Intelligent routing-based global heterogeneous message fusion method

Abstract

The invention discloses a global heterogeneous message fusion method based on intelligent routing, which relates to the technical field of network fusion communication, realizes wide access and deep decoupling of a heterogeneous system by constructing a unified message meta-model and a protocol adaptation layer, and remarkably improves the flexibility and expansibility of a framework. An intelligent routing decision algorithm based on multi-objective optimization is also introduced, so that an optimal transmission path can be dynamically selected according to a service strategy and real-time delay, cost and reliability indexes, and the end-to-end delivery quality of messages in a complex environment can be ensured through a built-in cross-domain high-reliability transmission and disaster recovery mechanism; finally, efficient, intelligent and reliable fusion of global messages is realized.

Description

Technical Field

[0001] This invention belongs to the field of network converged communication technology, and specifically relates to a global heterogeneous message fusion method based on intelligent routing. Background Technology

[0002] With the deepening of enterprise digital transformation and the development of cloud computing technology, information system architectures are exhibiting high complexity and heterogeneity. Core business processes often span multiple independent applications and platforms, such as traditional enterprise resource planning (ERP) systems, customer relationship management (CRM) systems, and emerging microservice applications and IoT devices. These systems are typically deployed in different environments such as on-premises data centers, private clouds, and public clouds, forming a hybrid or multi-cloud IT landscape. To ensure business continuity and data timeliness, modern enterprise IT architectures require efficient and reliable data exchange and message flow between heterogeneous systems. However, existing technologies, such as traditional point-to-point integration or simple message queue solutions, result in tight coupling between systems, making it difficult to adapt to rapid business changes. Furthermore, routing capabilities are usually based on static configuration, which not only fails to dynamically optimize based on real-time changing indicators such as network latency, operating costs, and system reliability, but also lacks reliability assurance mechanisms for long-distance transmission across regions, exhibiting poor adaptability and weak recovery capabilities in the face of complex network environments and system failures. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention provides a global heterogeneous message fusion method based on intelligent routing to solve the aforementioned technical problems.

[0004] The method for global heterogeneous message fusion based on intelligent routing includes the following steps:

[0005] The protocol adapter receives native messages from the source messaging system and encapsulates the native messages into messages in a unified message metamodel format.

[0006] By applying an intelligent routing decision algorithm, based on at least one evaluation dimension among latency, cost, and reliability, an optimal target system and its corresponding routing path are selected from multiple candidate target systems.

[0007] The message is sent to the optimal target system via the optimal routing path.

[0008] Preferably, after the step of encapsulating the raw message into a message in a unified message metamodel format...

[0009] The data payload of the message is validated according to the pre-registered data pattern, and a version conversion is performed according to the version of the data payload.

[0010] Preferably, the message is parsed based on routing rules defined using a domain-specific language to determine the multiple candidate target systems according to the message content, metadata, or context variables.

[0011] Preferably, when applying the intelligent routing decision algorithm, the specific steps include:

[0012] Establish and apply at least one of the delay optimization model, cost optimization model, and reliability assessment model to evaluate the candidate target system and routing path;

[0013] The comprehensive scoring algorithm calculates a comprehensive score by weighting the business weights set for the evaluation dimensions, and then selects the best candidate based on the comprehensive score.

[0014] Preferably, before sending the message to the optimal target system, the following steps are also included:

[0015] The message is processed by a message transformation engine, including at least one of the following: field mapping, data augmentation, data desensitization, format conversion, and content filtering.

[0016] Preferably, when the optimal target system is located in different geographical regions, the step of sending the message includes:

[0017] Multiple messages are batch-aggregated and compressed in the source domain to form transmission batches;

[0018] The transmission batch is sent to the target domain via a cross-domain transport layer.

[0019] Preferably, the step of transmitting via the cross-domain transport layer further includes at least one of the following steps:

[0020] Each message is assigned a globally unique identifier, and deduplication is performed in the target domain based on the globally unique identifier to ensure idempotency;

[0021] After a network transmission interruption, the transmission batch is resumed from the recorded breakpoint.

[0022] Preferably, the following steps are also included:

[0023] When the optimal target system is unavailable, the message is transferred to a local persistent dead-letter queue.

[0024] Once the system is available again, the messages are reread from the dead-letter queue and replayed at a controlled rate.

[0025] Preferably, the following steps are also included:

[0026] When a performance degradation or node failure is detected in the routing path, message traffic is switched to an alternative path through dynamic routing switching.

[0027] Preferably, the following steps are also included:

[0028] Inject or associate a globally unique tracking identifier into the message to enable end-to-end tracking and analysis of the entire message chain.

[0029] The beneficial effects of this invention are as follows: By constructing a unified message meta-model and protocol adaptation layer, it achieves broad access and deep decoupling of heterogeneous systems, significantly improving the flexibility and scalability of the architecture. It also introduces an intelligent routing decision algorithm based on multi-objective optimization, which can not only dynamically select the optimal transmission path according to business strategies and real-time latency, cost, and reliability indicators, but also ensure end-to-end message delivery quality in complex environments through built-in cross-domain high-reliability transmission and disaster recovery mechanisms, ultimately achieving efficient, intelligent, and reliable integration of global messaging. Attached Figure Description

[0030] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0031] Figure 1 A flowchart illustrating the steps of the global heterogeneous message fusion method based on intelligent routing provided by this invention;

[0032] Figure 2 This is a schematic diagram of the intelligent routing decision algorithm structure of the global heterogeneous message fusion method based on intelligent routing provided by the present invention. Detailed Implementation

[0033] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," "counterclockwise," "axial," "radial," and "circumferential" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0034] The following disclosure provides many different embodiments or examples for implementing different structures of the present invention. To simplify the disclosure of the present invention, the components and arrangements of specific examples are described below. Of course, these are merely examples and are not intended to limit the present invention.

[0035] The embodiments of the invention will now be described in detail with reference to the accompanying drawings.

[0036] like Figure 1 As shown, the global heterogeneous message fusion method based on intelligent routing includes the following steps:

[0037] The protocol adapter receives native messages from the source messaging system and encapsulates the native messages into messages in a unified message metamodel format.

[0038] By applying an intelligent routing decision algorithm, based on at least one evaluation dimension among latency, cost, and reliability, an optimal target system and its corresponding routing path are selected from multiple candidate target systems.

[0039] The message is sent to the optimal target system via the optimal routing path.

[0040] The core principle of this solution lies in achieving decoupling through standardized data abstraction and dynamic optimization through data-driven algorithms, thereby solving the problems of high system coupling, rigid routing, and poor reliability in existing technologies. The main execution component of this method is a global heterogeneous message fusion center deployed within the enterprise IT infrastructure. First, this method interacts with external heterogeneous systems through a protocol adapter. The protocol adapter has bidirectional conversion capabilities: when receiving native messages from the source messaging system, it parses and converts messages with different communication protocols and formats into an internal standard format, namely the unified message meta-model. This conversion is the foundation for all subsequent standardized processing. The unified message meta-model is the core custom data structure of this invention; it is not a specific message format, but rather an abstract data structure specification. It logically divides any message into two parts:

[0041] Standard Envelope: Encapsulates metadata unrelated to business content, used for routing, tracing, and control. The standard envelope must contain the following fields: globally unique ID, source system identifier, target system suggestion, timestamp, global tracing ID, message priority, and data mode version.

[0042] Business payload: The business data body that encapsulates the original message; it can be in any format.

[0043] This design separates "transmission control" from "service content," allowing the upper-layer processing logic of the fusion center to complete its work simply by manipulating standard envelopes without needing to understand the specific service payload, thus achieving complete decoupling from specific services and system protocols.

[0044] After the message is encapsulated into a unified message meta-model, it enters the core processing unit of this invention—the intelligent routing decision algorithm. The goal of this algorithm is to select the optimal target system and its corresponding routing path from multiple candidate target systems based on real-time status and business strategies. This step is detailed as follows:

[0045] First, establish and apply three evaluation models:

[0046] Delay optimization model: The round-trip time is obtained by sending probe messages to the target node of the candidate path, and the data can be smoothed by the exponential moving average algorithm to evaluate the network transmission delay.

[0047] Cost optimization model: Based on pre-configured cost factors, such as leased line bandwidth costs, public network traffic unit price, and computing resources consumed by message conversion, calculate the unit message cost through each path.

[0048] Reliability assessment model: Real-time monitoring of availability, error rate, message queue backlog depth, and target system CPU and memory usage of each path, and comprehensive calculation of a reliability index.

[0049] Then, a comprehensive scoring algorithm is used for decision-making. This algorithm normalizes the latency (L), cost (C), and reliability (R) of each candidate path, and then calculates a weighted sum (W_L, W_C, W_R) based on the weights set for the business scenario to obtain the comprehensive score S. For latency and cost, lower values ​​are generally better, so inverse normalization is required. The calculation formula is as follows:

[0050]

[0051] Finally, the algorithm selects the path with the highest overall score as the "optimal routing path," and the system it points to is the "optimal target system." Ultimately, the fusion center sends the message in the unified message metamodel format to the optimal target system via the selected optimal routing path. Before transmission, the protocol adapter again functions, converting the internal standard format message into the protocol and format required by the target system.

[0052] In Example 1, a payment message originating from the order system enters this method via the HTTP protocol. The protocol adapter encapsulates it into a unified message meta-model and sets its priority to "high". The intelligent routing decision algorithm obtains two candidate target paths: Path A, a cross-border leased line, with low latency and high reliability but high cost; Path B, a public VPN, with high latency, moderate reliability but low cost. Since the weight configuration for this business scenario is W_R=0.6, W_L=0.3, W_C=0.1, the comprehensive scoring algorithm calculates that the comprehensive score of Path A is significantly higher than that of Path B. Therefore, the system automatically selects Path A and sends the payment message to the target payment gateway, ensuring the real-time performance and stability of the transaction.

[0053] In Example 2, a batch of log data originating from a user behavior collection system enters this method via the MQTT protocol. The protocol adapter encapsulates it into a unified message meta-model with a "low" priority. The intelligent routing decision algorithm faces the same candidate paths. Since the weight configuration for this business scenario is W_C=0.7, W_L=0.2, W_R=0.1, the comprehensive scoring algorithm calculates that path B, with its lower cost, receives a higher comprehensive score. Therefore, the system selects path B for batch, asynchronous transmission, maximizing cost savings while meeting business requirements.

[0054] More specifically, after the step of encapsulating the raw message into a message in a unified message metamodel format,

[0055] The data payload of the message is validated according to the pre-registered data pattern, and a version conversion is performed according to the version of the data payload.

[0056] In the specific implementation process, an independent data schema registry center is set up to move the data structure validity verification and version compatibility work to the message fusion center, thereby decoupling the constraints and evolutionary complexities of data formats from various business systems. This center is a centralized metadata repository used to store and manage the data structure definitions, i.e., data schemas, of the "business payloads" of all messages flowing through the fusion center. The data schemas are defined using standardized languages ​​such as JSON Schema, Avro, or Protobuf, precisely describing the fields, data types, and constraints that the business payload should contain. When a message is encapsulated into a unified message metamodel format, this step is triggered: First, the method reads the "data schema version" identifier from the message's "standard envelope" and uses it as an index to retrieve the corresponding verification rules from the data schema registry center. Then, the method performs validity verification on the message's "business payload." If the structure or content of the payload does not match the data schema definition, the message will be rejected or routed to an exception handling queue, thus preventing "dirty data" from flowing into downstream systems. Next, the system will determine the data schema version that the target system expects to receive based on the routing results. If the target version differs from the current message version, the system will automatically perform a version conversion. This conversion is not arbitrary but based on a predefined rule script in the data schema registry that describes field mappings, additions, deletions, or logical transformations between versions. The system loads and executes this script to perform real-time conversion of the business payload, generating new version data that meets the requirements of the target system.

[0057] More specifically, the message is parsed based on routing rules defined using a domain-specific language to determine the multiple candidate target systems according to the message content, metadata, or context variables.

[0058] In a specific implementation, this solution provides a rule engine responsible for parsing and executing a set of routing rules written in a domain-specific language. A domain-specific language is a micro-programming language specifically designed for defining message routing logic, with concise and highly readable syntax. A routing rule text logically consists of two parts: a condition part and an action part. The condition part is typically introduced by a keyword indicating "when the following conditions are met," and contains one or more logical judgment expressions. These expressions can use comparison operators such as equals, greater than, and less than, and can be combined using logical connectors such as "AND" and "OR." The judgment object of the expression can directly reference data from the "Unified Message Metamodel," including metadata in its "standard envelope" or specific business fields in its "business payload."

[0059] The action section, typically guided by a keyword indicating "then perform the following action," defines the operation to be performed when the condition section is met. In this invention, this operation involves specifying identifiers for one or more systems and adding them to a list of candidate target systems.

[0060] In this implementation, when a message encapsulated as a "unified message metamodel" enters this step, the rule engine iterates through all deployed rules. For each rule, the engine parses its condition part and performs logical judgment based on the actual data in the message. If the final result of the condition part is true, the engine executes its action part, adding the target system identifier specified in the rule to a temporary candidate target system list. After all rules have been executed, this list contains all candidate targets that conform to the business logic and is submitted to the subsequent intelligent routing decision algorithm for final selection.

[0061] like Figure 2 As shown, more specifically, the application of the intelligent routing decision algorithm includes the following steps:

[0062] Establish and apply at least one of the delay optimization model, cost optimization model, and reliability assessment model to evaluate the candidate target system and routing path;

[0063] The comprehensive scoring algorithm calculates a comprehensive score by weighting the business weights set for the evaluation dimensions, and then selects the best candidate based on the comprehensive score.

[0064] In a specific implementation, after the present invention determines "multiple candidate target systems and corresponding routing paths" through the aforementioned routing rules, this step is initiated for optimization. First, the method performs parallel evaluations for each candidate routing path: First, a latency optimization model is established and applied. This model actively sends ICMP or TCP probe packets to the target node to obtain real-time network round-trip times, and uses an exponential moving average algorithm to smooth continuous probe results to calculate a stable and representative latency index for the path. Second, a cost optimization model is established and applied. This model quantifies the cost required to send a unit message through different paths based on a pre-configured cost factor table. The cost factors may include the hourly unit price of dedicated line bandwidth, the unit price per GB of public cloud outbound traffic, the CPU time consumed by message conversion, etc. The model weights and summarizes these factors to obtain a comprehensive cost index. Third, a reliability assessment model is established and applied. This model continuously monitors the health status of each node on the candidate path, the queue backlog depth of the message middleware, the CPU and memory load of the target system, and the success rate and error rate of historical message delivery to comprehensively calculate a quantified reliability score. After obtaining the evaluation data for the three dimensions of latency, cost, and reliability, this method makes decisions through a comprehensive scoring algorithm. The algorithm first normalizes the evaluation data from different models and units, mapping them to a unified [0,1] interval. Then, the algorithm loads business weights associated with the current business scenario. These weights are pre-set by the business administrator and directly reflect the different emphases the business places on latency, cost, and reliability.

[0065] More specifically, before sending the message to the optimal target system, the following steps are also included:

[0066] The message is processed by a message transformation engine, including at least one of the following: field mapping, data augmentation, data desensitization, format conversion, and content filtering.

[0067] In a specific implementation, after the intelligent routing decision algorithm determines the optimal target system, but before the message is finally sent, the message is handed over to a configurable message transformation engine. This engine performs one or more chained operations on the unified message metamodel instance of the message according to predefined processing rules associated with the routing path. These operations specifically include: performing field mapping, such as renaming the field name "user_id" in the source system's business payload to "customer Identifier" required by the target system; performing data enhancement, such as querying the geographical location information of the device from an external configuration library based on the device ID in the message metadata and adding it as a new field to the business payload; performing data desensitization, i.e., identifying and masking sensitive information in the business payload using regular expressions, such as converting "id_card":"310101...1234" to "id_card":"310101...XXXX"; performing format conversion, such as parsing and reconstructing the entire business payload from XML format to JSON format; and performing content filtering, i.e., removing debugging information or redundant fields in the business payload that are useless to the target system according to rules. After all processing steps are executed in a predetermined order, the modified message is delivered to the protocol adapter for final transmission. The key is to extract non-standardized data processing logic closely related to specific business operations from the endpoint system and centralize it within the message conversion engine of this method for unified execution, thereby achieving deep processing and value-added processing of message content.

[0068] More specifically, when the optimal target system is located in different geographical regions, the step of sending the message includes:

[0069] Multiple messages are batch-aggregated and compressed in the source domain to form transmission batches;

[0070] The transmission batch is sent to the target domain via a cross-domain transport layer.

[0071] In a specific implementation, when the optimal target system selected by the intelligent routing decision algorithm is identified as being located in a different geographical region, this invention does not immediately send a single message. Instead, it triggers a dedicated cross-domain transmission process: First, at the message center exit node of the source domain, multiple messages destined for the same target domain are batch-aggregated. This aggregation process can continuously accumulate messages until the resulting data set reaches a preset volume threshold or a preset time window times out. Subsequently, the aggregated message set is compressed as a whole to form a compact transmission batch. Next, a dedicated cross-domain transmission layer is activated. The specialization of this transmission layer is reflected in the fact that it is not a simple network connection, but a software layer with built-in advanced reliability mechanisms to solve the problem of cross-geographical region transmission. First, it achieves idempotency guarantees: before transmission, the source domain's transmission layer generates a globally unique batch ID for each transmission batch, and the receiving end of the target domain records all successfully processed batch IDs. When the target domain receives a batch, it first checks its ID. If it finds that the ID has already been recorded, it directly discards the duplicate batch, thereby fundamentally avoiding the problem of downstream systems repeatedly processing messages due to network retransmission. Second, it incorporates a built-in mechanism for resuming interrupted transmissions: before transmission, this dedicated layer logically divides a large transmission batch into multiple smaller data blocks and numbers them. It sends the data blocks one by one and waits for acknowledgment from the receiving end. If a cross-border network link experiences a momentary interruption, the transmission process is paused; once the network is restored, the sender can resume transmission from the interrupted data block based on the last received acknowledgment number, rather than retransmitting the entire large batch from the beginning.

[0072] More specifically, the step of sending via the cross-domain transport layer further includes at least one of the following steps:

[0073] Each message is assigned a globally unique identifier, and deduplication is performed in the target domain based on the globally unique identifier to ensure idempotency;

[0074] After a network transmission interruption, the transmission batch is resumed from the recorded breakpoint.

[0075] The two steps executed in the cross-domain transport layer are specific technical implementations of the aforementioned dedicated cross-domain transport layer. Their design principle lies in sinking and embedding the application layer's business data consistency guarantee and the network layer's transmission process reliability guarantee into the message infrastructure, thereby providing the host system with a transparent, self-healing, and wide-area communication service that guarantees correct business results. In specific implementations, these two steps are integrated into the collaborative work of the sending and receiving ends of the cross-domain transport layer. First, in the message processing pipeline of the source domain of this invention, when a message is encapsulated into a unified message meta-model, each original business message is assigned a globally unique and immutable identifier, such as a snowflake algorithm ID generated based on timestamps, machine codes, and sequence numbers, or a standard UUID. This unique identifier is stored in the message's "standard envelope" metadata area. When one or more messages are aggregated, compressed, and sent to the target domain as a "transmission batch," the dedicated transport layer receiver in the target domain immediately extracts this globally unique identifier when decompressing and parsing individual messages. The receiver maintains a time-sensitive cache of processed message IDs based on a high-performance key-value store. Before delivering a message to the downstream local business system, the receiving end first queries the cache using the message's unique identifier. If the identifier already exists, it indicates that the message was duplicated due to network retransmission or other reasons, and the receiving end will discard the message directly without any further processing. If the identifier does not exist, the receiving end delivers the message normally to the business system. After receiving confirmation of successful processing from the business system, the receiving end writes the message's unique identifier into the processed message ID cache and sets a reasonable lifespan. At the source domain sender, when a large transmission batch, such as 500MB, is ready to be sent, the dedicated transport layer logically divides it into a series of fixed-size, consecutively numbered data blocks. The sender and receiver maintain state synchronization for the transmission of this batch. The sender sends data block by block and updates a checkpoint for each successfully sent data block that receives confirmation from the other end. This checkpoint records the maximum consecutive number of the successfully transmitted data blocks. If, during transmission, the cross-border network link is interrupted, the TCP connection is broken, and transmission is paused. Once the network connection is restored, the dedicated transport layer at the sending end does not need to retransmit the entire 500MB batch. Instead, it first communicates with the receiving end to verify the last valid checkpoint recorded by both parties, and then resumes transmission from the first data block immediately following that checkpoint.

[0076] More specifically, it also includes the following steps:

[0077] When the optimal target system is unavailable, the message is transferred to a local persistent dead-letter queue.

[0078] Once the system is available again, the messages are reread from the dead-letter queue and replayed at a controlled rate.

[0079] In practice, when continuous health checks determine that the previously selected optimal target system has become unavailable—for example, due to multiple consecutive connection timeouts, receiving an HTTP 503 unavailable response, or a failed heartbeat—the system ceases sending messages. Instead, it immediately transfers the complete unified message metamodel instance to a disk-based persistent dead-letter queue located locally at the message fusion center. Persistence means that even if the message fusion center itself restarts, the messages in the queue will not be lost. A separate monitoring process is then started to continuously check the availability of the target system at a preset, gradually increasing probe interval—for example, every 10 seconds initially, then every 30 seconds, and then every minute. Once the target system is detected to have recovered service—for example, when the health check interface returns a successful response—the replay process is automatically triggered. This process does not send all the messages accumulated in the dead-letter queue at once, as this could potentially overwhelm the newly recovered target system again due to a surge in instantaneous traffic. Instead, this method rereads and resends messages from the head of the dead-letter queue at a controlled rate. This rate can be precisely controlled using rate limiting algorithms such as token bucket or leaky bucket, for example, configured to not exceed 50% of the target system's normal processing capacity. During replay, the system continuously monitors the target system's performance metrics, such as response latency and error rate, to dynamically adjust the replay rate, ensuring a smooth and safe recovery process.

[0080] More specifically, it also includes the following steps:

[0081] When a performance degradation or node failure is detected in the routing path, message traffic is switched to an alternative path through dynamic routing switching.

[0082] In its implementation, the message fusion center not only executes an intelligent routing decision algorithm once when selecting a path, but also continuously and in real-time monitors the performance of the currently used optimal routing path. The monitoring metrics directly reuse data from the aforementioned evaluation models for latency, cost, and reliability. This invention presets dynamic health thresholds for these key performance indicators; for example, latency exceeding the baseline value by 50% three consecutive times, or a message delivery error rate exceeding 2% within one minute. Once the monitoring system detects that any performance indicator of the current path has deteriorated and exceeded the preset threshold, or that any critical node on the path experiences a complete failure, the dynamic routing switching mechanism is immediately triggered. This mechanism immediately removes the problematic path from the current candidate list or sets its score to the lowest, then re-executes the intelligent routing decision algorithm to recalculate and evaluate from the remaining alternative paths, selecting a new "currently optimal" path. Subsequently, the traffic scheduler within the message fusion center seamlessly switches all subsequent messages to this newly selected alternative path, without any manual intervention.

[0083] More specifically, it also includes the following steps:

[0084] Inject or associate a globally unique tracking identifier into the message to enable end-to-end tracking and analysis of the entire message chain.

[0085] In practical implementation, when any external message first enters the entry node of this invention, the system immediately generates a globally unique trace identifier for it, such as a trace-id conforming to the W3C Trace Context specification, or a standard UUID. This identifier is injected into the "standard envelope" metadata area of ​​the message's unified message metamodel instance. From this point onward, all log records of every critical processing step the message undergoes internally will be automatically associated with this trace identifier. More importantly, when the message is sent to the downstream optimal target system through the protocol adapter, this trace identifier is also passed down by placing it in the target protocol's standard header, such as an X-Trace-ID header for an HTTP request, or a user-defined attribute for an AMQP message.

[0086] 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 them. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some or all of the technical features therein. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention, and they should all be covered within the scope of the claims and specification of the present invention.

Claims

1. A global heterogeneous message fusion method based on intelligent routing, characterized in that, Includes the following steps: The protocol adapter receives native messages from the source messaging system and encapsulates the native messages into messages in a unified message metamodel format. An intelligent routing decision algorithm is applied to select an optimal target system and its corresponding routing path from multiple candidate target systems based on at least one evaluation dimension of latency, cost, and reliability; the message is then sent to the optimal target system via the optimal routing path.

2. The global heterogeneous message fusion method based on intelligent routing according to claim 1, characterized in that, After the step of encapsulating the raw message into a message in a unified message metamodel format, the data payload of the message is validated according to the pre-registered data pattern, and a version conversion is performed according to the version of the data payload.

3. The global heterogeneous message fusion method based on intelligent routing according to claim 1, characterized in that, The message is parsed based on routing rules defined using a domain-specific language to determine the multiple candidate target systems according to the message content, metadata, or context variables.

4. The global heterogeneous message fusion method based on intelligent routing according to claim 1, characterized in that, When applying the intelligent routing decision algorithm, the specific steps include: establishing and applying at least one of the delay optimization model, cost optimization model, and reliability assessment model to evaluate the candidate target system and routing path; using a comprehensive scoring algorithm, performing weighted calculations based on the business weights set for the evaluation dimensions to obtain a comprehensive score, and selecting the best option based on the comprehensive score.

5. The global heterogeneous message fusion method based on intelligent routing according to claim 1, characterized in that, Before sending the message to the optimal target system, the process further includes the following steps: performing at least one processing on the message through a message conversion engine, the processing including: field mapping, data augmentation, data desensitization, format conversion, and content filtering.

6. The global heterogeneous message fusion method based on intelligent routing according to claim 1, characterized in that, When the optimal target system is located in different geographical regions, the step of sending the message includes: batch aggregating and compressing multiple messages in the source domain to form a transmission batch; and sending the transmission batch to the target domain through a cross-domain transmission layer.

7. The global heterogeneous message fusion method based on intelligent routing according to claim 6, characterized in that, The step of sending through the cross-domain transport layer further includes at least one of the following steps: assigning a globally unique identifier to each message and performing deduplication processing in the target domain based on the globally unique identifier to ensure idempotency; and resuming the transmission batch from the recorded breakpoint after network transmission is interrupted.

8. The global heterogeneous message fusion method based on intelligent routing according to claim 1, characterized in that, It also includes the following steps: When the optimal target system is unavailable, the message is transferred to a local persistent dead-letter queue. Once the system is available again, the messages are reread from the dead-letter queue and replayed at a controlled rate.

9. The global heterogeneous message fusion method based on intelligent routing according to claim 8, characterized in that, It also includes the following steps: When a performance degradation or node failure is detected in the routing path, message traffic is switched to an alternative path through dynamic routing switching.

10. The global heterogeneous message fusion method based on intelligent routing according to claim 1, characterized in that, It also includes the following steps: Inject or associate a globally unique tracking identifier into the message to enable end-to-end tracking and analysis of the entire message chain.

Images