Intelligent multi-channel fusion communication dispatching method based on edge computing

Abstract

The present application relates to the technical field of communication network resource scheduling, in particular to a kind of intelligent multi-channel converged communication scheduling method based on edge computing.The present application includes the following steps: multi-source data acquisition and preprocessing;Fragment group level scheduling strategy generation;Multi-channel grouping collaborative transmission execution;Scheduling effect verification and parameter updating.The present application collects the transmission dependency data between service fragments, constructs transmission dependency fragment group, and configures group-level transmission constraints and failure determination rules, which upgrades the scheduling unit from a single fragment to a fragment group that needs to be effectively guaranteed as a whole;According to the overall transmission result of the completed fragment group, dynamically adjust the group-level matching relationship, and update the channel state reference data and group-level matching degree calculation parameters.Thereby, it can reduce the risk of service analysis failure caused by key fragment loss, group time sequence overrun and whole group packet loss, and improve the reliability and adaptive scheduling capability of multi-channel converged communication in edge heterogeneous network.

Description

Technical Field

[0001] This invention relates to the field of communication network resource scheduling technology, and more specifically, to an intelligent multi-channel converged communication scheduling method based on edge computing. Background Technology

[0002] With the continuous development of edge computing and heterogeneous communication network technologies, multi-channel converged communication scheduling methods based on edge nodes have been widely used in fields such as the Industrial Internet, the Internet of Vehicles, the Internet of Power, and real-time multimedia transmission. This type of technology achieves multi-channel collaborative transmission and optimized resource allocation by real-time sensing of the operational status of multiple communication channels at the edge and making scheduling decisions based on the transmission requirements of business data.

[0003] Existing multi-channel converged communication scheduling schemes mainly include two typical technical paths: The first is a static or quasi-static scheduling method based on single-shard granularity. This type of method typically uses a single service shard as the scheduling object, performs matching calculations based on operational indicators such as bandwidth, latency, and packet loss rate of the communication channel, and allocates the shard to the optimal or suboptimal channel for transmission. This method is simple to implement and suitable for low-complexity service scenarios. The second is an adaptive scheduling method based on channel state feedback. This type of method introduces a transmission feedback mechanism on the basis of single-shard scheduling. By statistically analyzing the latency and packet loss of completed data transmissions, it dynamically updates the state parameters or weights of the communication channel, thereby improving the system's adaptability to network fluctuations to a certain extent. Both of these methods have been implemented in engineering applications in some edge communication systems, offering advantages such as low implementation cost and high deployment flexibility.

[0004] However, in complex business scenarios, business data often consists of multiple fragments with temporal correlations and business dependencies. These fragments exhibit overall constraints in terms of transmission order, parsing structure, and integrity. Simultaneously, when heterogeneous communication channels work in multi-channel collaboration, their actual transmission capacity is affected by link interference, resource contention, and dynamic fluctuations, exhibiting significant nonlinearity and combinatorial coupling characteristics. Against this backdrop, existing technologies have gradually revealed two common technical problems: First, the fragment dependencies within business data are not uniformly modeled at the scheduling level, resulting in scheduling objects remaining at the single-fragment granularity, failing to reflect the overall effectiveness constraints of the business, thus affecting the overall transmission consistency of complex businesses. Second, feedback mechanisms primarily operate at the channel state or single-fragment result level, lacking the ability to provide unified feedback and coordinated correction for sets of business fragments with inherent relationships, making it difficult for scheduling strategies to reflect the overall transmission behavior characteristics at the batch level. In engineering practice, these problems easily lead to inconsistencies in cross-channel transmission timing, mismatch of critical fragments, and business parsing failures, thereby affecting the overall service quality and stability of the system. In view of this, we propose an intelligent multi-channel converged communication scheduling method based on edge computing. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent multi-channel converged communication scheduling method based on edge computing, in order to solve the problems in existing multi-channel converged communication scheduling methods, such as the failure to uniformly model the service fragmentation dependency relationship at the scheduling level, the scheduling object still remaining at the single fragment granularity and making it difficult to reflect the overall effectiveness constraints of the service, and the existing feedback mechanism mainly acting on the channel status or single fragment transmission result level, making it difficult to uniformly provide feedback and linkage correction for service fragment sets with inherent correlation.

[0006] To address the aforementioned technical problems, this invention provides an intelligent multi-channel fusion communication scheduling method based on edge computing, comprising the following steps: S1. Multi-source data acquisition and preprocessing: The operation status data, service fragment data, and transmission dependency data between service fragments are acquired through edge computing nodes, and the operation status data, service fragment data, and transmission dependency data between service fragments are preprocessed. S2. Fragment Group-Level Scheduling Strategy Generation: Based on the preprocessed service fragmentation data and the preprocessed service fragmentation data, transmission dependency fragment groups are obtained. The matching relationship between the transmission dependency fragment groups and the multi-path heterogeneous communication channels is determined according to the preprocessed multi-path heterogeneous communication channel operation status data. The matching relationship is adjusted based on the overall transmission results of the completed transmission dependency fragment groups to generate a fragment group-level multi-channel converged communication scheduling strategy; where: The transmission-dependent fragmentation group refers to a multi-channel scheduling basic unit composed of several service fragments with transmission dependencies, which needs to ensure transmission effectiveness as a whole. S3. Multi-channel group collaborative transmission execution: Based on the multi-channel converged communication scheduling strategy at the fragment group level, the multi-channel heterogeneous communication channels are coordinated and managed. The multi-channel diversion and transmission of business fragment data is executed in units of transmission-dependent fragment groups, and the overall transmission results of the transmission-dependent fragment groups that have completed transmission are collected. S4. Scheduling effect verification and parameter update: Based on the overall transmission results of the transmission-dependent fragment groups that have completed transmission, verify the execution effect of the fragment group-level multi-channel converged communication scheduling strategy, and update the baseline data of the operating status of the multi-path heterogeneous communication channels and the basic parameters for calculating the group-level matching degree between the transmission-dependent fragment groups and the multi-path heterogeneous communication channels.

[0007] As a further improvement to this technical solution, in step S1, the operating status data of multiple heterogeneous communication channels, service fragmentation data, and transmission dependency data between service fragments are preprocessed, including the following steps: S1.1 Perform outlier removal and missing value filling on the operation status data, service fragmentation data, and transmission dependency data between service fragments of the multi-channel heterogeneous communication channels to obtain cleaned basic data; S1.2. The cleaned multi-channel heterogeneous communication channel operation status data, service fragment data, and transmission dependency data between service fragments are timestamped and unified to the same time series reference. S1.3 Perform dimension normalization processing on the time-aligned multi-channel heterogeneous communication status data, and perform structured encoding processing on the service fragment data and the transmission dependency data between service fragments. S1.4. The normalized multi-channel heterogeneous communication status data, the structured encoded service fragment data, and the transmission dependency data between service fragments are encapsulated in chronological order into a preprocessed dataset.

[0008] As a further improvement to this technical solution, in step S2, the generation of the shard group-level scheduling strategy includes the following steps: S2.1 Based on the transmission dependency data between preprocessed service fragments and the preprocessed service fragment data, several independent transmission dependency fragment groups are obtained according to the service transmission failure association dimension. S2.2 For each transmission-dependent fragment group, extract and configure the corresponding group-level transmission constraint parameters and group-level transmission failure judgment rules, and use them as the pre-constraint conditions for matching calculation. S2.3. Using group-level transmission constraint parameters as prerequisites, the initial matching relationship between transmission-dependent fragment groups and multiple heterogeneous communication channels is calculated based on the preprocessed operating status data of the multi-channel heterogeneous communication channels. S2.4. Based on the group-level transmission failure judgment rule, dynamically adjust the matching relationship between the same type of transmission dependent fragment group to be transmitted in the same batch and the corresponding multi-path heterogeneous communication channel according to the overall transmission result of the transmission dependent fragment group that has been transmitted. S2.5 Based on the adjusted matching relationship between the transmission dependency fragment group and the multiple heterogeneous communication channels, a fragment group-level multi-channel fusion communication scheduling strategy is generated in a structured manner according to the transmission timing sequence.

[0009] As a further improvement to this technical solution, step S2.1, which divides the data into several independent transmission-dependent fragment groups according to the service transmission failure correlation dimension, includes the following steps: S2.11. Parse the preprocessed service fragment data and extract the frame type identifier, timing number and service semantic attributes of each service fragment; S2.12. Based on the preprocessed transmission dependency data between service fragments, identify the set of service fragments with transmission failure associations; where: The transmission failure association refers to the fact that any abnormal transmission of any critical service fragment within the set will cause the entire set of service data to become unavailable. The critical business segment refers to the business segment that carries the core business semantics of the business segment set and determines the parseability and availability of the entire set of business data. Its corresponding attributes can be identified and determined through the frame type identifier and business semantic attributes of the business segment. S2.13. Based on the timing sequence number and semantic attributes of the service fragments, encapsulate the set of service fragments with transmission failure associations into independent transmission-dependent fragment groups.

[0010] As a further improvement to this technical solution, step S2.2, extracting and configuring the corresponding group-level transmission constraint parameters and group-level transmission failure determination rules, includes the following steps: S2.21. Based on the service type of the transmission-dependent fragmentation group, determine the three group-level transmission constraint parameters: the maximum allowable end-to-end delay of the entire group, the maximum allowable transmission timing difference of fragments within the group, and the maximum allowable number of packet losses in the entire group. S2.22. Set group-level transmission failure judgment rules: When any one of the following conditions is met, the transmission failure of the entire transmission dependent fragment group is judged to be a transmission failure. S2.23. Associate the group-level transmission constraint parameters and group-level transmission failure judgment rules as pre-constraint conditions with the attribute set of the corresponding transmission-dependent fragment group.

[0011] As a further improvement to this technical solution, step S2.3, calculating the initial matching relationship between the transmission-dependent fragment group and the multiple heterogeneous communication channels, includes the following steps: S2.31. Extract the available bandwidth attribute, average transmission delay attribute, steady-state packet loss rate attribute, and channel fluctuation amplitude attribute of the multi-channel heterogeneous communication channels from the preprocessed operation status data of the multi-channel heterogeneous communication channels. S2.32. Extract the total data size, importance level, latency sensitivity level, and group-level transmission constraint parameters of the transmission-dependent fragment group from the attribute set of the transmission-dependent fragment group. S2.33. Using group-level transmission constraint parameters as hard constraint boundaries, calculate the initial matching degree values ​​of each transmission-dependent fragment group and multiple heterogeneous communication channels through multi-attribute weighted matching operations. The initial matching relationship between the transmission-dependent fragment group and multiple heterogeneous communication channels is formed by all the initial matching degree values.

[0012] As a further improvement to this technical solution, step S2.4, dynamically adjusting the matching relationship between subsequent transmission-dependent fragment groups and multiple heterogeneous communication channels, includes the following steps: S2.41. Obtain the overall transmission result of the transmission-dependent fragment group that has been transmitted. Extract three types of feedback data: the overall transmission failure status, the timing difference deviation value within the group, and the overall packet loss deviation value. S2.42. Taking the overall transmission failure state as the core correction trigger condition, and combining the timing difference deviation value within the group and the overall packet loss deviation value, calculate the matching degree correction amount of the corresponding multi-path heterogeneous communication channels. S2.43 Under the dual constraints of the preset upper limit of the correction step size and the maximum attenuation amplitude, the matching degree correction amount is used to update the matching degree value of the same type of transmission dependent fragment group to be transmitted in the same batch and the corresponding multi-path heterogeneous communication channel, so as to complete the dynamic adjustment of the matching relationship between the transmission dependent fragment group and the multi-path heterogeneous communication channel.

[0013] As a further improvement to this technical solution, step S2.5, which involves structurally generating a fragment group-level multi-channel fusion communication scheduling strategy according to the transmission timing sequence, includes the following steps: S2.51. Based on the adjusted matching relationship between the transmission dependency fragment group and the multiple heterogeneous communication channels, match the main transmission channel and the alternative transmission channel for each transmission dependency fragment group. S2.52. Configure the overall transmission priority, offloading ratio, and fragmentation timing coordination rules within each transmission-dependent fragmentation group; S2.53. According to the transmission timing order of the transmission-dependent fragment groups, the channel matching results, overall transmission priority, splitting ratio and fragment timing coordination rules within the group of all transmission-dependent fragment groups are structurally encapsulated to generate a fragment group-level multi-channel fusion communication scheduling strategy.

[0014] As a further improvement to this technical solution, in step S3, the multi-channel packet cooperative transmission execution includes the following steps: S3.1 Analyze the multi-channel converged communication scheduling strategy at the fragment group level, extract the main transmission channel, alternative transmission channel, overall group transmission priority, diversion ratio and fragment timing coordination rules within the group corresponding to each transmission-dependent fragment group, and complete the transmission pre-configuration of multiple heterogeneous communication channels; S3.2. According to the priority of the whole group transmission, the service fragment data is distributed to the corresponding multi-way heterogeneous communication channels for offloading transmission, based on the transmission-dependent fragment group and the fragment timing coordination rules within the group. S3.3 After all transmission dependent fragments in a single transmission group have been transmitted, collect the overall transmission results of the completed transmission dependent fragment group and synchronously send the overall transmission results back to S2 and S4 for matching relationship adjustment and scheduling effect verification.

[0015] As a further improvement to this technical solution, step S4, the scheduling effect verification and parameter update, includes the following steps: S4.1 Based on the overall transmission results of the transmission-dependent fragment groups that have completed transmission, statistically analyze three types of performance indicators: the overall group transmission failure rate, the intra-group transmission timing difference compliance rate, and the overall group packet loss compliance rate. Verify the execution effect of the fragment group-level multi-channel fusion communication scheduling strategy and generate performance verification results. S4.2. Based on the effect verification results, update the operating status benchmark data of the multi-channel heterogeneous communication channels, and use the updated operating status benchmark data of the multi-channel heterogeneous communication channels as the benchmark for the next round of operating status data collection and evaluation of the multi-channel heterogeneous communication channels. S4.3. Based on the effect verification results, update the basic parameters for calculating the group-level matching degree of the transmission-dependent fragment group and the multi-path heterogeneous communication channel, and use the updated basic parameters for calculating the group-level matching degree of the transmission-dependent fragment group and the multi-path heterogeneous communication channel as the basis for the benchmark weight of the next round of group-level matching degree calculation.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention collects transmission dependency data between service fragments and divides the pre-processed service fragment data into transmission dependency fragment groups. This elevates the basic scheduling unit of multi-channel converged communication from a single service fragment to a transmission dependency fragment group that requires overall transmission effectiveness assurance. Simultaneously, by configuring group-level transmission constraint parameters and group-level transmission failure judgment rules, the scheduling process can uniformly model the timing, parsing, and integrity dependencies between service fragments. Therefore, it avoids the problem of existing single-fragment granular scheduling ignoring overall service effectiveness constraints, reduces the risk of entire group service parsing failures caused by factors such as critical service fragment loss and exceeding intra-group transmission timing differences, and improves the integrity, consistency, and reliability of multi-channel transmission in complex service scenarios. 2. This invention calculates the group-level matching relationship between transmission-dependent fragment groups and multiple heterogeneous communication channels. Based on the overall transmission results of transmission-dependent fragment groups that have completed transmission, it dynamically adjusts the matching relationship between transmission-dependent fragment groups of the same type to be transmitted within the same batch and their corresponding communication channels. Simultaneously, during the scheduling effect verification phase, it updates the baseline data of the operating status of multiple heterogeneous communication channels and the basic parameters for calculating the group-level matching degree. This ensures that transmission feedback is no longer limited to the level of a single channel status or a single fragment result, but can influence the fragment group-level scheduling decision-making process. Therefore, it improves the scheduling strategy's responsiveness to group-level transmission failures, intra-group timing deviations, and overall group packet loss deviations, enhancing the adaptability and continuous optimization capabilities of multi-channel converged communication scheduling in the fluctuating environment of edge heterogeneous networks. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall process flow of the present invention. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] like Figure 1As shown, this embodiment provides an intelligent multi-channel converged communication scheduling method based on edge computing, applied to a communication system consisting of edge computing nodes, multiple heterogeneous communication channels, a service data fragmentation module, and a transmission status acquisition module. The edge computing nodes are used to uniformly collect, preprocess, and make scheduling decisions regarding the operating status of each communication channel, service fragment data, and transmission dependencies between service fragments; the multiple heterogeneous communication channels are used to carry the collaborative transmission of different service fragments or fragment groups. This method uses transmission-dependent fragment groups as the basic unit of multi-channel scheduling, combining group-level transmission constraints, group-level failure judgment rules, and overall transmission result feedback to generate and dynamically optimize fragment group-level multi-channel converged communication scheduling strategies, thereby improving the transmission integrity, reliability, and adaptive scheduling capability of complex service data in an edge heterogeneous network environment. The method includes the following steps: S1. Multi-source Data Acquisition and Preprocessing: Edge computing nodes acquire operational status data, service fragmentation data, and transmission dependency data between service fragments from multiple heterogeneous communication channels. Preprocessing is then performed on this data. Specifically, edge computing nodes establish periodic data reporting links with the channel status monitoring agent of the multiple heterogeneous communication channels via a southbound interface, simultaneously obtaining the segmented service fragmentation data and corresponding inter-fragment transmission dependency descriptions from the service data fragmentation module. Before entering the scheduling calculation process, these three types of data undergo a pipelined processing of four sub-steps: cleaning, alignment, normalization and encoding, and encapsulation. This process eliminates the impact of noise, time deviations, and dimensional differences in the original data on subsequent matching calculations. The specific process is as follows: S1.1 Perform outlier removal and missing value filling on the operation status data, service fragmentation data, and transmission dependency data between service fragments of the multi-channel heterogeneous communication channels to obtain cleaned basic data; In this step, the channel operation status data is processed first. Edge computing nodes acquire a sequence of operation status indicators from each channel monitoring agent according to a preset collection cycle. These indicators typically include available bandwidth (Mbps), average transmission latency (ms), steady-state packet loss rate (dimensionless ratio or percentage), and channel fluctuation amplitude (dimensionless). Channel fluctuation amplitude characterizes the stability of channel transmission performance and is quantified as the normalized coefficient of variation of the average transmission latency within a preset statistical window, i.e., the ratio of the latency standard deviation to the latency mean. For each indicator, an anomaly detection method based on historical statistics is used: the mean value of that indicator for that channel within a preset historical time window is taken. and standard deviation If the value of the current collection point satisfy If a value is not found to be true, it is considered abnormal and removed. Gaps caused by removal, as well as gaps due to reporting timeouts or missing fields, are filled using linear interpolation between two adjacent valid data points before and after that time. This interpolation is only used to maintain the continuity of the time series and is not used as an accurate true value.

[0020] The cleaning of business fragmented data mainly checks for two types of structural issues. First, it checks whether the value of the frame type identifier field falls within the set of valid enumerated values ​​defined by the communication protocol. If the value is invalid, the entire fragment record is discarded. Second, it checks whether the value of the fragment payload length field matches the actual payload in bytes. If they do not match, the fragment record is also discarded. The discarded fragment records are synchronously written to the system log, along with the original timestamp and fragment identifier, for subsequent location.

[0021] Dependency data exists as ordered pairs between shards, with each record specifying a preceding shard identifier and a following shard identifier. During cleaning, all dependency records are traversed. If the preceding or following shard referenced by a record does not have a corresponding entity in the current batch of shard data, the dependency is determined to be an isolated record and directly removed, while a data integrity exception event notification is generated.

[0022] Through the above operations, errors and invalid records introduced by acquisition noise, transmission jitter, module malfunctions, etc. in the three types of data are removed, and the cleaned basic data is obtained.

[0023] S1.2. The cleaned multi-channel heterogeneous communication channel operation status data, service fragment data, and transmission dependency data between service fragments are timestamped and unified to the same time series reference. In this step, monitoring agents for multiple heterogeneous communication channels are deployed at the link or network layer of different channels, each maintaining its own local clock. The timestamps carried in the reported operational status data are the local time of each agent. The business data sharding module uses another set of local clocks. These clocks have a fixed or gradually varying deviation from the system clock of the edge computing nodes. To eliminate timing misalignment caused by different clock domains, the local system clock of the edge computing nodes is used as a unified timing reference, and the following alignment strategy is adopted: the edge computing node is used as each external data source. Maintain a clock skew This deviation is calculated by periodically sending bidirectional time synchronization probe messages and is continuously updated based on the most recent calculation result. During alignment, the original timestamps of each data source are used. Convert to a unified reference time .

[0024] After timestamp conversion, all data records are grouped into a pre-defined unified time window grid based on a unified reference time. The time window length of this grid is... The configuration can be adjusted based on the channel status change rate and service latency sensitivity, with a typical value of 10ms. Within the same time window, if a certain indicator has multiple numerical records, the arithmetic mean of these records is taken as the representative value of that indicator for that window. If there are no records in a window, the representative value of the previous window is kept unchanged, i.e., zero-order hold is used for filling until a new valid data point appears. After this processing, the three types of data are completely aligned to the same time series benchmark in the time dimension.

[0025] S1.3 Perform dimension normalization processing on the time-aligned multi-channel heterogeneous communication status data, and perform structured encoding processing on the service fragment data and the transmission dependency data between service fragments. In this step, the dimensions and numerical ranges of different indicators in the channel operation status data differ significantly. This difference in dimensions can cause indicators with larger numerical ranges to inappropriately dominate the weighting in subsequent multi-attribute weighted matching operations. Therefore, each indicator is independently subjected to Min-Max linear normalization.

[0026] Let the original value of a certain indicator in a certain channel be... The minimum value of this indicator within the preset historical statistical period is The maximum value is The normalized value Calculate using the following formula: ; in and The data is updated periodically by the edge computing nodes based on the historical operational data of this channel. (After normalization) The value range is [0, 1]. To avoid zero values ​​after normalization, which could cause anomalies in subsequent division or logarithmic operations, a base parameter is set. (For example, take) ),when At that time, Revised to .

[0027] The structured encoding of service fragment data involves organizing discrete fields such as frame type identifier, sequence number, service semantic attributes, fragment payload length, and fragment identifier of each fragment into a structured record according to a predefined key-value pair template, with fields separated by agreed-upon delimiters. The structured encoding of transmission dependency data encodes each dependency record as a triple (src_id, dst_id, dep_type), where src_id is the source fragment identifier, dst_id is the target fragment identifier, and dep_type is the dependency type field used to distinguish the nature of the dependency, such as decoding dependency, synchronization dependency, or semantic association dependency. After encoding, both types of service data have a unified internal representation format that is easily parsed by machines.

[0028] S1.4. The normalized multi-channel heterogeneous communication status data, the structured encoded service fragment data, and the transmission dependency data between service fragments are encapsulated in chronological order into a preprocessed dataset.

[0029] In this step, using the unified baseline timestamp generated in S1.2 as the primary index key, normalized channel operation status data, structured business sharding data, and structured transmission dependency data belonging to the same time window are organized into a preprocessed data record. All preprocessed data records are arranged in ascending order of timestamp, forming a time-ordered preprocessed dataset. This dataset is stored in the shared memory area of ​​the edge computing node in the form of a structured message queue, and provides data consumption services to the upper-layer scheduling module through a preset internal data access interface. The interface's read operation supports data retrieval by time window number or timestamp range, and downstream modules (such as the sharding group partitioning module and the matching calculation module) no longer need to concern themselves with the collection details and format differences of the original data.

[0030] Finally, after the overall processing of S1, the original data, which came from multiple clock domains, had different dimensions and formats, was transformed into a standardized dataset that was time-aligned, numerically normalized, structurally unified and sequentially encapsulated. This provided reliable and consistent data input for the fragment grouping based on transmission dependencies and the channel adaptation based on multi-attribute matching operations in S2.

[0031] S2. Segment Group-Level Scheduling Strategy Generation: Based on the preprocessed service segment transmission dependency data and the preprocessed service segment data, transmission dependency segment groups are obtained. The matching relationship between the transmission dependency segment groups and the multi-heterogeneous communication channels is determined according to the preprocessed multi-channel heterogeneous communication status data. The matching relationship is adjusted based on the overall transmission results of the completed transmission dependency segment groups, generating a segment group-level multi-channel converged communication scheduling strategy. Specifically, this step is executed by the scheduling strategy generation module in the edge computing node. A scheduling calculation is triggered after each service batch's data preprocessing is completed. The preprocessed dataset output from S1 is received as input. Using service batches as scheduling units (a service batch refers to a set of continuous service segments grouped according to a preset time window or preset number of segments), the scheduling strategy is generated through five stages: dividing transmission dependency segment groups, configuring group-level constraints and failure rules, calculating the initial matching relationship, dynamically adjusting the matching relationship based on the overall transmission results of the transmitted segment groups, and structurally encapsulating the scheduling strategy. This strategy is then distributed to S3. S2 includes the following sub-steps: S2.1. Based on the preprocessed transmission dependency data between service fragments and the preprocessed service fragment data, several independent transmission dependency fragment groups are obtained according to the service transmission failure correlation dimension; specifically, the following steps are included: S2.11. Parse the preprocessed service fragment data and extract the frame type identifier, timing number and service semantic attributes of each service fragment; In this step, the scheduling strategy generation module reads the service fragment data records in the preprocessed dataset in the order of the time window. Each fragment data record is organized in a uniform key-value pair format, and is parsed one by one according to the preset key name field to extract three key fields: frame type identifier, time sequence number, and service semantic attribute of the fragment.

[0032] Frame type identifiers are used to distinguish the role of fragments in the coding layer, such as intra-coded frames, forward predictive coded frames, bidirectional predictive coded frames, key parameter set frames, or application layer control frames; timing sequence numbers indicate the decoding or presentation order of fragments in the original service data stream; and service semantic attributes are high-level descriptive tags for the content carried by fragments, such as control signaling frames, measurement data frames, media base layer frames, and media enhancement layer frames.

[0033] The above three attributes provide a basis for identifying the correlation between transmission failures between fragments.

[0034] S2.12. Based on the preprocessed transmission dependency data between service fragments, identify a set of service fragments with transmission failure associations; wherein: the transmission failure association refers to the fact that any transmission failure of any critical service fragment within the set will cause the entire set of service data to become unavailable; the critical service fragment refers to the service fragment that carries the core service semantics of the service fragment set and determines the parseability and service availability of the entire set of service data, and its corresponding attributes can be identified and determined through the frame type identifier and service semantic attributes of the service fragment; In this step, the preprocessed transport dependency data is stored in the preprocessed dataset as triples, with the format (src_id, dst_id, dep_type). Here, src_id is the source shard identifier, dst_id is the target shard identifier, and dep_type is the dependency type field, used to distinguish different dependency types such as decoding dependencies, synchronization dependencies, or semantic association dependencies. The scheduling policy generation module reads all dependency triples within the current time window and constructs a directed dependency graph, using each shard identifier as a node and dependencies as directed edges. In the directed dependency graph, if shard node A can be reached from shard node B through one or more directed edges, then a transport dependency exists between shard A and shard B.

[0035] Based on the completed dependency graph, the fragment set is identified and determined according to the definition of transmission failure association. Transmission failure association means that if any critical service fragment in a fragment set experiences an anomaly during transmission—including loss, failure to arrive beyond the allowed delay, or inability to decode after arrival—it will cause all other fragments in the set to lose their service parsing capability at the receiving end, i.e., the entire set of service data becomes unusable.

[0036] Among them, critical service fragments refer to service fragments that carry the core service semantics of the fragment set and determine the parseability and service availability of the entire set of service data. Their corresponding attributes can be identified and determined through the frame type identifier and service semantic attributes of the service fragment. In actual determination, the edge computing node has two pre-set reference benchmarks: a critical frame type list and a critical semantic tag list. If the frame type identifier of a fragment falls into the critical frame type list, such as an intra-coded frame or a critical parameter set frame, or if its service semantic attributes fall into the critical semantic tag list, such as control signaling or media base layer data, then the fragment is determined to be a critical service fragment; fragments that do not fall into the above categories are non-critical service fragments, and their transmission anomalies usually only affect themselves and do not spread to the entire set.

[0037] The specific process for identifying the fragment set associated with transmission failures is as follows: In the directed dependency graph, each critical service fragment is used as the starting point for traversal. A reachability traversal is performed along the dependency edges, assigning the starting critical fragment and all its reachable fragment nodes to the same candidate set. If two candidate sets of different critical fragments intersect, the two candidate sets are merged. This merging process continues until all candidate sets are mutually exclusive. Each connected set obtained at the end represents a service fragment set associated with transmission failures.

[0038] S2.13. Based on the timing sequence number and semantic attributes of the service fragments, encapsulate the set of service fragments with transmission failure associations into independent transmission-dependent fragment groups.

[0039] In this step, for each service segment set divided in S2.12, the time sequence number of each segment within the set is extracted and arranged in ascending order to form the segment time sequence of that set. Simultaneously, the distribution of service semantic attributes of each segment within the set is statistically analyzed. Based on the service semantic attributes of the key service segments within the group, and combined with the majority principle, a service type label is assigned to the set; that is, the semantic attributes of key segments are preferentially used as the group label, and if multiple types of key segments exist, the semantic attribute with the highest percentage of segments is selected as the group label. The service type label is used to identify the service attribute category of the data in this group, and typical values ​​may include video frame groups, control signaling groups, measurement data groups, audio frame groups, etc.

[0040] After sorting and labeling, the fragment set, along with its metadata—including a list of fragment member identifiers, the total number of fragments, a list of key business fragment identifiers, business type tags, the first fragment's timing sequence number, and the last fragment's timing sequence number—is encapsulated into an independent data structure called a transmission-dependent fragment group. Each transmission-dependent fragment group is independent of each other in terms of transmission failure; that is, a fragment transmission failure in one group will not cause the business data in another group to become unavailable. Subsequent channel matching calculations and transmission control are all carried out using the fragment group as the smallest scheduling unit.

[0041] S2.2 For each transmission-dependent fragment group, extract and configure the corresponding group-level transmission constraint parameters and group-level transmission failure judgment rules, and use them as prerequisite constraints for matching calculations; specifically, this includes the following steps: S2.21. Based on the service type of the transmission-dependent fragmentation group, determine the three group-level transmission constraint parameters: the maximum allowable end-to-end delay of the entire group, the maximum allowable transmission timing difference of fragments within the group, and the maximum allowable number of packet losses in the entire group. In this step, the edge computing node has a built-in mapping between service types and group-level transmission constraint parameters. For each service type label, the following three constraint parameters are defined with the following values: The first term is the maximum permissible end-to-end delay for the entire group, denoted as... The unit is milliseconds (ms). It means the maximum end-to-end time elapsed from the first fragment of the group being sent by the sender to the last fragment of the group being successfully received by the receiver. This parameter reflects the timeliness requirements of the service for the group of data.

[0042] The second term is the maximum allowable transmission timing difference for fragments within the group, denoted as... The unit is milliseconds (ms), which represents the upper limit of the time difference between any two fragments arriving at the receiver within the group. This parameter constrains the transmission dispersion of each fragment within the group, ensuring that the receiver can receive all fragments within a controllable time window. In engineering applications, it is usually taken as... 20% to 50%.

[0043] The third item is the maximum allowed number of packets lost in the entire group, denoted as... The unit is "pieces", which means the maximum number of non-critical service fragments that can be lost during the transmission of this group. Since the loss of critical service fragments will directly trigger the failure of the entire group's transmission, this parameter only constrains the tolerance for the loss of non-critical fragments.

[0044] In one optional configuration, the group-level constraint parameter values ​​for each business type are shown in the following examples: Control signaling group It takes 20ms. It takes 5ms. There are 0; Real-time video frame groups It takes 50ms. It takes 15ms. One; Real-time audio frame groups It takes 30ms. It takes 10ms. One; Measurement data set It takes 500ms. It takes 200ms. There are 3; File Transfer Group It takes 2000ms. It takes 1000ms. There are 5.

[0045] Using the service type label marked for each transport-dependent fragment group in S2.13 as the query key, the corresponding data is retrieved from the above correspondence. , and The specific values ​​of the three parameters are then assigned and written to the attribute set of the corresponding shard group. For ease of subsequent reference, for the first... Each segment group The corresponding three group-level transmission constraint parameters are denoted as follows: and .

[0046] S2.22. Set group-level transmission failure judgment rules: When any one of the following conditions is met, the transmission failure of the entire transmission dependent fragment group is judged to be a transmission failure. In this step, the group-level transmission failure determination rule adopts a three-condition logical "OR" structure, that is, if any one of the following three conditions is met, the overall transmission of the transmission-dependent fragment group is determined to be a failure.

[0047] Condition 1: Any critical business shard within the group is lost. The criterion for loss is that the shard failed to... It is correctly received by the receiving end and decoded and verified within a certain time.

[0048] Condition 2: The timing difference between fragments transmitted within the group exceeds the maximum allowable timing difference between fragments within the group. Let the arrival time of the first fragment to arrive at the receiver within this group be denoted as . The arrival time of the final fragment at the receiving end is When satisfied This condition is triggered at that time.

[0049] Condition 3: The total number of lost packets in the entire group exceeds the maximum allowable number of lost packets for the entire group. Let the number of non-critical fragments lost during transmission in this group be denoted as . When satisfied This condition is triggered at that time.

[0050] The above three conditions cover three typical transmission anomalies: critical fragment loss, overall timing limit exceeding, and overall packet loss limit exceeding. They can make an overall judgment on the transmission results of fragment groups from the perspective of service integrity.

[0051] S2.23. Associate the group-level transmission constraint parameters and group-level transmission failure judgment rules as pre-constraint conditions with the attribute set of the corresponding transmission-dependent fragment group.

[0052] In this step, the three constraint parameters determined in S2.21 and the three failure judgment conditions set in S2.22 are organized into a structured pre-constraint data block and mounted into the attribute set of the corresponding transport dependency fragment group.

[0053] This pre-constraint data block is referenced in multiple places in subsequent processing flows: The initial matching calculation of S2.3 uses it as a hard constraint boundary to filter channels that do not meet the basic transmission conditions; The dynamic adjustment in S2.4 uses the failure determination rules therein as the basis for determining the correction trigger; The transmission execution of S3 uses the constraint parameters therein for the receiving end's collection wait and failure detection.

[0054] Therefore, group-level transmission constraints and failure determination rules maintain a globally consistent definition and reference throughout the entire scheduling closed loop.

[0055] S2.3. Using group-level transmission constraint parameters as prerequisites, and based on the preprocessed operational status data of the multi-channel heterogeneous communication channels, calculate the initial matching relationship between the transmission-dependent fragment group and the multi-channel heterogeneous communication channels; specifically, this includes the following steps: S2.31. Extract the available bandwidth attribute, average transmission delay attribute, steady-state packet loss rate attribute, and channel fluctuation amplitude attribute of the multi-channel heterogeneous communication channels from the preprocessed operation status data of the multi-channel heterogeneous communication channels. In this step, we assume that the total number of heterogeneous communication channels available in the current communication system is [number missing]. Items, numbered sequentially as follows For the first one Channel Read the four operational status attribute values ​​of this channel within the current time window from the preprocessed dataset output by S1, after dimension normalization: available bandwidth attribute Average transmission delay attribute Steady-state packet loss rate attribute and channel fluctuation amplitude attributes The range of the above normalized values ​​is... ,in This is a preset small positive base value used to prevent division by zero or logarithmic anomalies in subsequent calculations.

[0056] Simultaneously, the preprocessed dataset retains the original dimensional values ​​of the operating status attributes of each channel, which are read together in this step for use in hard constraint verification, including the original value of available bandwidth. (Unit: Mbps), raw average transmission delay (Unit: ms) and raw value of steady-state packet loss rate The normalized attribute values ​​and the original dimensional values ​​together constitute the channel. The current state description is used for subsequent matching degree calculation and feasibility screening.

[0057] S2.32. Extract the total data size, importance level, latency sensitivity level, and group-level transmission constraint parameters of the transmission-dependent fragment group from the attribute set of the transmission-dependent fragment group. In this step, let's assume there are a total of transport-dependent fragment groups to be scheduled within the current batch. 1, numbered sequentially as For the first one Each segment group Extract the following four types of information from its attribute set: The first category is the total scale of the data. The unit is bytes, which is the sum of the service fragment payload lengths within the group, reflecting the group's overall demand for channel bandwidth resources.

[0058] The second category is importance level. The value is an integer from 1 to 5. The larger the value, the more important the business data carried by the group is and the higher the requirement for transmission reliability. This level is pre-configured by the upper-layer business system or obtained by mapping the service type label of the fragment group.

[0059] The third category is the time delay sensitivity level. The value is an integer from 1 to 5. The larger the value, the more sensitive the group is to transmission latency. This level is also pre-configured by the upper-layer business system or corresponds one-to-one with the business type label.

[0060] The fourth category consists of group-level transmission constraint parameters configured in S2.2, namely the maximum permissible end-to-end delay for the entire group. Maximum allowable transmission timing difference for fragments within a group and the maximum allowed number of packets lost in the entire group The above four types of information provide the input basis for the subsequent hard constraint screening and matching degree weighted calculation on the fragment group side.

[0061] S2.33. Using group-level transmission constraint parameters as hard constraint boundaries, calculate the initial matching degree values ​​of each transmission-dependent fragment group and multiple heterogeneous communication channels through multi-attribute weighted matching operations. All initial matching degree values ​​constitute the initial matching relationship between the transmission-dependent fragment group and the multiple heterogeneous communication channels.

[0062] In this step, before calculating the matching degree, the feasibility of each fragment group and channel combination is screened using group-level transmission constraint parameters as hard constraint boundaries. For fragment groups... With channel The combination must simultaneously meet the following conditions: the original available bandwidth of the channel. Able to support Complete within the time limit The amount of data transmitted can be chosen when protocol overhead is ignored. As an approximate criterion, the original average transmission delay of the channel. No more than The preset ratio, which is determined according to the service type of the segment group, is 30% for the control signaling group and real-time audio / video group, and 50% for the measurement data group and file transfer group; the steady-state packet loss rate of the channel. The packet loss rate must not exceed the threshold value corresponding to the service type. The packet loss rate threshold is 1% for the control signaling group, 3% for the real-time audio and video group, 5% for the measurement data group, and 10% for the file transfer group. If any condition is not met, the initial matching degree of the combination is directly assigned to 0, indicating that the combination does not have basic transmission feasibility.

[0063] For combinations selected through hard constraints, perform multi-attribute weighted matching operations to calculate the fragment groups. With channel initial matching degree The calculation formula is: ; In the formula, These are the weighting coefficients for the four attributes: available bandwidth, transmission delay, packet loss rate, and channel fluctuation amplitude, respectively, satisfying the normalization constraint. The weighting coefficients are determined based on the partition groups. Importance level and latency sensitivity level Differentiated configurations are implemented, with the basic adjustment rule being: when the importance level... When the packet loss rate is high, appropriately increase the weight of the packet loss rate. Prioritize ensuring transmission reliability; when latency sensitivity level When the delay is high, appropriately increase the delay weight. Prioritizing real-time transmission; channel fluctuation amplitude weighting. In a typical configuration, a fixed value of 0.20 is maintained, but it can also be adjusted depending on the channel stability.

[0064] In one optional specific configuration: when and At that time, the weighted quadruple ( (Take (0.40, 0.20, 0.20, 0.20)). when and When the time is right, take (0.20, 0.50, 0.10, 0.20); when and When the time is right, take (0.20, 0.10, 0.50, 0.20). when and When the value is (0.10, 0.30, 0.40, 0.20), take (0.10, 0.30, 0.40, 0.20). when and When the time is right, take (0.25, 0.25, 0.30, 0.20).

[0065] For values ​​not covered above and For intermediate level combinations, linear interpolation of the weight values ​​corresponding to adjacent level combinations can be used. The specific values ​​of each weight can be calibrated and optimized based on the actual deployment scenario through offline testing or online learning during runtime.

[0066] All of the calculations constitute a Initial matching degree matrix of dimension .matrix elements in That is, a shard group With channel The initial matching degree value; the higher the value, the better the overall fit of the channel to the fragment group in the current state. Matrix This establishes the initial matching relationship between the transmission-dependent fragment group and the multiple heterogeneous communication channels, providing a basis for the subsequent channel selection in S2.5.

[0067] S2.4. Based on the group-level transmission failure judgment rule, dynamically adjust the matching relationship between the same type of transmission dependent fragment groups to be transmitted within the same batch and the corresponding multi-path heterogeneous communication channels according to the overall transmission results of the transmission dependent fragment groups that have completed transmission; specifically, it includes the following steps: S2.41. Obtain the overall transmission result of the transmission-dependent fragment group that has been transmitted. Extract three types of feedback data: the overall transmission failure status, the timing difference deviation value within the group, and the overall packet loss deviation value. In this step, during system operation, after each transmission-dependent fragment group completes multi-channel splitting transmission via step S3, S3 will collect the overall transmission results of the group and synchronously feed them back to S2 through the internal data backhaul channel. For fragment groups that have completed transmission... The following feedback data is extracted from the overall transmission results: The first category is a complete transmission failure. . It is a Boolean logic value. This means that at least one of the three failure criteria defined in S2.22 has been triggered, i.e., the group is determined to be a complete transmission failure; This indicates that none of the three conditions were triggered, and the entire transmission was successful.

[0068] The second category is the intra-group time series difference deviation value, denoted as... The unit is milliseconds (ms). It is defined as the deviation between the maximum arrival time difference within the group and the allowed upper limit during actual transmission: ; when When, it indicates that the actual timing difference exceeds the allowable upper limit; when When, it indicates that the timing difference is within the allowable range, with a margin of . .

[0069] The third category is the overall packet loss deviation value, denoted as... The unit is packets. It is defined as the deviation between the actual number of lost packets in the group and the allowed upper limit: ; when When, it indicates that the actual number of lost packets exceeds the allowed limit; when When the number of packets lost is within the allowed range, it indicates that the number of packets lost is within the allowed range.

[0070] In addition, the actual number of critical fragments lost in this group was extracted from the feedback data. This serves as a quantitative feedback quantity corresponding to failure condition one (loss of critical business fragments).

[0071] S2.42. Taking the overall transmission failure state as the core correction trigger condition, and combining the timing difference deviation value within the group and the overall packet loss deviation value, calculate the matching degree correction amount of the corresponding multi-path heterogeneous communication channels. In this step, we set up a fragment group. The channel actually used for transmission is .by To modify the core trigger conditions, combined with , and The value, calculation channel Matching degree correction for this type of shard group This correction amount applies only to transport-dependent fragment groups of the same type, and the correction amounts for different service types are independent of each other.

[0072] when That is, the matching degree correction amount when the entire group is successfully transmitted. Take a positive value or zero. The positive correction magnitude is based on... and Adjust the margin size. The calculation formula is: ; In the formula, It is a positive correction factor; and The contribution weights for timing difference margin and packet loss margin are respectively, satisfying... .when If the proportion of timing difference margin is less than the preset security threshold or the proportion of packet loss capacity margin is less than the preset security threshold, it indicates that although the entire group meets the transmission constraints, the security margin is insufficient. In this boundary case... Set to 0, and no positive correction is made.

[0073] when That is, the matching degree correction amount when the entire transmission fails. Take a negative value. The magnitude of the negative correction is determined comprehensively based on the specific conditions that trigger failure, as follows: ; In the formula, It is a negative correction factor; This represents the total number of critical fragments within the group. , , The contribution weights are respectively for three failure factors: critical fragment loss, timing difference exceeding the limit, and packet loss exceeding the limit, satisfying... When a failure condition is not triggered, the corresponding value is 0. After normalization, each deviation is limited to the range [0,1]. When the normalized deviation value of a certain item exceeds 1, it is taken as 1.

[0074] In one optional parameter configuration, the positive correction coefficient Take 0.02 as the negative correction factor. Take 0.15, and All values ​​are 0.5. Take 0.5, Take 0.3, The coefficient is set to 0.2. The values ​​of these coefficients reflect that the penalty for transmission failure is greater than the reward for successful transmission, and that critical fragment loss has the highest negative correction weight. Each coefficient can be adjusted and optimized based on the actual operating data of the communication system.

[0075] S2.43 Under the dual constraints of the preset upper limit of the correction step size and the maximum attenuation amplitude, the matching degree correction amount is used to update the matching degree value of the same type of transmission dependent fragment group to be transmitted in the same batch and the corresponding multi-path heterogeneous communication channel, so as to complete the dynamic adjustment of the matching relationship between the transmission dependent fragment group and the multi-path heterogeneous communication channel.

[0076] In this step, to avoid drastic changes in the matching degree caused by a single transmission result, which could lead to frequent switching of the scheduling strategy within a short period, two constraints are imposed on the matching degree correction. The first is the upper limit of the correction step size. The first term limits the absolute value of the maximum change in the matching degree value during a single correction operation; a typical value is 0.1. The second term represents the maximum attenuation magnitude. Limit the matching degree of a channel from its initial matching degree. The maximum cumulative decrease is used to prevent channels from being overly penalized due to short-term, occasional anomalies and losing scheduling opportunities for a long period of time. A typical value can be 0.4.

[0077] Set up a channel The current matching degree for this type of shard group is The correction amount calculated by S2.42 is The matching score update is performed according to the following process: First, limit the step size of the correction amount: if Then let .

[0078] Secondly, calculate the candidate new matching degree. .

[0079] Secondly, apply a maximum decay constraint to the candidate new matching degree: if ,but ;otherwise .

[0080] Finally, the matching score is cropped to the [0,1] range to ensure that the matching score is always within the normalized range.

[0081] The scope of the above update operation is strictly limited to fragment groups within the same batch that have not yet entered the transmission execution phase and have already completed transmission. Transport-dependent fragment groups with the same service type label. Different types of fragment groups do not share matching degree corrections to ensure the independence of channel evaluation for different service types and prevent the transmission performance of one service type from interfering with the channel selection decision of another service type. After the update is complete, the initial matching degree matrix... The matching degree values ​​of the corresponding fragment group and channel position are updated to the corrected new values, and the overall matching relationship is dynamically adjusted. The adjusted matching relationship matrix will replace the initial matching relationship and be used for channel matching decisions of the remaining fragment groups to be transmitted in the current batch in S2.5.

[0082] S2.5. Based on the adjusted matching relationship between transmission-dependent fragment groups and multiple heterogeneous communication channels, a fragment group-level multi-channel fusion communication scheduling strategy is generated in a structured manner according to the transmission timing sequence. Specifically, this includes the following steps: S2.51. Based on the adjusted matching relationship between the transmission dependency fragment group and the multiple heterogeneous communication channels, match the main transmission channel and the alternative transmission channel for each transmission dependency fragment group. In this step, for each transport-dependent fragment group currently to be scheduled... Read the group and all pairs from the adjusted matching matrix output by S2.4. The matching degree values ​​of each channel are used. Channels are sorted in descending order of their matching degree values. The channel with the highest matching degree value, which is greater than the preset minimum matching threshold, is selected as the primary transmission channel for this group. The preset minimum matching threshold can be set according to the system's conservative approach to scheduling quality, with a typical value of 0.3. If there is a channel with the second highest matching degree value that is also greater than the minimum matching threshold, it is used as a backup transmission channel for this group. If only one channel meets the threshold condition, the backup channel field is recorded as null. If no channel meets the threshold condition, the shard group is marked as having no available channels, and a channel resource insufficiency alarm event is output to the system log, indicating that manual intervention is required or waiting for the channel status to recover. This shard group is not added to the transmission queue but is placed in the waiting-to-be-scheduled buffer pool, where channel matching will be re-evaluated in the next scheduling cycle.

[0083] S2.52. Configure the overall transmission priority, offloading ratio, and fragmentation timing coordination rules within each transmission-dependent fragmentation group; In this step, the overall transmission priority is determined based on the importance level of the fragment group. and latency sensitivity level Determined comprehensively. Priority value. Calculate using the following formula: ; in and These are the weighting factors for importance and latency sensitivity, respectively, and their typical values ​​can both be set to 0.5. The smaller the value, the higher the priority. Fragments within the same batch are grouped according to... Ascending order determines the transmission sequence.

[0084] The offloading ratio defines the data volume allocation between the primary transmission channel and the backup transmission channel for fragmented data within a group. In the default configuration, all fragments are carried by the primary transmission channel, with an offloading ratio of 100% for the primary channel and 0% for the backup channel. When the available bandwidth of the primary channel is deemed insufficient for... Complete the entire group independently within the specified time. During data transmission, the splitting ratio is dynamically calculated based on the ratio between the original available bandwidth values ​​of the primary and backup channels. For example, suppose the available bandwidth of the primary channel is... The available bandwidth for the alternative channels is The data offloading ratio can then be set according to bandwidth proportion: the main channel handles the majority of the data volume. The alternative channel handles the majority of the data volume. This allows the two sets of data streams to be transmitted in parallel and then... The process is completed simultaneously within the same timeframe, thus satisfying the overall latency constraints through dual-channel collaborative transmission.

[0085] The intra-group fragment timing coordination rule is used to constrain the transmission timing of each fragment within the group and the receiver's collection waiting strategy. Specifically, the rule stipulates that: each service fragment within the group is sent sequentially from the sender in ascending order of its timing number; after receiving the first fragment of the group, the receiver's startup time is... The entire group is collected and a waiting timer is set up. If the receiver successfully collects all fragments in the group before the timer expires, the complete group data is submitted to the upper-layer service application. If the timer expires before all fragments are collected, the group-level transmission failure judgment rule agreed in S2.22 is used to determine whether to trigger the entire group transmission failure and the corresponding processing procedure.

[0086] S2.53. According to the transmission timing order of the transmission-dependent fragment groups, the channel matching results, overall transmission priority, splitting ratio and fragment timing coordination rules within the group of all transmission-dependent fragment groups are structurally encapsulated to generate a fragment group-level multi-channel fusion communication scheduling strategy.

[0087] In this step, the transmission order of each fragment group within this batch is determined by using the timing sequence number of the first fragment of each transmission-dependent fragment group as the sorting key, proceeding in ascending order. For each fragment group, its group identifier, service type label, primary transmission channel identifier, alternative transmission channel identifier (optional), and overall group transmission priority value are determined. The following elements are assembled into a structured scheduling entry: the traffic splitting ratio configuration (the percentage or absolute value of data volume handled by the primary and backup channels respectively), a complete description of the intra-group fragmentation timing coordination rules, and the group-level transmission constraint parameters and failure judgment rules configured in S2.2. All scheduling entries are arranged sequentially according to transmission timing and encapsulated into a complete fragmentation group-level multi-channel converged communication scheduling strategy. This scheduling strategy is distributed to S3 in the form of internal message packets through inter-process communication or shared memory mechanisms within the edge computing nodes, serving as the direct basis for subsequent multi-channel group collaborative transmission execution.

[0088] It is worth noting that, through the S2 fragment group-level scheduling strategy, the transmission dependencies within the service data are explicitly incorporated into the scheduling model. The scheduling granularity, based on fragment groups as the basic unit, ensures that channel matching decisions reflect the overall effectiveness constraints of the service data. Simultaneously, the introduction of group-level constraints and group-level feedback correction mechanisms enables the scheduling strategy to continuously and adaptively optimize based on service batch-level transmission results. These two mechanisms work together to reduce the probability of service parsing failures due to critical fragment mismatches or intra-group timing disorders in complex service scenarios and heterogeneous network conditions, thereby improving the integrity and reliability of multi-channel fused transmission.

[0089] S3. Multi-channel group collaborative transmission execution: Based on the fragment group-level multi-channel converged communication scheduling strategy, collaborative management and control of multiple heterogeneous communication channels are implemented. Multi-channel offloading and transmission of service fragment data is executed on a transmission-dependent fragment group basis, and the overall transmission results of the completed transmission-dependent fragment groups are collected. Specifically, this step receives the fragment group-level multi-channel converged communication scheduling strategy output from S2 as input. Based on the channel matching results, priority order, offloading ratio, and timing coordination rules specified in the strategy, unified management and control of multiple heterogeneous communication channels are implemented. Multi-channel offloading and transmission of service fragment data is organized on a transmission-dependent fragment group basis, and the overall transmission results are collected after each fragment group transmission is completed. These results are then sent back to S2 and S4 respectively, providing feedback data for dynamic adjustment of matching relationships and verification of scheduling effects. S3 includes the following sub-steps: S3.1 Analyze the multi-channel converged communication scheduling strategy at the fragment group level, extract the main transmission channel, alternative transmission channel, overall group transmission priority, diversion ratio and fragment timing coordination rules within the group corresponding to each transmission-dependent fragment group, and complete the transmission pre-configuration of multiple heterogeneous communication channels; In this step, the shard group-level multi-channel converged communication scheduling policy issued by S2 is read from the inter-process communication interface or shared memory area within the edge computing node. This policy is organized in the form of structured message packets, and is parsed field by field according to a preset message format to extract the scheduling entries for each transmission-dependent shard group encapsulated in the policy.

[0090] Each scheduling entry contains the following configuration information: a group identifier, used to uniquely identify the fragment group during transmission; a service type label, used to distinguish the service attribute category of the group; a primary transmission channel identifier and a backup transmission channel identifier, where the backup channel identifier can be null; and the overall group transmission priority value. This is used to determine the transmission order of the group within the same batch; the splitting ratio configuration specifies the data volume allocation relationship between the main transmission channel and the alternative transmission channel for the fragmented data in the group; the intra-group fragment timing coordination rules define the transmission timing of each fragment in the group and the receiving end's collection waiting strategy; and the group-level transmission constraint parameters and failure judgment rules configured in S2.2 are used for the receiving end to perform collection judgment and failure judgment during the transmission process.

[0091] After strategy parsing is completed, the corresponding multi-heterogeneous communication channels are pre-configured for transmission based on the primary and alternative transmission channel identifiers matched to each fragment group. Transmission pre-configuration includes: issuing channel enable commands to the channel interface drivers corresponding to the primary and alternative transmission channels (if any) to establish or activate the data transmission link from the edge computing node to the receiving end; allocating a transmission buffer and corresponding queue resources for each participating channel to temporarily store fragmented data to be sent; and initializing the transmission parameters of each channel according to the group-level transmission constraints in the strategy, such as maximum transmission rate and retransmission strategy. After the above pre-configuration is completed, the multi-heterogeneous communication channels are in a ready state, waiting for the distribution and transmission of fragmented data to begin.

[0092] S3.2. According to the priority of the whole group transmission, the service fragment data is distributed to the corresponding multi-way heterogeneous communication channels for offloading transmission, based on the transmission-dependent fragment group and the fragment timing coordination rules within the group. In this step, the transmission priority value of each fragment group is determined. The size of the data is used to sort all transport-dependent fragment groups within the same batch. A smaller value indicates a higher priority, meaning it enters the transmission queue first. For fragment groups with the same priority value, the transmission order is determined by the sequence number of their first fragment. The transmission process is then initiated for each transmission-dependent fragment group in the predetermined order.

[0093] For the current transmission-dependent fragment group to be transmitted First, all service fragment data contained in the group are read from the send buffer, and each fragment is indexed and located by its fragment identifier. Based on the traffic splitting ratio configuration of the group, the data volume allocation scheme for each fragment to be sent via the primary transmission channel and the backup transmission channel is determined.

[0094] Under the default traffic splitting configuration, where the primary channel handles 100% of the data and the backup channel handles 0%, all fragmented data within the group is sent through the primary transmission channel. Based on the transmission order specified in the group's fragmented timing coordination rules, each fragment within the group is extracted sequentially according to its timing number from smallest to largest, encapsulated into a data packet adapted to the primary transmission channel's communication protocol, and pushed frame by frame to the primary transmission channel's transmission buffer. The physical layer transmission is then driven by the channel interface.

[0095] When the available bandwidth of the main channel is assessed to be insufficient... Complete the entire group independently within the time limit When transmitting large amounts of data, the splitting ratio is configured to a non-default value, with the primary and backup channels each handling a portion of the data. In this case, the fragments within the group are divided into two fragment subsets according to the splitting ratio, corresponding to the primary channel and the backup channel, respectively. During partitioning, the temporal continuity of the fragments within the group is maintained; consecutive fragments with smaller time sequence numbers are assigned to the primary channel, and consecutive fragments with larger time sequence numbers are assigned to the backup channel. When the number of fragments is small and the timing consistency requirement is extremely high, they can also be allocated alternately in a round-robin manner according to the splitting ratio. Fragments within each of the two fragment subsets are sent sequentially according to their time sequence numbers from smallest to largest. The corresponding fragment data packets are synchronously pushed to the send buffers of the primary and backup channels, and the two sets of data streams are transmitted in parallel on the two channels to shorten the end-to-end transmission time of the entire group.

[0096] During transmission, in accordance with the group's fragmentation timing coordination rules, a transmission timestamp and a group sequence number are appended to each fragment's data packet for the receiver to calculate timing differences and determine completeness. Simultaneously, the transmission status of each fragment is tracked and recorded, including transmission time, transmission channel identifier, and packet sequence number, forming a transmission status record for the group.

[0097] S3.3 After all transmission dependent fragments in a single transmission group have been transmitted, collect the overall transmission results of the completed transmission dependent fragment group and synchronously send the overall transmission results back to S2 and S4 for matching relationship adjustment and scheduling effect verification.

[0098] In this step, for each transmission-dependent fragment group, all fragments within the group have been sent—that is, the sending end confirms that all fragmented data packets have been pushed to the corresponding channel's transmission buffer and submitted to the physical layer by the channel interface driver to complete the transmission action—and then starts the overall transmission result acquisition process for that group.

[0099] The acquisition process begins by obtaining the reception status report for the group from the receiving end. During transmission, the receiving end continuously records the actual arrival time, fragment identifier, arrival channel identifier, and decoding verification result of each fragment within the group. When the group's overall reception waiting timer times out or all fragments are received ahead of schedule, the receiving end determines the overall transmission result of the group according to the group-level transmission failure judgment rules agreed in S2.22 and generates a reception status report.

[0100] The receive status report contains the following information: group identifier; overall group transmission failure status. The value can be 1 (failure) or 0 (success); a list of arrival times for each fragment in the group, including the arrival time of the first arriving fragment. and the arrival time of the last fragment The actual number of fragments arriving and the number of fragments lost within the group, with the number of lost critical fragments denoted as [missing information]. The number of lost non-critical fragments is recorded as ; The specific condition number that triggers the failure determination (if ).

[0101] The received status report is compared with the transmitted status records on the sending end to verify whether there are any lost fragments that the sending end has sent but the receiving end has not received. False loss cases where the receiving end did not receive the data due to the sending end not sending the data are eliminated, ensuring the accuracy of the statistical data. After comparison and verification, the overall transmission result data for this transmission-dependent fragment group is generated. The overall transmission result data includes at least: the entire group's transmission failure status. Intra-group timing difference deviation value (Unit: ms) Overall packet loss deviation value (Unit: Number of pieces), and the actual number of critical fragments lost. .

[0102] After the overall transmission result is generated, it is synchronously sent to S2 and S4 via the data backhaul channel within the edge computing node. The overall transmission result sent to S2 is used to trigger the dynamic adjustment process of the matching relationship in S2.4; the overall transmission result sent to S4 is used to drive the scheduling effect verification and indicator statistics in S4.1. The above backhaul process adopts an asynchronous message mechanism, which does not block the transmission execution of subsequent shard groups within the current batch.

[0103] It is worth noting that through the multi-channel group collaborative transmission execution of S3, the scheduling strategy generated by S2 is transformed into actual transmission control operations for each channel. The transmission organization method based on transmission-dependent fragment groups, combined with fragment timing coordination rules within groups and traffic splitting ratio control, enables multiple heterogeneous communication channels to work collaboratively under a unified scheduling framework, completing parallel traffic splitting transmission of service data while meeting the transmission constraints of each group. Simultaneously, the complete acquisition and bidirectional feedback of the overall transmission results for each fragment group provides accurate and timely feedback data for online matching and correction in S2 and offline parameter updates in S4, ensuring the real-time performance and data consistency of the entire scheduling loop.

[0104] S4. Scheduling Effect Verification and Parameter Update: Based on the overall transmission results of the transmission-dependent fragment groups that have completed transmission, verify the execution effect of the fragment group-level multi-channel converged communication scheduling strategy, and update the baseline data of the operating status of the multi-channel heterogeneous communication channels and the basic parameters for calculating the group-level matching degree between the transmission-dependent fragment groups and the multi-channel heterogeneous communication channels. Specifically, this step receives the overall transmission results of the transmission-dependent fragment groups that have completed transmission as input from S3. By performing statistical analysis and effect verification on the accumulated overall transmission results within a certain period, evaluate the execution effect of the fragment group-level multi-channel converged communication scheduling strategy, and update the baseline data of the operating status of the multi-channel heterogeneous communication channels and the basic parameters for calculating the group-level matching degree between the transmission-dependent fragment groups and the multi-channel heterogeneous communication channels accordingly. S4 and S2.4 together constitute the two-level feedback correction mechanism of this method, where S2.4 is an online fast correction based on a single group of transmission results, and S4 is a periodic parameter update based on multiple groups of statistical results. The two complement each other in terms of time scale and scope. S4 includes the following sub-steps: S4.1 Based on the overall transmission results of the transmission-dependent fragment groups that have completed transmission, statistically analyze three types of performance indicators: the overall group transmission failure rate, the intra-group transmission timing difference compliance rate, and the overall group packet loss compliance rate. Verify the execution effect of the fragment group-level multi-channel fusion communication scheduling strategy and generate performance verification results. In this step, after each transmission-dependent fragment group completes its transmission, S3 asynchronously sends the overall transmission result data of that group to this step via the internal data backhaul channel. The overall transmission result data includes: the transmission failure status of the entire group. Intra-group timing difference deviation value (Unit: ms) Overall packet loss deviation value (Unit: Number of) and the actual number of critical fragments lost This step uses a preset statistical period as a unit to summarize and calculate the overall transmission results of all transmitted fragment groups accumulated within the statistical period. The statistical period can be set according to the time dimension, such as every 1 second; or it can be set according to the number of fragment groups, such as triggering a statistical analysis every 100 fragment groups completed. The specific setting depends on the timeliness requirements of the business scenario for evaluating the scheduling effect.

[0105] Within the statistical period, let the total number of completed tasks be... Each transmission depends on the transmission of fragment groups. This step iterates through the fragment groups. The overall transmission results of each fragment group are statistically analyzed based on the following three categories of performance metrics: The first type of indicator is the overall transmission failure rate, denoted as... Defined as the ratio of the number of fragments judged as having overall transmission failure within a statistical period to the total number of fragments: ; The value range is [0,1]. The lower the value, the stronger the scheduling strategy's ability to guarantee the overall transmission of the fragment group.

[0106] The second type of indicator is the intra-group transmission timing difference compliance rate, denoted as... Defined as the intra-group time series difference deviation value within the statistical period not exceeding the allowable upper limit (i.e. The ratio of the number of segments to the total number of segments: ; in This is an indicator function that takes the value 1 when the condition inside the parentheses is true, and 0 otherwise. The value range is [0,1]. A higher value indicates that the scheduling strategy performs better in terms of the consistency of the arrival time of the fragments within the control group.

[0107] The third type of indicator is the overall packet loss compliance rate, denoted as... Defined as the total packet loss deviation value within the statistical period not exceeding the allowable upper limit (i.e., The ratio of the number of segments to the total number of segments: ; The value range is [0,1]. A higher value indicates that the scheduling strategy performs better in controlling packet loss.

[0108] After calculating the three types of indicators, the values ​​of the indicators are compared with the preset effect verification thresholds to generate effect verification results.

[0109] Furthermore, the results of the effectiveness verification are divided into three levels: when and and If the current scheduling strategy is deemed to be performing well, the parameter update strategy is to maintain the existing parameter configuration.

[0110] When any indicator is worse than the above-mentioned good threshold, but the degree of degradation does not exceed 10 percentage points (i.e. and and If the condition is not met (and the excellent condition is not met), the execution effect is judged to be average, and the parameter update strategy is to trigger the minor fine-tuning update in S4.2 and S4.3.

[0111] When any indicator is worse than the above general threshold (i.e.) or or When the execution effect is deemed poor, the parameter update strategy is to trigger the large step size adjustment update in S4.2 and S4.3.

[0112] The effect verification results are generated in a structured format, including the start and end information of the current statistical period (timestamp or segment group number range), the specific values ​​of the three types of indicators, the effect level judgment results, and the corresponding parameter update strategy. This result will directly serve as the trigger signal and amplitude selection basis for parameter updates in S4.2 and S4.3.

[0113] S4.2. Based on the effect verification results, update the operating status benchmark data of the multi-channel heterogeneous communication channels, and use the updated operating status benchmark data of the multi-channel heterogeneous communication channels as the benchmark for the next round of operating status data collection and evaluation of the multi-channel heterogeneous communication channels. In this step, the baseline data for the operational status of the multi-channel heterogeneous communication channels refers to the historical minimum values ​​of each indicator used in S1.3 when performing Min-Max normalization on the channel operational status data. and maximum value S1.3 During normalization, the latest baseline data maintained in this step is read from shared memory or the parameter interface. The accuracy of the baseline data directly affects how well the normalized attribute values ​​reflect the true state of the channel; therefore, it needs to be periodically updated based on long-term changes in the channel's operating characteristics.

[0114] For the Channel Let a certain operational status indicator be defined as having a current normalized baseline minimum value. The maximum value is During this statistical period, the actual collected value sequence for this indicator is as follows: ,in This represents the number of times this indicator is collected within the statistical period. When the statistical period is set according to the time dimension, This represents the total number of times the edge computing nodes collected data on this metric within the specified time period; when the statistical period is set according to the number of shard groups, This represents the number of data collections within the time period covered by this statistical period. Calculate the minimum value of this sequence. and maximum value .

[0115] Based on the effect level in the effect verification results generated by S4.1, different smoothing strategies are used to update the baseline data.

[0116] When the performance level is excellent, it indicates that the current normalized baseline data matches the actual operating characteristics of the channel well, and an exponentially weighted moving average is used for smooth updates. ; ; in The smoothing factor is set to 0.8. The larger the value, the smoother the baseline data updates and the lower the sensitivity to short-term fluctuations.

[0117] When the effect level is moderate, it indicates that there may be some deviation between the current normalized baseline data and the actual operating characteristics of the channel, and the update speed of the baseline data needs to be accelerated to adapt to changes in the channel status. At this time, the smoothing factor... Take 0.5.

[0118] When the performance level is poor, it indicates that there may be a significant deviation between the current normalized baseline data and the actual operating characteristics of the channel. The minimum and maximum values ​​from this round of statistics are directly used as the new baseline data, which is equivalent to... : ; ; Updated runtime baseline data and The existing baseline data will be overwritten and written to shared memory or the parameter interface for use in the next round of S1.3's dimensional normalization processing. Through this hierarchical update mechanism, the normalized baseline can adaptively adjust its update rate based on the scheduling execution effect, achieving a balance between tracking changes in channel characteristics and maintaining baseline stability.

[0119] S4.3. Based on the effect verification results, update the basic parameters for calculating the group-level matching degree of the transmission-dependent fragment group and the multi-path heterogeneous communication channel, and use the updated basic parameters for calculating the group-level matching degree of the transmission-dependent fragment group and the multi-path heterogeneous communication channel as the basis for the benchmark weight of the next round of group-level matching degree calculation.

[0120] In this step, the basic parameters for calculating the group-level matching degree between transmission-dependent fragment groups and multiple heterogeneous communication channels refer to the weighting coefficients used in the multi-attribute weighted matching operation in S2.33. and at different importance levels and latency sensitivity level Configuration relationships under the combination. This step adjusts the above basic parameters according to the effect level and specific indicator deviation in the effect verification results generated by S4.1.

[0121] When the performance level is excellent, the current weight configuration can adapt well to the current channel environment and business characteristics, and the weights remain unchanged.

[0122] When the effect level is average or poor, the direction and amount of weight adjustment are determined based on the specific deviation of each performance indicator.

[0123] like Not meeting the excellent standard or If the failure to meet the excellent standard is primarily due to packet loss, or if the overall packet loss rate fails to meet the excellent standard, it indicates that the current scheduling strategy is inadequate in ensuring transmission reliability, and the weight of the packet loss rate attribute needs to be adjusted. The adjustment will be increased to enhance the scheduling strategy's sensitivity to channel packet loss characteristics. If... The failure to meet the excellent standard indicates that the current scheduling strategy is inadequate in terms of transmission timing consistency within the control group, and the weighting of delay attributes is insufficient. The weights are increased to enhance the scheduling strategy's sensitivity to channel latency characteristics. Simultaneously, the remaining weights within the same weight combination are scaled down proportionally to their original values ​​to maintain weight normalization constraints; that is, each remaining weight is multiplied by the same scaling factor so that the sum of all adjusted weights remains 1.

[0124] Weight adjustment The size is determined based on the effect level. When the effect level is moderate, a small adjustment step size is used. The weight is set to 0.05, meaning the target weight is increased by 0.05 each time. When the effect level is poor, a larger step size is used for adjustment. The step size is set to 0.10, meaning the target weight is increased by 0.10 each time. If a certain indicator triggers multiple weight increase requests simultaneously, the target weights are increased by the above step size, and the remaining weights in the group are decreased proportionally to maintain normalization.

[0125] Let the current value of a certain weight be... The adjusted candidate weights are as follows: (Upward adjustment) or the corresponding value after proportional downward adjustment. Apply amplitude constraints to the candidate new weights: the reasonable value range for each weight is [0.05, 0.60]. When When, take 0.05; when When the value is 0.60, after independently adjusting and constraining the magnitude of each weight, the affected weight combinations are renormalized to ensure that the sum of the weights in the same combination is 1.

[0126] The updated group-level matching degree calculation parameters will overwrite the original parameter configuration and be written to shared memory or parameter interface, serving as the basis for weight configuration in the next round of multi-attribute weighted matching operations in S2.3.3. Through the periodic parameter updates in S4.3, the group-level matching degree calculation parameters can be optimized in a targeted manner based on statistical feedback of actual scheduling effects during system operation, so that the weight configuration gradually approaches the optimal combination that adapts to the current channel environment and business characteristics.

[0127] It is worth noting that, after the overall processing of S4, the execution effect of the segmented group-level multi-channel converged communication scheduling strategy is quantitatively evaluated at the statistical level. The baseline data of channel operation status and the basic parameters for calculating group-level matching degree are updated in a targeted manner according to the evaluation results, so that the normalized baseline of S1 and the matching calculation weight of S2 can continuously adapt to the long-term changes in channel operation characteristics and business needs. The two-level adaptive mechanism of "online rapid correction combined with periodic parameter optimization" constituted by S4 and S2.4 ensures the long-term stability and continuous optimization capability of the entire scheduling method in complex heterogeneous network environments.

[0128] Those skilled in the art will understand that the process of implementing all or part of the steps of the above embodiments can be carried out by hardware or by a program instructing the relevant hardware.

[0129] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A method for scheduling intelligent multi-channel converged communication based on edge computing, characterized in that, Includes the following steps: S1. Multi-source data acquisition and preprocessing: The operation status data, service fragment data, and transmission dependency data between service fragments are acquired through edge computing nodes, and the operation status data, service fragment data, and transmission dependency data between service fragments are preprocessed. S2. Fragment Group-Level Scheduling Strategy Generation: Based on the preprocessed service fragmentation data and the preprocessed service fragmentation data, transmission dependency fragment groups are obtained. The matching relationship between the transmission dependency fragment groups and the multi-path heterogeneous communication channels is determined according to the preprocessed multi-path heterogeneous communication channel operation status data. The matching relationship is adjusted based on the overall transmission results of the completed transmission dependency fragment groups to generate a fragment group-level multi-channel converged communication scheduling strategy; where: The transmission-dependent fragmentation group refers to a multi-channel scheduling basic unit composed of several service fragments with transmission dependencies, which needs to ensure transmission effectiveness as a whole. S3. Multi-channel group collaborative transmission execution: Based on the multi-channel converged communication scheduling strategy at the fragment group level, the multi-channel heterogeneous communication channels are coordinated and managed. The multi-channel diversion and transmission of business fragment data is executed in units of transmission-dependent fragment groups, and the overall transmission results of the transmission-dependent fragment groups that have completed transmission are collected. S4. Scheduling effect verification and parameter update: Based on the overall transmission results of the transmission-dependent fragment groups that have completed transmission, verify the execution effect of the fragment group-level multi-channel converged communication scheduling strategy, and update the baseline data of the operating status of the multi-path heterogeneous communication channels and the basic parameters for calculating the group-level matching degree between the transmission-dependent fragment groups and the multi-path heterogeneous communication channels.

2. The intelligent multi-channel fusion communication scheduling method based on edge computing according to claim 1, characterized in that, In step S1, the preprocessing of the operating status data of multiple heterogeneous communication channels, service fragmentation data, and transmission dependency data between service fragments includes the following steps: S1.1 Perform outlier removal and missing value filling on the operation status data, service fragmentation data, and transmission dependency data between service fragments of the multi-channel heterogeneous communication channels to obtain the cleaned basic data; S1.

2. The cleaned multi-channel heterogeneous communication channel operation status data, service fragment data, and transmission dependency data between service fragments are timestamped and unified to the same time series reference. S1.3 Perform dimension normalization processing on the time-aligned multi-channel heterogeneous communication status data, and perform structured encoding processing on the service fragment data and the transmission dependency data between service fragments. S1.

4. The normalized multi-channel heterogeneous communication status data, the structured encoded service fragment data, and the transmission dependency data between service fragments are encapsulated in chronological order into a preprocessed dataset.

3. The intelligent multi-channel fusion communication scheduling method based on edge computing according to claim 1, characterized in that, In step S2, the generation of the shard group-level scheduling policy includes the following steps: S2.1 Based on the transmission dependency data between preprocessed service fragments and the preprocessed service fragment data, several independent transmission dependency fragment groups are obtained according to the service transmission failure association dimension. S2.2 For each transmission-dependent fragment group, extract and configure the corresponding group-level transmission constraint parameters and group-level transmission failure judgment rules, and use them as the pre-constraint conditions for matching calculation. S2.

3. Using group-level transmission constraint parameters as prerequisites, the initial matching relationship between transmission-dependent fragment groups and multiple heterogeneous communication channels is calculated based on the preprocessed operating status data of the multi-channel heterogeneous communication channels. S2.

4. Based on the group-level transmission failure judgment rule, dynamically adjust the matching relationship between the same type of transmission dependent fragment group to be transmitted in the same batch and the corresponding multi-path heterogeneous communication channel according to the overall transmission result of the transmission dependent fragment group that has been transmitted. S2.5 Based on the adjusted matching relationship between the transmission dependency fragment group and the multiple heterogeneous communication channels, a fragment group-level multi-channel fusion communication scheduling strategy is generated in a structured manner according to the transmission timing sequence.

4. The intelligent multi-channel fusion communication scheduling method based on edge computing according to claim 3, characterized in that, In step S2.1, dividing the data into several independent transmission-dependent fragment groups based on the service transmission failure correlation dimension includes the following steps: S2.

11. Parse the preprocessed service fragment data and extract the frame type identifier, timing number and service semantic attributes of each service fragment; S2.

12. Based on the preprocessed transmission dependency data between service fragments, identify the set of service fragments with transmission failure associations; where: The transmission failure association refers to the fact that any abnormal transmission of any critical service fragment within the set will cause the entire set of service data to become unavailable. The critical business segment refers to the business segment that carries the core business semantics of the business segment set and determines the parsability and availability of the entire set of business data. S2.

13. Based on the timing sequence number and semantic attributes of the service fragments, encapsulate the set of service fragments with transmission failure associations into independent transmission-dependent fragment groups.

5. The intelligent multi-channel fusion communication scheduling method based on edge computing according to claim 3, characterized in that, In step S2.2, extracting and configuring the corresponding group-level transmission constraint parameters and group-level transmission failure determination rules includes the following steps: S2.

21. Based on the service type of the transmission-dependent fragmentation group, determine the three group-level transmission constraint parameters: the maximum allowable end-to-end delay of the entire group, the maximum allowable transmission timing difference of fragments within the group, and the maximum allowable number of packet losses in the entire group. S2.

22. Set group-level transmission failure judgment rules: When any one of the following conditions is met, the transmission failure of the entire transmission dependent fragment group is judged to be a transmission failure. S2.

23. Associate the group-level transmission constraint parameters and group-level transmission failure judgment rules as pre-constraint conditions with the attribute set of the corresponding transmission-dependent fragment group.

6. The intelligent multi-channel fusion communication scheduling method based on edge computing according to claim 3, characterized in that, In step S2.3, calculating the initial matching relationship between the transmission-dependent fragment group and the multiple heterogeneous communication channels includes the following steps: S2.

31. Extract the available bandwidth attribute, average transmission delay attribute, steady-state packet loss rate attribute, and channel fluctuation amplitude attribute of the multi-channel heterogeneous communication channels from the preprocessed operation status data of the multi-channel heterogeneous communication channels. S2.

32. Extract the total data size, importance level, latency sensitivity level, and group-level transmission constraint parameters of the transmission-dependent fragment group from the attribute set of the transmission-dependent fragment group. S2.

33. Using group-level transmission constraint parameters as hard constraint boundaries, calculate the initial matching degree values ​​of each transmission-dependent fragment group and multiple heterogeneous communication channels through multi-attribute weighted matching operations. The initial matching relationship between the transmission-dependent fragment group and multiple heterogeneous communication channels is formed by all the initial matching degree values.

7. The intelligent multi-channel fusion communication scheduling method based on edge computing according to claim 3, characterized in that, In step S2.4, dynamically adjusting the matching relationship between subsequent transmission-dependent fragment groups and multiple heterogeneous communication channels includes the following steps: S2.

41. Obtain the overall transmission result of the transmission-dependent fragment group that has been transmitted. Extract three types of feedback data: the overall transmission failure status, the timing difference deviation value within the group, and the overall packet loss deviation value. S2.

42. Taking the overall transmission failure state as the core correction trigger condition, and combining the timing difference deviation value within the group and the overall packet loss deviation value, calculate the matching degree correction amount of the corresponding multi-path heterogeneous communication channels. S2.43 Under the dual constraints of the preset upper limit of the correction step size and the maximum attenuation amplitude, the matching degree correction amount is used to update the matching degree value of the same type of transmission dependent fragment group to be transmitted in the same batch and the corresponding multi-path heterogeneous communication channel, so as to complete the dynamic adjustment of the matching relationship between the transmission dependent fragment group and the multi-path heterogeneous communication channel.

8. The intelligent multi-channel fusion communication scheduling method based on edge computing according to claim 3, characterized in that, In step S2.5, the structured generation of a fragment group-level multi-channel fusion communication scheduling strategy according to the transmission timing order includes the following steps: S2.

51. Based on the adjusted matching relationship between the transmission dependency fragment group and the multiple heterogeneous communication channels, match the main transmission channel and the alternative transmission channel for each transmission dependency fragment group. S2.

52. Configure the overall transmission priority, offloading ratio, and fragmentation timing coordination rules within each transmission-dependent fragmentation group; S2.

53. According to the transmission timing order of the transmission-dependent fragment groups, the channel matching results, overall transmission priority, splitting ratio and fragment timing coordination rules within the group of all transmission-dependent fragment groups are structurally encapsulated to generate a fragment group-level multi-channel fusion communication scheduling strategy.

9. The intelligent multi-channel fusion communication scheduling method based on edge computing according to claim 1, characterized in that, In step S3, the multi-channel packet cooperative transmission execution includes the following steps: S3.1 Analyze the multi-channel converged communication scheduling strategy at the fragment group level, extract the main transmission channel, alternative transmission channel, overall group transmission priority, diversion ratio and fragment timing coordination rules within the group corresponding to each transmission-dependent fragment group, and complete the transmission pre-configuration of multiple heterogeneous communication channels; S3.

2. According to the priority of the whole group transmission, the service fragment data is distributed to the corresponding multi-way heterogeneous communication channels for offloading transmission, based on the transmission-dependent fragment group and the fragment timing coordination rules within the group. S3.3 After all transmission dependent fragments in a single transmission group have been transmitted, collect the overall transmission results of the completed transmission dependent fragment group and synchronously send the overall transmission results back to S2 and S4 for matching relationship adjustment and scheduling effect verification.

10. The intelligent multi-channel fusion communication scheduling method based on edge computing according to claim 1, characterized in that, In step S4, the scheduling effect verification and parameter update include the following steps: S4.1 Based on the overall transmission results of the transmission-dependent fragment groups that have completed transmission, statistically analyze three types of performance indicators: the overall group transmission failure rate, the intra-group transmission timing difference compliance rate, and the overall group packet loss compliance rate. Verify the execution effect of the fragment group-level multi-channel fusion communication scheduling strategy and generate performance verification results. S4.

2. Based on the effect verification results, update the operating status benchmark data of the multi-channel heterogeneous communication channels, and use the updated operating status benchmark data of the multi-channel heterogeneous communication channels as the benchmark for the next round of operating status data collection and evaluation of the multi-channel heterogeneous communication channels. S4.

3. Based on the effect verification results, update the basic parameters for calculating the group-level matching degree of the transmission-dependent fragment group and the multi-path heterogeneous communication channel, and use the updated basic parameters for calculating the group-level matching degree of the transmission-dependent fragment group and the multi-path heterogeneous communication channel as the basis for the benchmark weight of the next round of group-level matching degree calculation.

Images