Multi-network converged communication data processing method and system

Abstract

The invention relates to the technical field of digital information transmission, and discloses a multi-network converged communication data processing method and system, and the method comprises the steps: pre-assembling a metadata header mirror image containing a slow data abstract and a state value through a slow path processing unit at a network node, and storing the metadata header mirror image into a stream cache; and when the fast path forwarding unit receives the fast data packet, retrieving a metadata header mirror image from the stream cache, and executing hardware rewriting on the fast data packet to form an enhanced data packet. According to the invention, the fast and slow path architecture is utilized on the transmission layer, and the slow data context information is embedded into the fast data packet on the hardware level in a header mirror image mode; the same-packet and low-delay transmission of the slow data context and the fast data payload is realized, so that the dilemma of waiting or blind movement of a receiving end caused by time sequence mismatching of a heterogeneous network is solved, and the timeliness of a fast data stream is ensured.

Description

TECHNICAL FIELD

[0001] The present application relates to a multi-network converged communication data processing method and system, belonging to the technical field of digital information transmission. BACKGROUND

[0002] At present, especially in systems such as industrial control and smart grid that require high reliability decision-making, the making of decisions often depends on the fusion of multiple data streams from different sources. The existing transmission and processing methods generally rely on the premise that logically associated data streams are roughly synchronized in transmission timing or their time delay difference is within a controllable range, and the receiving end can use conventional buffering mechanisms to achieve the final data alignment and fusion. With the deep integration of high-speed low-latency networks and low-speed high-latency networks, for example, when a millisecond-level response 5G network and a minute-level reporting NB-IoT network are used together for the same decision logic, the above transmission method based on timing synchronization assumption encounters a fundamental challenge. The transmission delay difference between the two types of networks can reach several orders of magnitude, and this huge timing mismatch fundamentally destroys the transmission basis for effective data fusion at the receiving end. In this case, the processing logic at the receiving end faces a technical choice resulting from the transmission mechanism itself: if you choose to wait for the slow data stream to arrive to achieve data integrity, the fast data stream that has arrived will be forced to stay in the buffer for a long time, which makes its millisecond-level low-latency transmission value be offset. This is unacceptable for scenarios that require instantaneous response. If you choose to give up waiting and make an immediate decision based only on the fast data stream, the transmission system is in an unknown state for the logical context carried by the slow data stream, leading to serious errors in decision-making due to the lack of necessary context support. This limitation of the transmission layer that cannot coexist with timeliness and integrity is an issue that existing transmission methods have failed to effectively address.

[0003] In the technical exploration of solving network fusion timing mismatch, there are also deficiencies in how to efficiently utilize the idle computing resources of existing devices, especially the scheduling and allocation of idle computing resources in home networks. For example, the Chinese invention patent with publication number CN115866045A discloses a resource processing method, system, device and medium based on multi-network fusion computing. The patent proposes that the target data packets that need to be migrated in the edge gateway node of the communication network are transferred to the television node of the television network for calculation. By controlling the television node to play pre-cached resources such as advertisements or pre-stored television resources, the computing power is released to complete the calculation of the target data packets. Although this scheme utilizes the computing power of the television node, its core idea is to shift the computing power and occupy the playing period. Specifically, the computing power must be released by stopping the task of real-time analysis of television signals when the television node is in an idle state or during a specific time window when advertisements or pre-cached television resources are played. The essence of this mechanism is to make a trade-off between user viewing experience and computing power scheduling, and forcibly occupy the user's playing time to complete the computing task. Especially when processing high-time-efficiency data or fast data streams, this scheduling method that relies on playing interruption or pre-stored content playing cannot guarantee the immediacy, stability and certainty of the calculation. In particular, in the case of heterogeneous network fusion, the contradiction between the time efficiency of fast data streams and the context integrity of slow data streams is a difficult problem that cannot be solved by existing technologies.

[0004] Therefore, how to provide a data transmission processing method, which enables high-speed data streams to instantly obtain the context information of low-speed data streams logically associated with them while maintaining their low-latency characteristics, to solve the problems of low utilization of transmission information and decision-making disability caused by timing mismatch of heterogeneous networks, has become a technical problem to be solved by the present application. SUMMARY

[0005] The present application provides a multi-network fusion communication data processing method, which mainly aims to solve the problem that in the prior art, due to the large difference in transmission timing of heterogeneous networks, the receiving end cannot balance the time efficiency of fast data streams and the context integrity of slow data streams, thereby facing limitations in transmission information utilization and decision-making.

[0006] To achieve the above-mentioned purpose, the present application provides a multi-network fusion communication data processing method, which is executed at a first network node in a multi-network fusion communication system to process data sent to a receiving end. The first network node includes a slow path processing unit and a fast path forwarding unit, as well as a flow cache associated with the fast path forwarding unit. The method comprises:

[0007] Step 101, the slow path processing unit configures a data fusion transmission strategy, which defines the logical association between the fast data stream of high-speed network transmission and the slow data stream of low-speed network transmission, and defines the summary generation rule of the slow data stream.

[0008] Step 102: The slow path processing unit attempts to obtain a slow data summary in the background according to the summary generation rules.

[0009] Step 103 is executed by the slow path processing unit: when the slow data digest is successfully obtained, the slow data digest and the first state value are assembled into a metadata header image; when the slow data digest is not obtained, the second state value is assembled into a metadata header image, and the metadata header image is stored in the flow buffer associated with the fast path forwarding unit.

[0010] Step 104: The fast path forwarding unit monitors the fast data stream in real time and performs flow matching on the fast data packets that conform to the data fusion transmission strategy.

[0011] Step 105: Based on the result of flow matching, the fast path forwarding unit retrieves the metadata header image from the flow buffer, and at the hardware forwarding level, uses the metadata header image to rewrite one or more preset metadata fields of the fast data packet to form an enhanced data packet.

[0012] Step 106: The fast path forwarding unit sends the enhanced data packet to the receiving end through the high-speed network.

[0013] Preferably, the method further includes: step 201, at the first network node, monitoring the source address of data packets in the slow data stream in real time to maintain an active source list; step 202, counting the number of active source ends in the active source list to obtain a digest source count; and in step 103, when a slow data digest is successfully obtained, assembling the digest source count together with the slow data digest and the first status value into a metadata header image.

[0014] Preferably, step 102 further includes: using the generation time of the slow data digest as the digest timestamp, and in step 103, when the slow data digest is successfully obtained, assembling the digest timestamp, the slow data digest, and the first state value into a metadata header image.

[0015] Preferably, the data fusion transmission strategy defines a consistent digest group including at least two slow data digests; step 102 includes: independently generating at least two slow data digests and their respective corresponding digest timestamps in the background; step 103, before assembling the metadata header image, further includes: step 401, obtaining at least two digest timestamps; step 402, calculating the time difference between at least two digest timestamps. Step 403, when the time difference When the time difference exceeds the preset maximum allowable time deviation, the second state value is assembled into a metadata header image; step 404, when the time difference... At least two slow data digests and the first state value will be assembled into a metadata header image only when the time deviation is less than or equal to the preset maximum allowable time deviation.

[0016] Preferably, the method further includes: step 501, processing the time difference. Events exceeding the preset maximum allowable time deviation are counted to obtain a failure count value; step 502, it is determined whether the failure count value exceeds the preset network degradation threshold within the preset time window; step 503, when the failure count value exceeds the preset network degradation threshold, a network management alarm message is generated and sent, the alarm message is used to indicate that the transmission quality of the low-speed network associated with the consistency summary group has deteriorated.

[0017] Preferably, the summary generation rule is a non-predictive aggregation algorithm selected from at least one of summation, counting, averaging, and state list statistics for slow data streams.

[0018] Preferably, the first network node is an edge gateway, router, 5G base station, or software-defined network (SDN) switch; the preset metadata field is the IP header options, TCP header options, or extended header of the encapsulation protocol of the fast data packet; the method further includes the following steps performed at the receiving end: step 801, receiving the enhanced data packet; step 802, parsing the fast data from the payload of the enhanced data packet; step 803, parsing the slow data digest, first status value, or second status value from the preset metadata field of the enhanced data packet.

[0019] Preferably, the method further includes: step 901, when the first state value is parsed, performing fusion data processing based on the fast data and slow data digest; step 902, when the second state value is parsed, executing a security policy based on missing context.

[0020] Preferably, in step 103, when the slow data digest is successfully obtained, the digest timestamp, digest source count, slow data digest and the first state value are assembled into a metadata header image.

[0021] A multi-network converged communication data processing system, the system comprising:

[0022] Slow path processing unit, fast path forwarding unit, and first-class cache associated with fast path forwarding unit;

[0023] The slow path processing unit is used to: configure a data fusion transmission strategy, which defines the logical relationship between fast data streams transmitted over a high-speed network and slow data streams transmitted over a low-speed network, and defines the digest generation rules for slow data streams; according to the digest generation rules, attempt to obtain slow data digests in the background; when the slow data digest is successfully obtained, assemble the slow data digest and a first status value into a metadata header image; when obtaining the slow data digest fails, assemble a second status value into a metadata header image; and store the metadata header image in the stream buffer.

[0024] The fast path forwarding unit is used to: monitor fast data streams in real time and perform flow matching on fast data packets that conform to the data fusion transmission strategy; retrieve metadata header images from the flow buffer based on the flow matching results; at the hardware forwarding level, use the metadata header images to rewrite one or more preset metadata fields of the fast data packets to form enhanced data packets; and send the enhanced data packets to the receiving end through the high-speed network.

[0025] Compared with the prior art, the beneficial effects of the present invention are:

[0026] 1. At the first network node, which serves as the convergence point of heterogeneous networks, a slow data digest is pre-generated using its background caching mechanism and embedded into the preset metadata field of the fast data packet. This enables the transmission of slow data context information and fast data payload in the same packet. This method changes the traditional data alignment method in digital information transmission that relies on the receiver's buffering and waiting. It transforms the time problem of waiting at the transport layer into the space problem of field occupancy at the transport layer. This allows the high-speed fast data stream to retain its millisecond-level timeliness when it arrives at the receiver, while also obtaining the necessary decision context. From the perspective of transmission mechanism, this avoids the dilemma of waiting or blind action caused by timing mismatch in the system.

[0027] 2. By setting a digest status identifier field in the preset metadata field, and actively setting the identifier field to the second preset status value when the digest acquisition fails, such as when the cache is empty or the data is invalid, it is ensured that the enhanced data packet sent by the first network node to the receiving end always carries a clear transmission context status signaling. This mechanism, combined with the mechanism of performing time-series consistency verification on the consistency digest group in claim 7, not only avoids the transmission risk of misleading information caused by time-series tearing through verification at the data level, but also solves the information ambiguity problem faced by the receiving end when facing extreme working conditions such as cache non-ready through status codes at the signaling level. Together, they construct an information delivery method that has both security and state determinism at the transmission level.

[0028] 3. The digest embedding action is further restricted to the fast path forwarding unit of the first network node. The slow path processing unit pre-assembles the digest, timestamp, or status identifier into a complete metadata header image in the background and stores it in the stream buffer. When a fast data packet arrives, the fast path forwarding unit does not process it through the CPU, but directly uses the header image to rewrite or replace the data packet at the hardware forwarding level. This architecture, which transforms the packet-by-packet dynamic calculation embedding action into a stream-by-stream static replacement hardware line-rate operation, reduces the latency of the digest embedding action itself from microsecond-level software processing to nanosecond-level hardware forwarding. This ensures that this method will not interfere with or damage the original low-latency and low-jitter transmission characteristics of the fast data stream when facing extreme traffic conditions such as fast data packet storms. Attached Figure Description

[0029] Fig. 1 This is a schematic diagram of the architecture of a multi-network converged communication data processing method and system according to the present invention;

[0030] Fig. 2 This is a schematic diagram illustrating the relationship between the key parameters of this invention and their impact on performance.

[0031] Fig. 3 This is a timing diagram illustrating the generation of metadata header images by the slow path processing unit of the present invention. Detailed Implementation

[0032] To make the technical means, creative features, objectives and effects of this invention easier to understand, the invention will be further described below in conjunction with specific embodiments. It should be noted that the description herein is only for explaining the invention and does not constitute a limitation on the scope of the invention.

[0033] This invention provides a multi-network converged communication data processing method and system. Its operating mechanism is deployed at a key aggregation node in the multi-network converged communication system, namely the first network node. This node undertakes the data aggregation and forwarding functions of high-speed, low-latency networks and low-speed, high-latency networks. The first network node utilizes mature fast and slow path processing mechanisms in network devices. Its system correspondingly includes a slow path processing unit, a fast path forwarding unit, and a flow buffer highly associated with the fast path forwarding unit. The operating logic of this solution is as follows: the slow path processing unit is typically powered by a general-purpose CPU. The execution control plane task is responsible for performing relatively complex logical processing asynchronously and non-real-time in the background. This includes parsing slow data streams, generating context summary information according to policies, and pre-compiling a metadata header image that can be directly used by the hardware forwarding layer. This header image is pushed down and stored in the stream buffer associated with the fast path forwarding unit. The fast path forwarding unit (usually executed by hardware such as ASICs or FPGAs to perform data plane tasks) is responsible for line-rate processing of high-speed fast data streams. Through hardware stream matching, it retrieves the corresponding header image within nanoseconds and rewrites it in pure hardware. By embedding the slow data context or its state into the metadata field of the fast data packet, the context information and fast data are transmitted in the same packet with low latency priority, thus solving the timing mismatch problem in heterogeneous networks at the transport layer. To ensure that the fast path forwarding unit can always obtain deterministic context information at hardware line speed in step 105, the update and invalidation management procedure of the metadata header image in the stream cache A1 is executed by the slow path processing unit. Specifically, when the slow path processing unit writes the metadata header image to the stream cache in step 103, it synchronously calculates the cache entry based on the service characteristics of the slow data stream. A Time-to-Live (TTL) value is calculated and set. For example, the TTL value can be set to between 1.5 and 2.0 times the known maximum reporting period of the slow data stream. The flow buffer hardware associated with the fast path forwarding unit is configured to automatically replace the data content of the entry with a context failure image containing a second status value when the TTL expires, instead of physically deleting the buffer entry. This procedure ensures that the hardware flow matching action of the fast path forwarding unit can always retrieve a result within a certain clock cycle, that is, obtain a valid metadata header image or obtain a clear context failure status signaling.

[0034] In step 102, the slow path processing unit performs background actions to obtain slow data summaries according to the summary generation rules. The triggering mechanism can be based on a periodic aggregation window procedure. Under this procedure, the slow path processing unit starts a background timer with a fixed period. The period value of this timer is determined based on the expected arrival frequency of the slow data stream and the freshness requirements of downstream services; for example, it can be set to 500 milliseconds or 1 second. Each time the timer expires, the slow path processing unit executes the aggregation algorithm defined by the summary generation rules on all associated slow data stream data packets received and temporarily stored within the time window. This includes operations such as summation, average calculation, or state list statistics. The generated slow data summary result is then used to update the corresponding metadata header image in the stream cache A1. If no slow data stream data packets are received within the time window, the unit can choose to maintain the summary data from the previous period or update it to a context-invalidated image containing a second state value according to a strategy. In a specific application scenario, this is used for transient stability control of a smart grid. For example, the method is executed at the first network node, which can be an edge computing gateway, in the following steps: Step 101 is executed, where the slow path processing unit configures the data fusion transmission strategy. In the transmission network, the system must establish logical associations between different data streams to identify which fast data requires context support from which slow data. For this purpose, the strategy is a set of rules defined by the network administrator, and its input is the header characteristics of the data packets. The strategy can be defined as follows: all data packets with a source IP address of 192.168.1.0 / 24 (PMU aggregation area) and a UDP destination port of 5000 are identified as fast data streams; and the fast data stream is logically associated with the data stream with a source IP address of 10.10.0.0 / 16 (NB-IoT wide area measurement and control area) and a protocol type of CoAP. The strategy further defines the summary generation rules for slow data streams. These rules are limited to a non-predictive aggregation algorithm, and their purpose is not complex analysis in the G06F domain, but rather context aggregation at the transport layer. The rule can be defined as: AVG(Payload_Byte[4:8]), which calculates the average of bytes 4 to 8 of the payload of all associated slow packets, assuming it is the total regional load value, and uses it as the slow data digest. Then, the slow path processing unit asynchronously executes step 102 in the background, that is, according to the rule defined in step 101, it attempts to obtain the slow data digest in the background. When one or more CoAP slow packets arrive, the AVG() algorithm is triggered to obtain a slow data digest with a value of 150.75MW.

[0035] To address the transmission problem of digest information distortion caused by dynamic changes in slow network topology, such as device disconnection, this method also executes steps 201 and 202 in parallel. Specifically, during step 102, the IP address of the source address L3 of data packets in the slow data stream is monitored, and an active source list is maintained in memory. This list can be implemented as a hash table with the source IP as the key and the last active timestamp as the value. A lifespan (Lv) is set for each list item, which can be set to 15 minutes. Simultaneously with digest generation, step 202 counts the number of active sources in the list that have not timed out, obtaining a digest source count of 98. To characterize the freshness of the digest and for timing consistency verification in subsequent transmissions, step 102 further includes: simultaneously with digest generation, the slow path processing unit queries its own system clock to obtain the current time, specifically UTC14:30:05.120, and uses this as the digest timestamp; step 103 is then executed, where the slow path processing unit, based on the above... The result is used to perform state judgment and metadata header image assembly. This step is used to handle different states during transmission to ensure the determinism of transmission: Case 1: Successful acquisition: When the slow data digest is successfully acquired, that is, the digest 150.75MW has been generated, the slow path processing unit will assemble the digest source count (98), digest timestamp (14:30:05.120), slow data digest (150.75), and a first state value representing data validity (its specific value is 0x01) into a complete binary metadata header image according to the protocol format of the preset metadata field, which can be selected by a custom TCP option or VXLAN extended header; Case 2: Acquisition failure: When the acquisition of the slow data digest fails, taking the example that the backend cache is empty due to the node's cold start, or the digest in the cache expires due to timeout, the slow path processing unit will assemble a second state value representing context not ready (its specific value is 0x02) into a metadata header image.

[0036] In more complex fusion scenarios, to avoid internal time-series tearing within the data context (i.e., excessively large differences in timestamps among multiple related digests), this data fusion transmission strategy can also define a consistent digest group including at least two slow data digests (taking digest A and digest B as an example), and preset a maximum allowable time deviation of 10 seconds. In this case, before assembling the header image, step 103 will further execute steps 401 and 402, that is, obtain the timestamps of the corresponding digests within the group and calculate the time difference between them. ,like Greater than the preset maximum allowable time deviation (if) =30 seconds), then this situation is judged as acquisition failure, and step 403 is executed to forcibly assemble a header image containing the second state value, in order to actively prevent the transport layer from transmitting misleading information about timing tears, only when If the deviation is less than or equal to this value, step 404 is executed to assemble a header image containing the first state value and two digests. Finally, at the end of step 103, the slow path processing unit stores the metadata header image generated in any of the above cases in the stream cache associated with the fast path forwarding unit, for example, a TCAM or a hardware hash table, and binds it to the fast data stream identification features defined in step 101. When the fast data stream arrives, the processing flow switches to the fast path forwarding unit, for example, an ASIC chip: Step 104 is executed, where the fast path forwarding unit monitors the fast data stream in real time. When a fast data packet arrives at the hardware port, for example, a UDP packet from 192.168.1.10 with a destination port of 5000, its header is immediately parsed, and hardware stream matching is performed. Then, step 105 is executed. Based on the stream matching result, the hardware, for example, uses TCAM to hit the entry in the stream cache within nanoseconds and retrieves the corresponding binary metadata header image from the stream cache. The fast path forwarding unit (ASIC) then... At the hardware forwarding level, the header mirroring is used to rewrite one or more preset metadata fields of the current fast data packet. The preset metadata fields are standard fields in the protocol stack, such as the IP options field in the IPv4 header, the TCP options field in the TCP header, or a custom field in the extended header of a tunneling encapsulation protocol (such as VXLAN). This hardware rewrite operation transforms the original fast data packet into an enhanced data packet. The payload of the enhanced data packet remains unchanged and is still the original fast data, but its metadata fields (headers) now carry the context information assembled in step 103. Finally, step 106 is executed. The fast path forwarding unit immediately sends the enhanced data packet to the receiving end through a high-speed network (such as a 5G core network) without any CPU (slow path) intervention. This hardware rewrite mechanism of the fast path reduces the generation time of the enhanced data packet from milliseconds (software processing) to nanoseconds (hardware forwarding), which helps maintain the low latency and low jitter transmission characteristics of the fast data stream.

[0037] At the other end of the transmission link, the receiving end (such as the central controller) executes step 801 to receive the enhanced data packet. At the same time, it executes step 802 to parse the fast data from its payload and executes step 803 to parse the metadata header image from its preset metadata field, separating the slow data digest (if present), digest timestamp (if present), digest source count (if present), and a first or second state value. The downstream processing logic of the receiving end can switch strategies based on the deterministic signaling provided by the transport layer: when step 901 is executed, i.e., the first state value is parsed, the receiving end confirms that the transmission context is valid and immediately performs fused data processing based on the fast and slow data digests. At the same time, it can use the digest timestamp to evaluate the freshness and use the digest source count to compare with the locally configured expected number of sources, such as 100, to determine the completeness of the digest. When step 902 is executed, i.e., the second state value is parsed, the receiving end is explicitly informed of the current context at the transport layer. If a missing event occurs, such as a cache not being ready or a timing tear, the system will enforce a security policy based on the missing context, for example, switching to a security watch mode or refusing to execute critical control commands. In addition, to achieve proactive operation and maintenance awareness of the slow network transmission quality, this method may further include steps 501, 502 and 503. In the slow path processing unit of the first network node, the events of step 403, i.e., consistency verification failure, are counted to obtain a failure count value. In step 502, it is determined whether the failure count value exceeds a preset network degradation threshold within a preset time window (the window is set to 1 minute) (the threshold is set to 10 times). If it exceeds the threshold, in step 503, a network management alarm message is actively generated and sent. This can be an SNMPTrap message. The alarm message is used to indicate that the transmission quality of the low-speed network associated with the consistency digest group has deteriorated. This mechanism transforms the data plane verification failure event into a management plane QoS monitoring signal.

[0038] Example 1: In a transient stability control transmission scenario of a wide-area power system, control decisions rely simultaneously on fast data streams with millisecond-level responses and slow data streams with minute-level reporting. These two streams differ by several orders of magnitude in their transmission timing. In a specific operating condition, the first network node is a 5G edge gateway deployed at a regional aggregation point. It simultaneously connects to a 5G private network (high-speed network) for carrying PMU phasor measurement unit data and an NB-IoT network (low-speed network) for carrying the entire network topology and reserve capacity reporting. This gateway system includes a slow path processing unit, a fast path forwarding unit, and a stream buffer. During the stable operation phase of the power grid, at 14:30:00, the slow path processing unit receives multiple low-speed network data packets (slow data streams) from the NB-IoT network in the background. Based on a preset data fusion transmission strategy and summary generation rules, the slow path processing unit parses these data packets, performs summation and status list statistics, and generates a slow data summary. This summary contains information on the total available reserve capacity of the entire network (1500MW) and the connectivity of the critical section topology (ABC). Simultaneously, the slow path processing unit obtains the summary... The system requires a timestamp of 14:30:00.510 and a digest source count of 50 substations. To cope with fast packet storms, the slow path processing unit uses all the information it has acquired to pre-assemble a complete binary metadata header image conforming to the IP options field format in the background. This header image is then bound to a first status value representing valid data, 0x01, and both are pushed down and stored in the flow buffer associated with the fast path forwarding unit. This buffer entry is also linked to the five-tuple identifying the PMU fast data flow, such as the source IP segment and destination IP address. Related to the above, at 14:31:05.005, a serious line fault occurred in the regional power grid. The PMU deployed near the fault point immediately began sending fast data packets at a frequency of one packet every 20 milliseconds through the 5G private network. At 14:31:05.020, the first fast data packet arrived at the first network node, namely the 5G edge gateway. The fast path forwarding unit, which is an ASIC, immediately performed hardware flow matching on the fast data packet and hit the corresponding entry in the flow buffer within nanoseconds, retrieving the pre-assembled metadata header image.

[0039] At the hardware forwarding level, the fast path forwarding unit, without any CPU (i.e., slow path unit) intervention, directly uses the metadata header mirror to perform a hardware rewrite operation on the IP header option field of the fast data packet, forming an enhanced data packet. This rewrite operation is completed in nanoseconds, and the original low-latency transmission characteristics of the fast data packet are not affected. The fast path forwarding unit then sends the enhanced data packet to the high-speed network. At 14:31:05.090, 85 milliseconds after the fault occurred, the receiving end of the remote control center receives the enhanced data packet. In the same data packet, the receiving end simultaneously parses the fast data, i.e., the real-time phase offset of the fault point, from its payload, and parses the slow data digest from its IP header option field. This means that the entire network has a backup capacity of 1500MW and the topology ABC is connected, as well as the digest timestamp 14:30:00.510 and the first state value. The control center logic immediately confirms that the transmission context is valid, and performs a fusion decision based on the fast data and slow data digest. At 14:31:05.095, it issues a control command to disconnect the faulty line and start the backup unit. This method, through the coordination of fast and slow path architecture at network nodes, utilizes the background context preprocessing of the slow path and the hardware line-speed rewriting of the fast path to realize the context-based packet carrying of heterogeneous network data at the transmission layer. It provides a processing approach that balances timeliness and information integrity for digital information transmission tasks that require instantaneous response and rely on multi-source context.

[0040] Example 2: This example aims to objectively verify the effectiveness of the aforementioned technical solution in maintaining the low latency characteristics of fast data streams and ensuring the success rate of slow data digest packet transmission under extreme conditions of fast data packet storms through a transmission performance test. The test platform is built in a hardware-in-the-loop (HIL) simulation environment. The first network node, as the device under test, is an edge gateway prototype. Its internal system explicitly includes a slow path processing unit for performing control plane tasks, which is based on an ARM Cortex-A72 processor, and a fast path forwarding unit for performing data plane forwarding. Based on a Xilinx Zynq UltraScale+ FPGA, the stream buffer is deployed in the FPGA's on-chip memory. The test platform also includes: a high-speed network traffic generator and analyzer to simulate fast data streams, which consist of UDP packets marked with specific 5-tuples for matching and to measure their transmission performance; and a PC to simulate low-speed networks, sending slow data streams at 60-second intervals with ±1.5-second random jitter, which consist of CoAP packets, each carrying an incrementing sequence number as a slow data digest.

[0041] The experiment consisted of two test groups: one was the sample group of this invention, which fully implemented the fast and slow path coordination method disclosed in the aforementioned specific embodiments, i.e., the slow path processing unit generated metadata header images and pushed them down to the flow buffer, and the fast path forwarding unit performed hardware matching and hardware rewriting operations; the other was control group 1, which served as a partially missing control group, using the same hardware but with modified processing logic, forcing all fast data packets to be sent to the slow path processing unit, where the CPU performed digest retrieval and software rewriting of the data packets; the experimental process was as follows: the slow data stream generator ran for 10 minutes to ensure that the flow buffer of the sample group of this invention contained valid metadata header images; the high-speed network traffic generator started sending fast data streams, with the transmission load starting from 100Mbps and increasing sequentially to 500Mbps, 800Mbps, and 950Mbps, a rate close to 1Gbps line speed, with each load level lasting 60 seconds; the analyzer at the receiving end simultaneously measured and recorded the average end-to-end delay of fast data packets at each load level. Average latency jitter, and the success rate of correctly carrying the latest slow data digest in the data packet. .

[0042] Table 1: Performance Comparison Data Between Sample Group 1 and Control Group 1 under Different Transmission Loads

[0043]

[0044] Referring to Table 1, the experimental data shows that the transmission performance of the sample group of this invention is stable, with its average latency and average jitter remaining in the single-digit microsecond range, both less than 6.0 μs, and is basically unaffected by the increase in fast data load from 100 Mbps to 950 Mbps. At the same time, its context success rate remains at 100%, which confirms that the hardware rewriting mechanism of the fast path forwarding unit can complete metadata embedding at line speed without causing performance loss to the data plane. In contrast, the performance of control group 1 shows a typical CPU processing bottleneck: at a low load of 100 Mbps, its latency of 85.3 μs is significantly higher than that of the sample group of this invention; as the load increases, its average latency and jitter deteriorate sharply. At a load of 950 Mbps, the average latency reaches 7500.9 μs, and the context success rate drops to 80.5%. This is because the CPU resources of the slow path processing unit are exhausted by the software rewriting processing of high-concurrency data packets, making it unable to process all data packets in time, and affecting the normal update and acquisition of the background slow data digest.

[0045] Example 3: This example is a comparative example, employing a conventional transmission method that relies on receiver buffer alignment, and placing it under the same transmission scenario and fault conditions as Example 1. In this method, the first network node is a standard forwarding device that does not perform any fast / slow path collaborative processing or digest embedding of this invention. Instead, it independently forwards the fast data stream and slow data stream to the receiver at the layer level. The receiver is responsible for performing data fusion. At 14:30:00.510, the slow data stream in the NB-IoT network is sent by the source. At 14:31:05.005, a power grid fault occurs. At 14:31:05.020, the fast data stream, i.e., the PMU packet, is sent by the source.

[0046] At 14:31:05.090, the receiving end received the fast data packet via the high-speed network. The receiving end's fusion logic was triggered, but its local cache query revealed that the latest slow data digest timestamp was 14:25:00, which was data from the previous cycle, indicating an outdated transmission context. Therefore, the receiving end logic had to enter a waiting state, placing the fast data packet in a buffer and waiting for the slow data packet with a network transmission timestamp of 14:30:00.510 to arrive. Due to the transmission latency of the low-speed network, the slow data packet with a timestamp of 14:30:00.510 did not arrive until 14:31:10.000. Upon reaching the receiving end, the transmission time was close to 70 seconds. At 14:31:10.005, the receiving end fusion logic finally obtained the required slow data digest, i.e., 1500MW capacity, and made a fusion decision with the fast data in the buffer, i.e., the phase offset. At 14:31:10.010, the control center issued a control command. This result shows that, using the conventional receiving end buffering method, the millisecond-level transmission timeliness of the fast data stream was offset by the transmission delay of the slow data stream, resulting in the overall response time of the transmission link being extended from 85 milliseconds in Example 1 to 5.01 seconds, which could not meet the instantaneous requirements of transient stability control for transmission.

[0047] Example 4: This example combines Figs. 1 to 3 This describes a multi-network converged communication data processing method and system, such as... Fig. 1As shown, its core lies in the collaborative processing of fast and slow paths. The slow path processing unit receives slow data streams from external entities in the low-speed network in the background and performs background processing to generate digests and assemble header images. The metadata header image, containing a digest, status value, timestamp, and source count, is stored in the A1: flow buffer. The fast path forwarding unit receives fast data packets from external entities in the high-speed network, performs metadata header image retrieval from the A1: flow buffer, and rewrites and forwards them through hardware line-speed matching, forming an enhanced data packet with fast data payload + slow data header, which is sent to the receiving end. When the receiving end parses the enhanced packet and makes decisions, it performs fused data processing based on the first status value or initiates a security policy based on the second status value, depending on the parsing result (status one or status two).

[0048] like Fig. 2 As shown, the horizontal axis represents the maximum permissible time deviation in seconds, with data points ranging from 5 seconds to 60 seconds. This chart reveals the performance trade-offs through two curves, both corresponding to changes in the horizontal axis: The first curve, a solid line, corresponds to the left vertical axis, context success rate (%). This curve shows that the context success rate increases non-linearly with the relaxation of the maximum permissible time deviation, rapidly increasing from 95% at 5 seconds to 99% at 10 seconds, and finally reaching 100% at 60 seconds. The second curve, a dashed line, corresponds to the right vertical axis, average processing latency in μs, with a scale range from 0 to 15 μs. This curve shows that the average processing latency also increases almost steadily with the increase of the maximum permissible time deviation, increasing from 4.8 μs at 5 seconds to 12.5 μs at 60 seconds.

[0049] like Fig. 3 As shown, the process begins with the slow path processing unit receiving slow data packets (NB-IoT / CoAP) from the low-speed network. Step 102 is executed: generating a digest, including executing aggregation algorithms such as AVG / SUM to generate a slow data digest, such as 150.75MW. In parallel, in steps 201-202: maintaining active sources, the unit interacts with the active source list by extracting source addresses, updating the active list, and counting the number of active sources that have not timed out, to return a digest source count, such as 98. Then, the unit queries the system clock for the current time and returns a digest timestamp, such as 14:30:05.120. Finally, in step 103: successful acquisition scenario, the unit assembles the digest, timestamp, source count, and first state value 0x01 into a metadata header image, and stores it in the stream cache by storing the complete metadata header image. After receiving a successful storage response, the image is ready and waiting for the fast path call.

[0050] Example 5: This example illustrates the specific engineering procedures for offline calibration and configuration of key strategy parameters in the data fusion transmission strategy on the slow path processing unit of the first network node, i.e., the edge gateway, to solve the technical problem of lack of objective basis for parameter setting. Before system deployment, one of the challenges is to set a reasonable maximum allowable time deviation for the consistency digest group. This parameter is related to the timing security and availability of the data delivered by the transport layer. The determination of this parameter does not depend on network testing, but is determined by the business logic of the downstream application, i.e., the receiving end. Taking the power grid scenario in Example 1 as an example, if the digest group is used for transient stability decision, its business layer has a low tolerance for context timing tearing, and the maximum allowable time deviation is set to a small value, such as 8 seconds. If the digest group is only used for daily load curve statistics, the business tolerance is higher, and the value can be relaxed to 60 seconds. In this example, for high timeliness requirements, the maximum allowable time deviation is determined and configured to be 8 seconds.

[0051] Therefore, a reproducible procedure is needed to determine the network degradation threshold to avoid false or missed HOE4L alarms. To this end, a 24-hour network baseline calibration test is performed on the first network node. During this period, the system only runs the consistency check logic and records failure events. At this time, the low-speed network, i.e., the NB-IoT network, is in a normal, non-congested operating state. The test log shows that 52 consistency check failure events were recorded in 24 hours, an average of 2.17 events per hour. This data reflects the inherent, random transmission jitter characteristics of the low-speed network. To ensure the effectiveness of the alarm, the network degradation threshold is set to a statistically significant deviation from the baseline. It can be calculated by adding three times the baseline standard deviation to the baseline mean or by a simplified engineering value, such as setting it to a cumulative failure count exceeding 12 within 15 minutes, i.e., 4.6 times the baseline average rate. Once transmission quality degradation causes the failure rate to trigger this threshold, step 5 is initiated. The network management alarm message 03, and to ensure the topology awareness accuracy of the digest source count, has its active source list lifetime parameter calibrated to 1.5 times the maximum reporting cycle of the terminal devices (i.e., NB-IoT modules) in the slow data stream. Given that the longest sleep reporting cycle set by the devices in this network is 10 minutes, the lifetime is configured to 15 minutes to ensure that the devices remain in the active list even if a regular packet loss occurs. Finally, the selection of the digest generation rule is also configured according to the downstream business requirements. If the business needs to obtain the average load of the slow data stream, the rule is configured as the AVG algorithm. If the business needs to perceive the status list of all terminals in the slow data stream, the rule is configured as status list statistics. Through the above calibration procedure, the data fusion transmission strategy of the slow path processing unit has obtained a determined parameter configuration, namely, the maximum allowable time deviation is 8 seconds, the network degradation threshold is 12 times / 15 minutes, and the source lifetime is 15 minutes.

[0052] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0053] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A multi-network converged communication data processing method, characterized in that, This method is executed at a first network node in a multi-network converged communication system to process data sent to a receiving end. The first network node includes a slow path processing unit, a fast path forwarding unit, and a first-order buffer associated with the fast path forwarding unit. The method includes: Step 101: The slow path processing unit configures the data fusion transmission strategy. The strategy defines the logical relationship between the fast data stream transmitted over the high-speed network and the slow data stream transmitted over the low-speed network, and defines the summary generation rules for the slow data stream. Step 102: The slow path processing unit attempts to obtain a slow data summary in the background according to the summary generation rules. Step 103 is executed by the slow path processing unit: when the slow data digest is successfully obtained, the slow data digest and the first state value are assembled into a metadata header image; when the slow data digest is not obtained, the second state value is assembled into a metadata header image, and the metadata header image is stored in the flow buffer associated with the fast path forwarding unit. Step 104: The fast path forwarding unit monitors the fast data stream in real time and performs flow matching on the fast data packets that conform to the data fusion transmission strategy. Step 105: Based on the result of flow matching, the fast path forwarding unit retrieves the metadata header image from the flow buffer, and at the hardware forwarding level, uses the metadata header image to rewrite one or more preset metadata fields of the fast data packet to form an enhanced data packet. Step 106: The fast path forwarding unit sends the enhanced data packet to the receiving end through the high-speed network.

2. The multi-network converged communication data processing method according to claim 1, characterized in that, The method also includes: step 201, at the first network node, monitoring the source address of data packets in the slow data stream in real time to maintain an active source list; step 202, counting the number of active source ends in the active source list to obtain a digest source count; and in step 103, when a slow data digest is successfully obtained, assembling the digest source count, the slow data digest, and the first status value together into a metadata header image.

3. The multi-network converged communication data processing method according to claim 1, characterized in that, Step 102 further includes: using the generation time of the slow data digest as the digest timestamp, and in step 103, when the slow data digest is successfully obtained, assembling the digest timestamp, the slow data digest, and the first state value into a metadata header image.

4. The multi-network converged communication data processing method according to claim 3, characterized in that, The data fusion transmission strategy defines a consistent digest group including at least two slow data digests; step 102 includes: independently generating at least two slow data digests and their respective corresponding digest timestamps in the background; step 103, before assembling the metadata header image, further includes: step 401, obtaining at least two digest timestamps; step 402, calculating the time difference between at least two digest timestamps. Step 403, when the time difference When the time difference exceeds the preset maximum allowable time deviation, the second state value is assembled into a metadata header image; step 404, when the time difference... At least two slow data digests and the first state value will be assembled into a metadata header image only when the time deviation is less than or equal to the preset maximum allowable time deviation.

5. The multi-network converged communication data processing method according to claim 4, characterized in that, The method further includes: step 501, processing the time difference. Events exceeding the preset maximum allowable time deviation are counted to obtain a failure count value; step 502, it is determined whether the failure count value exceeds the preset network degradation threshold within the preset time window; step 503, when the failure count value exceeds the preset network degradation threshold, a network management alarm message is generated and sent, the alarm message is used to indicate that the transmission quality of the low-speed network associated with the consistency summary group has deteriorated.

6. The multi-network converged communication data processing method according to claim 1, characterized in that, The summary generation rule is a non-predictive aggregation algorithm selected from at least one of summation, counting, averaging, and state list statistics for slow data streams.

7. The multi-network converged communication data processing method according to claim 1, characterized in that, The first network node is an edge gateway, router, 5G base station, or software-defined network (SDN) switch; the preset metadata field is the IP header options, TCP header options, or extended header of the encapsulation protocol of the fast data packet. The method also includes the following steps performed at the receiving end: step 801, receiving the enhanced data packet; step 802, parsing the fast data from the payload of the enhanced data packet; step 803, parsing the slow data digest, first status value, or second status value from the preset metadata field of the enhanced data packet.

8. The multi-network converged communication data processing method according to claim 7, characterized in that, The method further includes: step 901, when the first state value is parsed, performing fusion data processing based on the fast data and slow data digest; step 902, when the second state value is parsed, executing a security policy based on missing context.

9. A multi-network converged communication data processing method according to claim 2, characterized in that, In step 103, when the slow data digest is successfully obtained, the digest timestamp, digest source count, slow data digest and first state value are assembled into a metadata header image.

10. A multi-network converged communication data processing system, used to implement the multi-network converged communication data processing method according to claim 1, characterized in that the system... include: Slow path processing unit, fast path forwarding unit, and first-class cache associated with fast path forwarding unit; The slow path processing unit is used to: configure data fusion transmission strategies, which define the logical correlation between fast data streams transmitted over high-speed networks and slow data streams transmitted over low-speed networks, and define the digest generation rules for slow data streams; attempt to obtain slow data digests in the background according to the digest generation rules; and assemble the slow data digests and the first status value into a metadata header image when the slow data digests are successfully obtained. When obtaining a slow data digest fails, a second status value is assembled into a metadata header image; And store the metadata header image in the stream cache; The fast path forwarding unit is used to: monitor fast data streams in real time, and perform flow matching on fast data packets that conform to the data fusion transmission strategy upon receipt; Based on the results of stream matching, retrieve the metadata header image from the stream cache; At the hardware forwarding level, metadata header mirroring is used to rewrite one or more preset metadata fields of the fast data packet to form an enhanced data packet; Enhanced data packets are sent to the receiving end via a high-speed network.

Images