trend of each path on the time and space axes in the watermark propagation spatiotemporal mapping structure, accumulating the phase changes caused by each jump node on the path to form a continuous phase drift curve. When constructing the trajectory record, each path should be assigned a unique number, marking key parameters such as the path's starting anchor point time, the position of each jump node, the phase drift amplitude, the phase change direction, and the rate of change. These parameters will serve as the basic dataset for subsequent generation of alignment guidance textures and formulation of staggered embedding strategies. Through the systematic organization of the trajectory records, the phase change pattern and evolution law of each path during propagation can be clearly understood, thus enabling differentiated guidance preparation in advance when facing complex path superposition, avoiding watermark cancellation or distortion caused by excessive trajectory overlap.

[0030] S300, based on phase drift trajectory recording, generates alignment guide texture within the watermark carrier, sets an independent start scale for each data return path, and constructs a hierarchical alignment index; This step proposes a processing method based on phase drift trajectory recording to generate alignment guidance textures within the watermark carrier. Furthermore, by setting independent start-point scales for each data return path, a hierarchical alignment index is constructed to achieve differentiated embedding control of the watermark signal in multi-path convergence scenarios. The specific steps are as follows: Based on the phase evolution trend and jump node information of each data path contained in the phase drift trajectory record, a guiding reference structure is constructed in the watermark carrier to guide the watermark embedding process. This structure is established using both temporal order and path numbering as dual dimensions, mapping the node information, time span, and phase drift change amplitude experienced by all paths during propagation to a unified watermark embedding reference interval. Each path occupies an independent calibration channel within this interval to record its phase change behavior from the anchor point to the point of intersection. During construction, it is crucial to ensure that the arrangement order of each path in the guiding reference structure is consistent with its temporal order during actual propagation, preventing embedding overlap caused by path location intersections. Simultaneously, to avoid overlapping identifiers in high-density path segments, equally spaced separation segments can be set between adjacent paths on the time axis, thus forming a temporally separated guiding reference framework, providing clear structural boundaries for subsequent texture generation.

[0031] Based on the constructed guidance reference structure, and combined with the phase drift trajectory data of each path, a unique guidance texture is generated for each path within the watermark carrier. The guidance texture is a signal structure formed within the watermark embedding carrier through nesting or superposition, used to characterize the rhythmic guidance pattern of a specific path during the embedding process. During generation, characteristic parameters such as amplitude changes, directional trends, number of inflection points, and change density in the phase drift trajectory should be used as the basis for texture morphology construction, ensuring that the guidance texture corresponding to each path possesses specificity, distinctiveness, and trajectory consistency. The shape of the texture can exhibit periodic undulations, linear gradations, or a jump-type structure. Its inherent regularity must be consistent with the phase drift rhythm of the path during propagation, forming a guidance channel embedded within the watermark carrier, serving as a positioning reference for subsequent watermark signal embedding. Simultaneously, overlapping areas should be avoided between guidance textures, especially during high-risk periods corresponding to conflict windows. The boundary clarity between textures should be enhanced, and the risk of overlap interference should be reduced through spatial isolation or texture frequency adjustment.

[0032] For each generated guide texture, an independent start-up scale is set based on the anchor point time position during the propagation process to determine the initial embedding time point of the watermark signal. Setting the start-up scale is crucial for achieving multi-path staggered embedding control. By staggering the embedding start times of different paths, multiple watermark signals from different paths are prevented from entering the embedding process simultaneously, thus reducing the probability of signal superposition within the conflict window. The specific position of the start-up scale can be calculated comprehensively based on the path propagation delay, anchor point time position, and guide texture density, ensuring that the watermark of an earlier-arriving path enters the embedding channel earlier, while the watermark of a later-arriving path starts the embedding process at a later time point. To enhance the recognition capability of the start-up scale, a weak time marker signal and path number label can be embedded within the carrier to mark the start reference lines of different paths, thereby achieving dual positioning in both physical space and logical time axis.

[0033] After all paths have completed independent start-up scale settings, the guide textures and start-up scales of each path are vertically combined to construct a hierarchical alignment index to support differentiated embedding control of watermark signals before convergence. The hierarchical structure can be divided according to path priority, embedding order, or phase change complexity. Each level corresponds to a specific set of embedding channels, ensuring that paths start at off-peak times, are spatially independent, and operate alternately in the frequency domain, thus forming a watermark guidance framework with triple alignment capabilities of temporal structure, frequency domain structure, and path structure. The generation of the alignment index requires the unified collection of start-up scale calibration information, guide texture spatial distribution information, and path number mapping information for each path, organized in an embedding control reference table. This allows subsequent watermark signals to be indexed according to the graph and embedded in an orderly manner within predetermined paths and time nodes. This index table can also be linked with the conflict window structure constructed in the previous stage to dynamically adjust the embedding strategy parameters to adapt to real-time conflict states between paths.

[0034] S400 performs a frequency division and layered staggered peak embedding operation based on the layered alignment index, redistributing the watermark signal in the overlapping conflict area to the adjacent time window to generate a staggered peak embedding time spectrum. This step proposes a technique for performing frequency-division hierarchical staggered embedding operations based on a hierarchical alignment index. By redistributing watermark signals from overlapping conflict areas to adjacent time windows, a set of staggered embedding time spectra is constructed to control the watermark embedding rhythm. This method is based on the previously constructed hierarchical alignment index and comprehensively considers the phase drift behavior between paths, the guiding texture distribution, and the positional relationship of the conflict windows. This ensures that each path reasonably adjusts its embedding plan within a given time range, thereby avoiding mutual superposition and interference of watermark signals at key nodes. The specific steps are as follows: After constructing the hierarchical alignment index, the hierarchical structure, starting scale, and guide texture distribution within the index should be used as references to identify high-density conflict areas where paths tend to overlap during watermark embedding. Conflict area identification requires combining the previously constructed conflict window. Within the timeframe before path convergence, intervals in the index structure where paths overlap, frequencies are similar, or guide textures are concentrated should be marked and graded according to the degree of conflict. Particular attention should be paid to the watermark start-up time overlap phenomenon near nodes where multiple paths simultaneously enter the embedding process, caused by overlapping starting scales. This should be initially mitigated by adjusting the path order or re-dividing texture boundaries. Based on this, several typical conflict areas should be extracted as key targets for subsequent peak-shifting adjustments, and each conflict area should be assigned a unique time index label to ensure precise control over its position and extent in subsequent steps.

[0035] Around the identified conflict areas, analyze whether there are time buffer segments in adjacent time segments that can be used for watermark redistribution, and construct a time buffer structure containing multiple micro-time windows. This structure should extract idle time periods from the original embedded time series that are not occupied by the watermark signal, have sparse path distribution, or have low signal interference, and classify them according to the duration, location, and path density of each segment. When constructing the buffer structure, the sensitivity of different paths to time precision should be considered to ensure that the adjusted embedding points do not disrupt the continuity and rhythmic integrity of the guiding texture in the path. In addition, the buffer segments should be compared and calibrated on the time axis with the original embedding intervals, and the time distance, signal interference probability, and path compatibility of each buffer segment relative to the conflict area should be recorded to provide a set of optional targets for the next embedding redistribution operation.

[0036] After constructing the buffer structure, based on the path information in the hierarchical alignment index, the watermark embedding points originally located in conflict areas are reallocated. These original embedding points are moved one by one to their corresponding buffer segments to achieve spatial misalignment and temporal rhythm separation. During this reallocation, the embedding order should be determined primarily by the path's hierarchical level. High-priority paths occupy the buffer time slice closest to their original position, while low-priority paths are sequentially deferred until the embedding requirements of all paths to be adjusted are covered. During the reallocation, the mapping relationship between each new embedding point and the original guiding texture should be maintained continuously. If necessary, the texture should be fine-grained extended to cover the newly added time range. Simultaneously, the starting scale of each path needs to be dynamically fine-tuned to ensure that the paths maintain overall rhythm consistency and response logic coherence even with staggered starts.

[0037] After redistributing and restructuring the embedding points, all adjusted embedding time points, corresponding path numbers, adjusted guide texture segments, and their respective hierarchical information are uniformly merged to generate a staggered embedding time spectrum covering the entire path watermark embedding behavior. This time spectrum uses the time axis as its main axis, vertically distributing the embedding point information for each path and horizontally arranging the start-up rhythm between paths, forming a time guidance framework that can be used for embedding control. The time spectrum should clearly indicate the path to which each embedding point belongs, its relative embedding time, signal adjustment amplitude, and the number of its buffer segment, allowing for indexing and sequential control during embedding operations. With the assistance of this time spectrum, the watermark signal can be implanted staggered according to a set rhythm, thus avoiding large-scale superposition interference before data convergence, providing a more stable and structurally ordered input foundation for the next stage of phase convergence adjustment.

[0038] The S500, based on the staggered embedding time spectrum, injects an anti-phase sentinel watermark during data backhaul aggregation, performs echo phase-locked signal adjustment, and dynamically converges the phase through a rhythm scaling control mechanism to maintain the continuity and stability of the traceability link. This step proposes a processing method based on staggered embedding time spectrum, injecting an anti-phase sentinel watermark during the data backhaul aggregation stage, and dynamically aligning the watermark phase through echo phase-locked signal adjustment and rhythm scaling control mechanisms. This method is built upon the previously generated staggered embedding time spectrum. Addressing the potential phase inconsistencies and interference superposition issues that may occur in multi-path watermark signals at the aggregation node, a series of dynamic control measures are employed to gradually synchronize the watermark signals to the target rhythm, thereby maintaining the continuity and stability of the tracing link in the final stage. Specifically, it includes the following steps: Based on the previously constructed staggered embedding time spectrum and the convergence times of each data path, a unified control interval suitable as the starting point for watermark aggregation is identified. This interval should be located in the transition period before all paths enter the result feedback aggregation node, and its time boundary should cover the idle period before the earliest path arrives at the aggregation node to the critical moment when the latest path is about to complete embedding. The purpose of this control interval is to provide a controllable buffer area, allowing watermark signals from different paths and at different time points to be pre-tuned and prepared for aggregation during this stage. During the identification process, the tail structure of the time spectrum should be focused on by comparing it with the last embedding time point of each path in the staggered embedding time spectrum, analyzing the time overlap area and path convergence trend, and extracting potential phase convergence windows. At the same time, to prevent aggregation failure caused by inconsistent watermark rhythms between paths, the required time length of the aggregation interval should be determined according to the rhythm density and phase change frequency of each path, and a rhythm buffer band should be preset.

[0039] Within the convergence control interval, a reverse-phase sentinel watermark is injected directionally, guided by path phase differences. This sentinel watermark is a set of auxiliary signals structurally independent of the main watermark information channel. It is used to create a small reverse disturbance before path convergence, thereby guiding the phase towards convergence within a common frequency range. The injection of the sentinel watermark is not for information embedding, but rather for phase correction. Its form is a signal beat with a phase opposite to the original path watermark, a weak amplitude, and a uniform period, used to intervene in the natural expansion trend of the path watermark, causing it to have a certain rhythmic convergence tendency in the time dimension. The injection of sentinel watermarks for each path should be differentiated according to its embedding rhythm in the staggered time spectrum. If the path is in a leading position in rhythm, a delayed reverse-phase sentinel signal should be injected; if it is in a lagging position, an advanced intervention signal should be injected to guide it towards the convergence centerline in time. At the same time, to prevent the sentinel signal from interfering with the original watermark, its effective range should be set to the central region within the convergence interval, avoiding coverage of the high-density section of the original watermark.

[0040] After the sentinel watermark is injected, an echo phase-locked loop (ELLLOL) signal adjustment mechanism is applied based on the response behavior of the path watermark within the convergence interval. This guides the watermark signals of all paths to resonate periodically in phase rhythm, thus entering a unified tempo. The core of this adjustment mechanism lies in setting a finely adjustable phase reflection segment at the end of the embedded signal of each path. This segment generates an echo waveform that corresponds to the rhythm of the preceding sentinel watermark as it approaches the end of the convergence interval. These echo segments, through a self-adjusting frequency construction method, gradually couple with the sentinel watermark in frequency, manifesting as several echo convergence peaks on the time scale. This forces the rhythmic spacing of the path signals to gradually converge, ultimately leading to phase alignment at the end of the convergence interval. Structurally, this process does not interfere with the original watermark recognition channel but adds a response feedback control channel. This allows watermark signals, originally distributed across different time periods and exhibiting significant phase deviations, to gradually converge to a unified phase reference point as they approach the final convergence node, forming a watermark overlap interval. This facilitates subsequent unified extraction and path recognition.

[0041] After the echo-locked loop (ELL) adjustment takes effect, to prevent the watermark signal from misaligning again due to slight differences between paths, a rhythmic scaling control mechanism needs to be introduced. This mechanism slightly stretches or compresses the watermark rhythm at the end of the path, ensuring its length and frequency on the time axis are ultimately aligned. The execution of rhythmic scaling control requires bidirectional matching based on the echo response effect and the sentinel watermark injection intensity. By applying small adjustments to the rhythm interval at the end of the path, the beat is aligned with the target convergence rhythm line. If a path exhibits a slower rhythm at the end of the convergence interval, the time interval of its last few rhythm cycles should be slightly compressed to allow its rhythm to enter the overlapping interval earlier; if the rhythm is faster, the rhythm interval should be slightly stretched to delay the beat alignment point. During the adjustment process, it is essential to ensure that the adjusted segments do not affect the logical coherence of the original path data structure. Simultaneously, the position information and scaling ratio of all adjusted segments should be marked on the time spectrum so that the complete watermark trajectory can be recovered based on this information during the subsequent extraction and recognition stages. Through dynamic control of rhythmic scaling, all path watermark signals are eventually synchronized before the data convergence point, constructing a watermark fusion structure with unified rhythm and consistent phase.

[0042] This invention effectively addresses the path drift and phase misalignment issues arising from multi-path, multi-node financial data during privacy computation by constructing a spatiotemporal mapping structure and hierarchical alignment index for watermark propagation. By precisely extracting phase anchor points and drift trajectories, it dynamically guides the watermark embedding position and rhythm, ensuring the watermark signal maintains structural coherence and rhythmic stability throughout complex data backhaul links. This improves the tracking continuity of data identifiers and prevents identifier failure due to signal drift.

[0043] This invention employs a staggered embedding and echo-locked loop adjustment mechanism to proactively absorb rhythm differences and converge phases before the watermark signals converge, ensuring the uniformity of the watermark structure and the clarity of recognition at the final node. By dynamically adjusting the embedding time spectrum and rhythmic scaling operations, the watermark signals achieve a rhythmic coordination state during the data aggregation stage, effectively avoiding cancellation or interference problems caused by signal superposition, thereby maintaining the integrity of the data tracking chain and the stability of watermark recognition.

[0044] This invention provides, for example Figure 2 The enterprise financial data sharing security processing system based on privacy computing shown includes a watermark spatiotemporal mapping construction module, a conflict detection and trajectory analysis module, an alignment guidance index generation module, a peak-shifting embedding scheduling module, and a phase-locked loop control and echo correction module. The watermark spatiotemporal mapping construction module constructs a spatiotemporal mapping structure for watermark propagation based on the time sequence and node topology information of the enterprise financial data backhaul link, and extracts the starting position of phase offset to generate a list of phase anchor points. The collision detection and trajectory analysis module, based on the phase anchor point list, identifies overlapping collision areas between paths before data aggregation, establishes collision windows, and extracts the phase drift trajectory of each path to form a trajectory record. The alignment guide index generation module generates alignment guide textures within the watermark carrier based on phase drift trajectory records, sets independent start-up scales for each data return path, and constructs a hierarchical alignment index. The staggered embedding scheduling module performs frequency division and hierarchical staggered embedding operation based on the hierarchical alignment index, redistributing the watermark signal in the overlapping conflict area to the adjacent time window to generate the staggered embedding time spectrum. The phase-locked loop (PLL) and echo correction module injects an anti-phase sentinel watermark during data backhaul aggregation based on the staggered embedding time spectrum, performs echo phase-locked loop signal adjustment, and dynamically converges the phase through a rhythm scaling control mechanism to maintain the continuity and stability of the traceability link.

[0045] The privacy-based computation-based enterprise financial data sharing security processing method provided in this embodiment of the invention is implemented through the aforementioned privacy-based computation-based enterprise financial data sharing security processing system. For details of the specific methods and processes of the privacy-based computation-based enterprise financial data sharing security processing system, please refer to the embodiments of the aforementioned privacy-based computation-based enterprise financial data sharing security processing method, which will not be repeated here.

[0046] The foregoing has only described certain exemplary embodiments of the present invention by way of illustration. Undoubtedly, those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the foregoing drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A method for securely processing shared enterprise financial data based on privacy-preserving computing, characterized in that: Includes the following steps: S100 constructs a spatiotemporal mapping structure for watermark propagation based on the time sequence and node topology information of the enterprise financial data backhaul link, and extracts the starting position of phase offset to generate a list of phase anchor points. S200, based on the phase anchor point list, identifies overlapping and conflicting areas between paths before data aggregation, establishes conflict windows, and extracts the phase drift trajectory of each path to form a trajectory record. S300, based on phase drift trajectory recording, generates alignment guide texture within the watermark carrier, sets an independent start scale for each data return path, and constructs a hierarchical alignment index; S400 performs a frequency division and layered staggered peak embedding operation based on the layered alignment index, redistributing the watermark signal in the overlapping conflict area to the adjacent time window to generate a staggered peak embedding time spectrum. The S500, based on the staggered embedding time spectrum, injects an anti-phase sentinel watermark during data backhaul aggregation, performs echo phase-locked signal adjustment, and dynamically converges the phase through a rhythm scaling control mechanism to maintain the continuity and stability of the traceability link.

2. The method for secure processing of enterprise financial data sharing based on privacy computing according to claim 1, characterized in that, Step S100 includes: Based on the collaborative path of enterprise financial data in privacy computing scenarios, the data flow links participating in joint computing are organized at the node level, a data backhaul link topology covering the entire process is constructed, and the connection relationship and path weight of each node are marked. Based on the link topology, the time sequence information of enterprise financial data during transmission is extracted, and a time sequence chain with hierarchical relationship is constructed to mark the position of data segments between nodes on the global time axis. By combining the node topology and the time sequence chain, a spatiotemporal mapping structure for watermark propagation is established, and the watermark propagation trajectory is mapped to the spatial node position and the time node sequence to form a three-dimensional spatiotemporal mapping surface. Based on the spatiotemporal mapping structure, the propagation characteristics of the watermark in each path are analyzed, the starting position of the phase shift of the watermark signal is marked, a list of phase anchor points is generated, and the path identifier, offset node and corresponding time coordinate information are recorded.

3. The method for secure processing of enterprise financial data sharing based on privacy computing according to claim 2, characterized in that, When generating the phase anchor point list, the starting offset position of the watermark signal is synchronously associated with the spatial coordinates in the node topology and the time nodes in the time sequence chain. The initial phase change point is determined by tracing back the path propagation process, and the anchor point correspondence is established based on the propagation direction information to achieve accurate positioning of the phase offset position and unified identification of the anchor point data.

4. The method for secure processing of enterprise financial data sharing based on privacy computing according to claim 3, characterized in that, Step S200 includes: Based on the generated spatiotemporal mapping structure and phase anchor point list of watermark propagation, we conduct centralized analysis on multiple backhaul paths of enterprise financial data, identify the intersection nodes of each path in the data flow convergence process, and establish a node correspondence mapping. Based on the identified path intersection nodes, combined with the offset starting point of each path in the phase anchor point list, potential conflict sections between paths are identified, and the time boundary and spatial span of the overlapping sections are determined. Based on the identified potential conflict zones, a conflict window structure is constructed around the path combinations with intersecting relationships. Continuous time segments are established by combining the time dimension and path distribution, and the path participation relationship and overlap density level are marked. Based on the constructed conflict window, the phase drift trajectory of each path from the phase anchor point to the intersection node is extracted, and the path start time, node position, drift amplitude and change direction are recorded to form a trajectory record.

5. The method for secure processing of enterprise financial data sharing based on privacy computing according to claim 4, characterized in that, When constructing the conflict window structure, the path is divided into continuous time segments on the time axis. By classifying the degree of overlap of watermark signals within the time segments, the overlap density level and interference direction are determined. A joint temporal and spatial mapping is established based on the differences in path distribution to limit the conflict range and interference boundary of watermark signals before data convergence.

6. The method for secure processing of enterprise financial data sharing based on privacy computing according to claim 4, characterized in that, Step S300 includes: Based on the phase evolution trend and jump node information of each data path in the phase drift trajectory record, a guiding reference structure is constructed in the watermark carrier to map the node information, time span and phase change amplitude of each path to a unified reference interval. By combining the phase drift trajectory data of each path, a path-specific guide texture is generated in the guide reference structure to ensure that the guide texture is consistent with the phase drift rhythm of the path and to avoid overlapping areas between textures. For each generated guide texture, an independent starting scale is set based on the anchor point time position of the path to determine the initial embedding time point of the watermark signal, and a time identifier signal and path number label are embedded in the carrier. After all paths have been set with independent starting scales, the guide textures and starting scales of each path are combined vertically to construct a hierarchical alignment index, which is used to guide the differentiated embedding control of the watermark signal before convergence.

7. The method for secure processing of enterprise financial data sharing based on privacy computing according to claim 6, characterized in that, The hierarchical structure of the bit index is divided into levels according to path priority and embedding order. Each level corresponds to a specific embedding channel. The paths start sequentially in time, are independently distributed in space, and run interleaved in the frequency domain, so that the watermark signal maintains a continuous rhythm before convergence and avoids embedding interference.

8. The method for secure processing of enterprise financial data sharing based on privacy computing according to claim 6, characterized in that, Step S400 includes: Based on the hierarchical structure, starting scale and guide texture distribution information in the hierarchical alignment index, high-density conflict areas with overlapping trends in the watermark embedding process are identified, and the conflict areas are hierarchically labeled. Around the identified conflict areas, analyze the time buffer segments in adjacent time segments that can be redistributed for watermarking, construct a time buffer structure, and record the time distance and path compatibility between the buffer segments and the conflict areas. Based on the time buffer structure, a reallocation operation is performed on the watermark embedding points located in the conflict area, moving the embedding points to the corresponding buffer segment, and dynamically fine-tuning the path start scale to maintain rhythm consistency. After the embedding points are redistributed, the embedding time points, guiding texture segments and hierarchical information of each path are merged to generate a staggered embedding time spectrum that covers the watermark embedding behavior of the entire path.

9. The method for secure processing of enterprise financial data sharing based on privacy computing according to claim 6, characterized in that, Step S500 includes: Based on the staggered embedding time spectrum and the convergence time of each data path, a unified control interval suitable as the starting point for watermark aggregation is identified, and the time boundary and rhythm buffer zone of the aggregation interval are determined. Within the convergence control range, guided by the path phase difference, an anti-phase sentinel watermark is injected, causing the watermark signal to exhibit a rhythmic convergence tendency in the time dimension, while avoiding interference with the original high-density watermark segment. After the sentinel watermark is injected, an echo phase-locked signal is applied to the path watermark to make the path watermark signals resonate periodically in phase rhythm and form a phase alignment trend at the end of the convergence interval. After echo lock-in modulation, a rhythm stretching control mechanism is introduced to slightly stretch or compress the watermark rhythm at the end of the path, so that all path watermark signals can complete rhythm convergence and maintain phase consistency before the data convergence point.

10. A privacy-based computation-based secure processing system for sharing enterprise financial data, used to implement the privacy-based computation-based secure processing method for sharing enterprise financial data as described in any one of claims 1-9, characterized in that, It includes a watermark spatiotemporal mapping construction module, a collision detection and trajectory analysis module, an alignment guidance index generation module, a peak-shifting embedding scheduling module, and a phase-locked loop control and echo correction module. The watermark spatiotemporal mapping construction module constructs a spatiotemporal mapping structure for watermark propagation based on the time sequence and node topology information of the enterprise financial data backhaul link, and extracts the starting position of phase offset to generate a list of phase anchor points. The collision detection and trajectory analysis module, based on the phase anchor point list, identifies overlapping collision areas between paths before data aggregation, establishes collision windows, and extracts the phase drift trajectory of each path to form a trajectory record. The alignment guide index generation module generates alignment guide textures within the watermark carrier based on phase drift trajectory records, sets independent start-up scales for each data return path, and constructs a hierarchical alignment index. The staggered embedding scheduling module performs frequency division and hierarchical staggered embedding operation based on the hierarchical alignment index, redistributing the watermark signal in the overlapping conflict area to the adjacent time window to generate the staggered embedding time spectrum. The phase-locked loop (PLL) and echo correction module injects an anti-phase sentinel watermark during data backhaul aggregation based on the staggered embedding time spectrum, performs echo phase-locked loop signal adjustment, and dynamically converges the phase through a rhythm scaling control mechanism to maintain the continuity and stability of the traceability link.

Images