Method and system for joint scheduling of multi-link communication based on task adaptation

Abstract

The application discloses a multi-link communication joint scheduling method and system based on task adaptation, relates to the technical field of communication path scheduling, and obtains the running states of a graph data transmission link, a fiber link and a 4G / 5G link and the properties of to-be-transmitted tasks, forms multi-link observation data, analyzes the data volume, delay requirement, security level, continuous transmission demand and mobile accompanying characteristics of the tasks, and generates a transmission task portrait; according to the transmission task portrait, the task adaptation value, the link degradation value and the switching influence value of each link are calculated, a set of link scheduling factors is formed, and then a communication path scheduling scheme is generated, path optimization, shunt transmission and link pre-switching control are performed, and therefore the adaptability, continuity and transmission reliability of path scheduling in a multi-link communication scene are improved.

Description

Technical Field

[0001] This invention relates to the field of communication path scheduling technology, and more specifically to a multi-link communication joint scheduling method and system based on task adaptation. Background Technology

[0002] In multi-link communication scenarios, image and data transmission links, fiber optic links, and 4G / 5G links typically have different bandwidth, latency, security attributes, and stability characteristics. Existing communication path scheduling methods mostly rely on single link quality indicators or fixed priority rules for path selection, making it difficult to make differentiated judgments based on the data volume, latency requirements, security level, continuous transmission needs, and mobile accompanying characteristics of the task to be transmitted, resulting in insufficient matching between link selection and task constraints.

[0003] Meanwhile, when link status degrades or signal changes continuously in mobile scenarios, existing solutions often only perform handover after link failure or significant degradation in transmission quality. This lacks a coordinated assessment of link degradation trends, handover impacts, and traffic offloading order, easily leading to transmission interruptions, uneven utilization of link resources, and instability in high-priority data transmission. Therefore, there is an urgent need for a fusion assessment of multi-link status, task constraints, and handover costs to improve the continuity, adaptability, and reliability of communication path scheduling. Summary of the Invention

[0004] This invention addresses the shortcomings of existing technologies by proposing a multi-link communication joint scheduling method and system based on task adaptation.

[0005] The technical solution to achieve the objective of this invention is as follows:

[0006] On the one hand, the multi-link communication joint scheduling method based on task adaptation includes the following steps:

[0007] The system acquires the link operation status and pending transmission task attributes of image data transmission links, fiber optic links, and 4G / 5G links, performs time stamp alignment, link identification, and anomaly removal, and generates multi-link observation data.

[0008] Based on multi-link observation data analysis, the data volume, latency requirements, security level, continuous transmission requirements, and mobile accompanying characteristics of the task to be transmitted are analyzed to generate a transmission task profile;

[0009] Based on the transmission task profile, calculate the task adaptation value, link degradation value, and handover impact value of each communication link, eliminate links that do not meet connectivity and security requirements, and form a set of link scheduling factors;

[0010] Based on the link scheduling factor set, the particle swarm optimization algorithm is used to globally optimize the link role allocation, data diversion ratio and pre-switching trigger threshold, determine the main transmission link, backup transmission link, cooperative diversion link and data carrying order, generate a communication path scheduling scheme and execute data transmission, link diversion and pre-switching control.

[0011] Furthermore, the formation of the link scheduling factor set includes:

[0012] Extract link security constraints and latency constraints from the transmission task profile, and extract core parameters of the link operation status of the current observation time slice from multi-link observation data;

[0013] Based on the link connectivity status and link security constraints, links that do not meet the connectivity and security requirements are eliminated to obtain candidate links;

[0014] The task adaptation value is calculated based on the available bandwidth, end-to-end one-way delay, and packet loss count of the candidate links; the link degradation value is calculated based on the received signal strength, end-to-end one-way delay, packet loss count, and anomaly markers of the candidate links; and the handover impact value is calculated based on the currently used main link and candidate links.

[0015] The link identifier, task adaptation value, link degradation value, and handover impact value of the candidate links are associated and sorted to form a set of link scheduling factors.

[0016] Furthermore, the generation of the communication path scheduling scheme includes:

[0017] The comprehensive scheduling score is calculated based on the task adaptation value, link degradation value, and handover impact value of each candidate link in the link scheduling factor set.

[0018] The primary and backup transmission links are determined based on the comprehensive scheduling score ranking, and the collaborative diversion links are determined based on the continuous transmission requirements and the carrying capacity of the primary links;

[0019] Based on security level and latency requirements, the priority of data to be transmitted is determined, and the data transmission order and splitting ratio are determined.

[0020] The main transmission link, backup transmission link, collaborative diversion link, data carrying order, diversion ratio and pre-switching judgment rules are integrated to generate a communication path scheduling scheme.

[0021] Furthermore, the generation of the transmission task profile includes:

[0022] Extract the attributes of the task to be transmitted from multi-link observation data and generate a standardized dataset of task attributes;

[0023] Based on the standardized dataset of task attributes, the data volume level, latency constraint parameters and link security constraints are analyzed.

[0024] The continuous transmission requirement level is determined based on the data volume level and latency constraint parameters, and the estimated transmission duration of a single link is calculated. At the same time, the mobility-related characteristics are determined based on the change pattern of the received signal strength of the corresponding link at the transmission start and end nodes.

[0025] By binding the data volume level, estimated single-link transmission duration, latency constraint parameters, link security constraints, continuous transmission requirement level, mobile accompanying characteristic identifier, and task unique identifier, a transmission task profile is generated.

[0026] Furthermore, the formation of multi-link observation data includes:

[0027] Collect raw data on the operational status of image and data transmission links, fiber optic links, and 4G / 5G links, and obtain attribute data of tasks to be transmitted from the upper-layer task scheduling queue;

[0028] Timestamp compensation and time axis resampling are performed on the raw acquired data based on the global reference clock; the link identifier mapping table is queried according to the acquisition port number and the device number to which it belongs, and the link identifier is bound;

[0029] Abnormal data are removed based on physical range, historical operating benchmarks, mutation judgment thresholds, and logical consistency. The data is then compiled according to observation time slices and link identifiers to form multi-link observation data.

[0030] Further calculation of the task adaptation value includes:

[0031] Extract available bandwidth, end-to-end one-way latency, and packet loss count from the core parameters of the candidate link's operational status;

[0032] Bandwidth suitability is determined based on available bandwidth and candidate link available bandwidth benchmarks; latency satisfaction is determined based on end-to-end one-way latency and the maximum allowable end-to-end one-way latency threshold; and packet loss suitability is determined based on packet loss count.

[0033] The bandwidth adaptability, latency satisfaction, and packet loss adaptability are fused according to preset weights to obtain the task adaptability value.

[0034] Further calculation of the link degradation value includes:

[0035] Retrieve the core parameters and anomaly markers of the candidate link's operational status within the preset observation time slot;

[0036] The signal strength degradation rate is obtained based on the decreasing state of the received signal strength relative to the historical average, the delay degradation rate is obtained based on the increasing state of the end-to-end one-way delay relative to the historical average, and the packet loss degradation rate is obtained based on the increasing state of the packet loss rate relative to the historical average.

[0037] Determine the anomaly labeling correction factor based on the number of anomaly labels;

[0038] The link degradation value is obtained by fusing the signal strength degradation rate, latency degradation rate, packet loss degradation rate, and anomaly labeling correction coefficient according to preset weights.

[0039] Further calculations of the switching impact value include:

[0040] If there is no currently active primary link, set the switching impact value of the candidate link to the initial value;

[0041] When a currently used primary link exists, the inherent switching loss is determined based on the inherent switching delay benchmark of the candidate link and the maximum allowable end-to-end one-way delay threshold in the transmission task profile.

[0042] Performance difference loss is determined based on the end-to-end one-way latency difference between candidate links and currently used main links;

[0043] The handover impact value is obtained by integrating the inherent handover loss and performance difference loss according to the preset weights.

[0044] Furthermore, the execution of data transmission and link offloading includes:

[0045] According to the communication path scheduling scheme, the data to be transmitted is divided into transmission queues based on the bearer priority;

[0046] Control the main transmission link to carry high-priority data and the corresponding proportion of low-priority data, and control the cooperative diversion link to carry the remaining low-priority data;

[0047] During transmission, update the core parameters of the link's operating status and verify the link's connectivity and performance indicators.

[0048] When the link degradation value of the collaborative offloading link reaches the offloading adjustment condition, the offloading ratio of the collaborative offloading link is reduced, and the corresponding data is switched back to the main transmission link.

[0049] Secondly, a multi-link communication joint scheduling system based on task adaptation includes a link acquisition module, a task profiling module, a link evaluation module, and a path scheduling module:

[0050] The link acquisition module obtains the link operation status and transmission task attributes of the image data transmission link, fiber optic link and 4G / 5G link, performs time stamp alignment, link identification and anomaly removal, and forms multi-link observation data;

[0051] The task profiling module analyzes the data volume, latency requirements, security level, continuous transmission needs, and mobile accompanying characteristics of the task to be transmitted based on multi-link observation data to generate a transmission task profile;

[0052] The link evaluation module calculates the task adaptation value, link degradation value and handover impact value of each communication link based on the transmission task profile, eliminates links that do not meet connectivity and security requirements, and forms a set of link scheduling factors.

[0053] The path scheduling module uses a particle swarm optimization algorithm based on the link scheduling factor set to globally optimize the link role allocation, data diversion ratio and pre-switching trigger threshold. It determines the main transmission link, backup transmission link, cooperative diversion link and data carrying order, generates a communication path scheduling scheme and executes data transmission, link diversion and pre-switching control.

[0054] Compared with the prior art, the advantages of this invention are as follows:

[0055] 1. Based on multi-link observation data, a transmission task profile is generated, and the task adaptation value, link degradation value and handover impact value of each communication link are calculated in combination with the transmission task profile. After removing links that do not meet connectivity and security requirements, a set of link scheduling factors is formed, so that the communication path selection no longer depends on the quality of a single link, but can simultaneously consider task constraints, link status changes and handover costs, thereby improving the accuracy and adaptability of multi-link scheduling decisions.

[0056] 2. Based on the link scheduling factor set, determine the main transmission link, backup transmission link, cooperative diversion link and data carrying order, and combine data priority, diversion ratio and link degradation threshold to perform data transmission, link diversion and pre-switching control, so that high priority data is carried by stable links first, low priority data can be diverted according to link bandwidth, and backup link pre-connection is completed before the main link degrades, reducing the risk of transmission interruption caused by link switching. Attached Figure Description

[0057] Figure 1 This is a flowchart of the multi-link communication joint scheduling method based on task adaptation in this invention;

[0058] Figure 2 This is a schematic diagram illustrating the formation of multi-link observation data in this invention;

[0059] Figure 3 This is a schematic diagram illustrating the generation of the link scheduling factor set in this invention;

[0060] Figure 4 This is a schematic diagram of communication path scheduling and pre-switching control in this invention. Detailed Implementation

[0061] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments.

[0062] Example 1

[0063] This invention discloses a multi-link communication joint scheduling method based on task adaptation, comprising the following steps:

[0064] S1: Obtain the link operation status and pending transmission task attributes of the image data transmission link, fiber optic link, and 4G / 5G link, perform time stamp alignment, link identification, and anomaly removal to form multi-link observation data.

[0065] S2: Based on multi-link observation data, analyze the data volume, latency requirements, security level, continuous transmission requirements, and mobile accompanying characteristics of the task to be transmitted to generate a transmission task profile.

[0066] S3: Calculate the task adaptation value, link degradation value and handover impact value of each communication link based on the transmission task profile, eliminate links that do not meet connectivity and security requirements, and form a set of link scheduling factors.

[0067] S4: Based on the link scheduling factor set, the particle swarm optimization algorithm is used to globally optimize the link role allocation, data diversion ratio and pre-switching trigger threshold, determine the main transmission link, backup transmission link, cooperative diversion link and data carrying order, generate a communication path scheduling scheme, and execute data transmission, link diversion and pre-switching control according to the communication path scheduling scheme.

[0068] refer to Figure 1 , Figure 1 The flowchart shows the multi-link communication joint scheduling method based on task adaptation.

[0069] Figure 1 The diagram illustrates the overall process of acquiring the link operation status and transmission task attributes of graph data transmission links, fiber optic links, and 4G / 5G links, forming multi-link observation data, further generating transmission task profiles and link scheduling factor sets, and generating and executing communication path scheduling schemes based on these data.

[0070] It should be noted that, Figure 1 This is for illustrative purposes only, intended to help understand the process structure and data organization, and does not represent the precise working state in actual operation.

[0071] In step S1, the link operating status and pending transmission task attributes of the image data transmission link, fiber optic link, and 4G / 5G link are obtained. Time-stamp alignment, link identification, and anomaly removal are performed to form multi-link observation data, including:

[0072] S101: Multi-source raw data acquisition.

[0073] For image data transmission links, fiber optic links, and 4G / 5G links, raw data on the link operation status are collected according to a preset fixed sampling period of 200ms. This sampling period can cover the typical variation range of fast fading in 4G / 5G links and beam jitter in image data transmission links. At the same time, the total amount of collected data is controlled within the computing power carrying capacity threshold of the embedded scheduling unit. The attribute data of the tasks to be transmitted are obtained from the upper-layer task scheduling queue and summarized to form the raw data collection dataset.

[0074] The raw data of the link operation status is collected sequentially according to the sampling period. The heartbeat message reception status is directly obtained by each link receiver recording the heartbeat response message reception results in the corresponding period. The received signal strength value is collected by the transceiver module of each link receiver reading the signal strength value in the hardware register in each sampling period and reporting it. The transmission and reception time of the probe message is collected by the sender embedding the transmission timestamp in the probe message and the receiver synchronously recording the message arrival time to obtain paired time data. The packet loss count is calculated by the receiver counting the difference between the actual number of received messages in a single sampling period and the nominal number of messages sent by the sender.

[0075] Based on the aforementioned raw data of the link's operational status, the complete core parameters of the link's operational status are calculated. The link connectivity status is determined by the reception result of a single-cycle heartbeat response; a successful reception of a heartbeat response within a single sampling period indicates connectivity, while failure to receive a heartbeat response for two consecutive sampling periods indicates disconnection. The end-to-end one-way delay is obtained by subtracting the reception and transmission times of the same probe message, calculated as: One-way delay = Reception time - Transmission time. Available bandwidth is calculated iteratively and convergently using the packet loss count of the current sampling period and the historical throughput benchmark value as inputs. The historical throughput benchmark value is taken from the statistical average throughput of nearly 300 sampling periods during the link's steady-state operation. The attribute collection items for the task to be transmitted include the task's unique identifier, data volume, latency requirement level, security level, and transmission start and end nodes.

[0076] S102: Global timescale alignment processing.

[0077] Using the global reference clock synchronized by the system's PTP as a unified time source, standard timestamps are added to all data samples in the original acquisition dataset. Time axis resampling and alignment are performed on various types of data at different sampling frequencies according to a preset 1-second observation time slice granularity. This time slice granularity matches the minimum decision cycle of communication link scheduling, avoiding scheduling oscillations caused by fine granularity while ensuring real-time scheduling response. Linear interpolation is used to complete the values ​​at non-aligned moments, mapping all data samples to the same time reference sequence.

[0078] The local clock offset of each acquisition node is calculated through bidirectional synchronization message interaction: the master node, where the global reference clock is located, sends a synchronization message carrying the master node's sending timestamp T1 to each acquisition node at a fixed period. After receiving the synchronization message, the acquisition node records the local receiving timestamp T2 and immediately returns an acknowledgment message carrying the acquisition node's sending timestamp T3 to the master node. After receiving the acknowledgment message, the master node records the local receiving timestamp T4. The clock offset calculation formula is Offset=[(T2-T1)-(T4-T3)] / 2, where Offset is the offset value of the acquisition node's local clock relative to the global reference clock.

[0079] Timestamp offset compensation is performed on a per-node, per-sample basis: For all raw data samples reported by a single acquisition node, its original timestamp is superimposed with the clock offset Offset corresponding to that node to obtain the corrected standard timestamp, thereby eliminating the clock asynchronous difference between acquisition nodes of different links and completing the time stamp alignment of all data.

[0080] S103: Link identifier mapping and binding.

[0081] A link identifier mapping table is pre-built. The construction process is as follows: All physical communication links accessing the scheduling system are traversed, and the transmitting device number, receiving device number, and corresponding data acquisition port number of each link are registered sequentially according to the type order: image data transmission links, fiber optic links, and 4G / 5G links. A globally unique structured link identifier is generated according to the fixed rule of "link type code + transmitting device number + receiving device number". The link type code is set according to the link type: code 01 for image data transmission links, code 02 for fiber optic links, and code 03 for 4G / 5G links. The generated link identifiers are associated and stored one-to-one with the device number and acquisition port number of the corresponding link to form the link identifier mapping table. After time-stamp alignment, each link data sample queries the link identifier mapping table according to its acquisition port number and the device number it belongs to, matches the corresponding link identifier, and binds it to the data sample, so that each data sample carries a clear link affiliation identifier.

[0082] S104: Identification and Removal of Abnormal Data.

[0083] Based on the physical range of each data indicator, the historical operating benchmark interval, and the mutation judgment threshold, multi-dimensional abnormal data judgment rules are constructed in advance. For all data samples that have completed link identification binding, abnormal judgment is performed sample by sample to identify and remove invalid data samples.

[0084] The specific rules for identifying multi-dimensional anomalies include:

[0085] Among them, numerical out-of-range anomalies are determined by comparing the values ​​of each indicator in the data sample with the physical theoretical upper and lower limits of the corresponding indicators for that type of link. Samples with values ​​exceeding the upper and lower limits are judged as abnormal. The range threshold of received signal strength is set according to the rated parameters of each link's hardware. Values ​​exceeding the range have no physical meaning and are considered acquisition fault data.

[0086] The continuous missing data anomaly is determined by counting the number of valid data reports in consecutive observation time slots of the same link. A period of no valid data reports in three consecutive observation time slots is determined to be a data anomaly. This duration exceeds the maximum duration of normal fluctuations such as temporary link blockage and fast channel fading, and can be identified as a failure of the acquisition link or physical link.

[0087] Abnormal abrupt changes are determined by calculating the magnitude of the numerical change of the same indicator in two adjacent observation time slices. Samples whose magnitude of change exceeds the abrupt change determination threshold and have no corresponding link status change associated with supporting records are determined to be abnormal. The abrupt change determination threshold is preset to 80% of the historical baseline value of the indicator, which is used to filter numerical changes caused by normal channel fluctuations in the link, and at the same time effectively identify unreasonable jumps caused by abnormal acquisition.

[0088] Logical contradiction anomalies are determined by verifying the logical consistency of multiple indicators within the same data sample. The link connectivity status is marked as connected, but samples with infinite latency, 100% packet loss rate and no valid signal strength value are determined as logical contradiction anomalies.

[0089] After anomaly removal is completed, an anomaly marker is added to the observation time slice position corresponding to the removed sample, and the timestamp and link identifier corresponding to the position are retained for subsequent link degradation analysis.

[0090] S105: Compilation of multi-link observation data.

[0091] After time-stamp alignment, link identifier binding, and anomaly removal, the valid data samples are structured and organized according to the chronological order of the observation time slices and the classification of link identifiers to form standardized multi-link observation data. Each multi-link observation data contains four structured fields: standard timestamp, link identifier, link operation status parameter set, and attribute of the task to be transmitted. The core parameters of the link operation status include link connectivity, received signal strength, end-to-end one-way delay, available bandwidth, and packet loss count. The attribute data of the task to be transmitted includes a unique task identifier, data volume, latency requirement level, security level, and transmission start and end nodes. The final output of the multi-link observation data serves as the input basis for generating subsequent transmission task profiles.

[0092] refer to Figure 2 , Figure 2 This is a schematic diagram illustrating the formation of multi-link observation data.

[0093] Figure 2 This illustrates the process by which input data is processed through multi-source raw data acquisition, global time-stamp alignment, link identifier mapping and binding, abnormal data identification and removal, and multi-link observation data compilation, resulting in the output of multi-link observation data.

[0094] It should be noted that, Figure 2 This is for illustrative purposes only, intended to help understand the process structure and data organization, and does not represent the precise working state in actual operation.

[0095] In step S2, based on multi-link observation data, the data volume, latency requirements, security level, continuous transmission needs, and mobility characteristics of the task to be transmitted are analyzed to generate a transmission task profile, including:

[0096] S201: Extraction and standardization mapping of attributes of the task to be transmitted.

[0097] Extract the attribute data of the task to be transmitted from the multi-link observation data, perform standardized mapping processing on the extracted attributes, and unify the numerical expression form and judgment criteria.

[0098] Data volume is uniformly converted into scalar values ​​in bytes; latency requirement levels adopt a five-level quantization mapping rule, converting the original level identifier into a continuous quantization value from 1 to 5, with higher levels corresponding to larger quantization values, representing stronger latency constraints; security levels adopt a five-level quantization mapping rule, converting the original level identifier into a continuous quantization value from 1 to 5, with higher levels corresponding to larger quantization values, representing stronger security protection requirements. All attribute data after standardization mapping is bound one-to-one with the unique task identifier, forming a standardized task attribute dataset.

[0099] S202: Analysis of data volume characteristics.

[0100] Based on the data volume values ​​in the standardized dataset according to task attributes, data volume levels are categorized. Simultaneously, available bandwidth parameters for each communication link over the past 10 observation time slots are retrieved from the multi-link observation data to calculate the statistical mean of available bandwidth per link, further calculating the estimated transmission time of the task on a single link. Data volume levels are divided into three thresholds: less than 10MB is considered small volume, 10MB to 1GB is considered medium volume, and greater than 1GB is considered large volume. This threshold range matches the typical bandwidth carrying capacity of the three types of communication links, supporting rapid matching of subsequent link offloading strategies. The statistical mean of available bandwidth per link is the arithmetic mean of the available bandwidth parameters for the corresponding link over the past 10 observation time slots; the estimated transmission time per link is calculated by dividing the data volume value by the statistical mean of available bandwidth per link.

[0101] S203: Delay Requirement Feature Analysis.

[0102] Based on the quantized values ​​corresponding to the latency requirement levels in the standardized dataset of task attributes, latency constraint parameters corresponding to the task are mapped, including the maximum allowable end-to-end one-way latency threshold and the latency jitter threshold. When the latency requirement level quantization value is 1, the corresponding maximum allowable end-to-end one-way latency threshold is 1000ms; for each increase of one quantization level, the maximum allowable end-to-end one-way latency threshold decreases by 50%; when the quantization value is 5, the corresponding maximum allowable end-to-end one-way latency threshold is 10ms. The latency jitter threshold is taken as 20% of the corresponding maximum allowable end-to-end one-way latency threshold. This latency threshold gradient covers the typical latency range of three types of communication links and can adapt to transmission tasks with different real-time requirements.

[0103] S204: Security Level Feature Analysis.

[0104] Based on the quantified values ​​corresponding to the security levels in the standardized dataset of task attributes, link security constraints for each task are mapped, including link transmission permissions and data encryption requirements. Three types of communication links are pre-classified according to their transmission attributes: image-data transmission links and fiber optic links belong to private network attributes, while 4G / 5G links belong to public network attributes. For tasks with a security level quantified value of 3 or higher, core data is only allowed to be carried through private network attributes, and data encryption is enabled throughout the transmission process; for tasks with a security level quantified value below 3, public network attributes are allowed to participate in data transmission. This constraint rule balances the confidentiality of data transmission with the utilization of link resources.

[0105] S205: Analysis of Continuous Transmission Requirements and Accompanying Characteristics of Mobility.

[0106] The continuous transmission requirement level is determined by a combination of quantified values ​​for data volume level and latency requirement level. When the latency requirement level is greater than or equal to 3 and the data volume level is medium or higher, it is considered a high continuous transmission requirement; if only one condition is met, it is considered a medium continuous transmission requirement; if neither condition is met, it is considered a low continuous transmission requirement. A high continuous transmission requirement corresponds to zero-interruption transmission, meaning no continuous link disconnection is allowed during transmission; a low continuous transmission requirement allows for short-term transmission interruptions and data retransmission.

[0107] The motion-accompaniment characteristic is determined based on the historical received signal strength variation patterns of the links corresponding to the transmission start and end nodes. Received signal strength parameters for nearly 300 observation time slots of the three types of communication links under the corresponding transmission start and end nodes are retrieved, and the fluctuation amplitude of the received signal strength within a unit time period is statistically analyzed. If the fluctuation amplitude of the received signal strength exceeds 20dB, and the signal strength shows a continuous gradual change without sudden jumps, then the transmission task is determined to have motion-accompaniment characteristics. The 20dB fluctuation amplitude threshold can distinguish between sudden fluctuations caused by random obstruction from fixed nodes and the continuous signal changes of mobile nodes, ensuring the accuracy of the determination results.

[0108] S206: Structured generation of transmission task profiles.

[0109] The data volume level, estimated single-link transmission duration, latency constraint parameters, link security constraints, continuous transmission requirement level, mobile accompanying characteristic identifier, and task unique identifier are associated and bound to form a structured transmission task profile. The final output transmission task profile serves as the input basis for the subsequent generation of link scheduling factor set.

[0110] In step S3, based on the transmission task profile, the task adaptation value, link degradation value, and handover impact value of each communication link are calculated. Links that do not meet connectivity and security requirements are eliminated, forming a set of link scheduling factors, including:

[0111] S301: Initial screening of link admission criteria.

[0112] Link security constraints and latency constraints are extracted from the transmission task profile. Core parameters of the link's operational status in the current observation time slice are extracted from multi-link observation data. Connectivity and security checks are performed on each link. The connectivity check rule is as follows: links whose connectivity status is disconnected within the current observation time slice are considered to fail the connectivity requirements and are directly eliminated. The security check rule is as follows: if the security level quantification value in the transmission task profile is greater than or equal to 3, public network attribute links are considered to fail the security requirements and are directly eliminated; if the security level quantification value is less than 3, all links meet the security requirements. The links retained after the initial screening are designated as candidate links and proceed to the subsequent scheduling factor calculation stage.

[0113] S302: Calculation of task adaptation value.

[0114] For each candidate link, based on the core parameters of the link's operating status and the transmission task profile, three sub-items are calculated: bandwidth adaptability, latency satisfaction, and packet loss adaptability. The task adaptability value of the link is obtained by weighted summation. The task adaptability value ranges from [0,1]. The higher the value, the higher the degree of matching between the link and the transmission task.

[0115] Bandwidth adaptability is used to characterize the relative carrying capacity of the available bandwidth of a link, and is calculated using the following formula:

[0116] ,

[0117] in, For bandwidth adaptation of a single candidate link, This represents the available bandwidth of the current observation time slice for the candidate link. This represents the maximum available bandwidth for the current observation time slice across all candidate links. When hour, The value is set to 0 to avoid calculation errors caused by a zero denominator.

[0118] Delay satisfaction is used to characterize the degree to which link delay performance satisfies task delay constraints, and is calculated using the following formula:

[0119] ,

[0120] in, For the latency satisfaction of a single candidate link, The end-to-end one-way delay of the current observation time slice for this candidate link. This is the maximum permissible end-to-end one-way latency threshold for the task in the transmission task profile. When hour, The value is 1.

[0121] Packet loss adaptability is used to characterize the reliability level of link transmission and is calculated using the following formula:

[0122] ,

[0123] in, For packet loss adaptation of a single candidate link, This represents the packet loss rate of the candidate link in the current observation time slot. The packet loss rate is calculated as the ratio of the packet loss count of the link in the current observation time slot to the nominal number of packets sent by the sender. The packet loss count is taken from the core parameters of the link operation status in the multi-link observation data, and the nominal number of packets sent by the sender is synchronously reported to the acquisition unit by the sender of the corresponding link.

[0124] Task adaptation value Calculate using the following weighted formula:

[0125] ,

[0126] Among them, the weighting coefficient , , The weight allocation focuses on bandwidth transmission capacity and latency performance, matching the core requirements of transmission efficiency and real-time performance in industrial communication scenarios.

[0127] S303: Calculation of link degradation value.

[0128] For each candidate link, the core parameters of the link's operating status and the anomaly marker records within the last 10 observation time slices are retrieved. The three sub-items of signal strength degradation rate, delay degradation rate, and packet loss degradation rate are calculated respectively. Combined with the anomaly marker correction coefficient, the link degradation value of the link is obtained by weighted summation. The link degradation value ranges from [0,1]. The higher the value, the more severe the link status degradation and the lower the transmission reliability.

[0129] Signal strength degradation rate is calculated using the following formula:

[0130] ,

[0131] in, The signal strength degradation rate of a single candidate link. This is the arithmetic mean of the received signal strength over the last 10 observation time slots for the candidate link. The received signal strength of the candidate link in the current observation time slot; the lower limit of the calculation result is 0, that is, the degradation rate is 0 when the current signal strength is higher than the historical average. hour, The value is set to 0 to avoid calculation errors caused by a zero denominator.

[0132] The latency degradation rate is calculated using the following formula:

[0133] ,

[0134] in, The latency degradation rate of a single candidate link. This is the arithmetic mean of the end-to-end one-way delay over the last 10 observation time slices of the candidate link. This represents the end-to-end one-way delay of the current observation time slice for the candidate link; the lower limit of the calculation result is set to 0. When hour, The value is 0.

[0135] Packet loss degradation rate is calculated using the following formula:

[0136] ,

[0137] in, This represents the packet loss degradation rate for a single candidate link. This is the arithmetic mean of the packet loss rate over the past 10 observation time slices for this candidate link. This represents the packet loss rate of the candidate link in the current observation time slice; the lower limit of the calculation result is set to 0. When hour, The value is set to 0 to avoid calculation errors caused by a zero denominator.

[0138] The anomaly marker correction coefficient is determined based on the number of anomaly markers within the last 10 observation time slices. Its initial value is set to 0, corresponding to a steady-state link operation scenario with no anomaly markers within the statistical window. This represents the absence of additional degradation correction caused by abnormal data. For each anomaly marker present, the correction coefficient increases by 0.1, up to a maximum of 0.3, denoted as . The unit increment of 0.1 is obtained by engineering calibration and corresponds to the marginal attenuation value of the link steady-state transmission reliability of a single abnormal time slice. The upper limit of 0.3 corresponds to the extreme degradation scenario where abnormalities occur in three consecutive observation time slices. This matches the rule in the abnormal data identification process that if there is no valid data in three consecutive observation time slices, it is judged as a link failure. This prevents the accumulation of abnormal markers from causing the link degradation assessment results to deviate excessively from the actual operating state.

[0139] Link degradation value Calculate using the following weighted formula:

[0140] ,

[0141] Among them, the weighting coefficient , , The weight allocation focuses on the impact of packet loss performance on link degradation, which aligns with the core evaluation criteria for data transmission reliability.

[0142] S304: Calculation of switching impact value.

[0143] For each candidate link, the inherent switching loss and performance difference loss are calculated by combining the current operating status and transmission task profile of the main link in use. The switching impact value of the link is obtained by weighted summation. The switching impact value ranges from [0,1]. The higher the value, the greater the impact of transmission interruption and performance fluctuation caused by switching to the link.

[0144] If there is no primary link currently in use, the switching impact value for all candidate links is set to 0.

[0145] If a primary link is currently in use, the inherent switching loss is used to characterize the impact of the inherent interruption duration during the link switching process on the task, and is calculated using the following formula:

[0146] ,

[0147] in, The inherent handover loss of a single candidate link, This serves as the inherent handover latency benchmark for the candidate link type; the inherent handover latency benchmark for fiber optic links is 5ms, for image and data transmission links it is 20ms, and for 4G / 5G links it is 50ms. This benchmark corresponds to the typical time consumption for link layer handover processing for various types of links. The maximum allowable end-to-end one-way latency threshold for the task in the transmission task profile is set; the upper limit of the calculation result is 1.

[0148] Performance difference loss is used to characterize the transmission fluctuation impact caused by the difference in link delay performance before and after handover, and is calculated using the following formula:

[0149] ,

[0150] in, The performance difference loss for a single candidate link. The end-to-end one-way delay of the current observation time slice for this candidate link. This represents the end-to-end one-way delay of the current observation time slice of the currently used main link. This parameter is taken from the core parameters of the link's operational status corresponding to the link identifier in the multi-link observation data; the upper limit of the calculation result is 1. When hour, The value is set to 0 to avoid calculation errors caused by a zero denominator.

[0151] Switch Influence Value Calculate using the following weighted formula:

[0152] ,

[0153] Among them, the weighting coefficient , The weight allocation balance takes into account the inherent time consumption of the switching process and the impact of performance fluctuations after the switching.

[0154] S305: Structured generation of link scheduling factor set.

[0155] The link identifier, task adaptation value, link degradation value, and handover impact value corresponding to all candidate links are structurally associated to form single-link scheduling factor entries. All single-link scheduling factor entries are then sorted from highest to lowest task adaptation value and compiled into a link scheduling factor set. This link scheduling factor set serves as the input for generating subsequent communication path scheduling schemes.

[0156] refer to Figure 3 , Figure 3 A schematic diagram for generating the link scheduling factor set.

[0157] Figure 3 This paper illustrates the process of initial screening of link admission criteria based on transmission task profiles and multi-link observation data, obtaining candidate links through connectivity and security checks, and then calculating task adaptation value, link degradation value, and handover impact value respectively. After being associated and sorted by link identifiers, a set of link scheduling factors is formed.

[0158] It should be noted that, Figure 3This is for illustrative purposes only, intended to help understand the process structure and data organization, and does not represent the precise working state in actual operation.

[0159] In step S4, the primary transmission link, backup transmission link, cooperative offloading link, and data carrying order are determined based on the link scheduling factor set, a communication path scheduling scheme is generated, and data transmission, link offloading, and pre-switching control are executed according to the communication path scheduling scheme, including:

[0160] S401: Hierarchical selection of roles in the link.

[0161] Based on the task adaptation value, link degradation value, and handover impact value of each candidate link in the link scheduling factor set, the comprehensive scheduling score of a single candidate link is calculated. The comprehensive scheduling score ranges from [0,1], and the higher the value, the higher the priority of the comprehensive scheduling of the link.

[0162] The overall scheduling score is calculated using the following formula:

[0163] ,

[0164] in, The comprehensive scheduling score for a single candidate link. For the task adaptation value of the corresponding link, This represents the link degradation value for the corresponding link. The switching impact value for the corresponding link; weighting coefficient. , , The weight allocation prioritizes task suitability while considering link stability and switching costs, aligning with the comprehensive decision-making logic of multi-link scheduling. Calculation results less than 0 are assigned a value of 0 to avoid negative scores leading to abnormal sorting.

[0165] A particle swarm optimization algorithm is used to jointly optimize link role allocation, data diversion ratio, and pre-switching trigger threshold. The candidate link role code, diversion ratio coefficient, and two-level switching threshold are used as the dimensional components of the particles, and the particle position vector corresponds to a complete set of scheduling parameters. The overall performance of the scheduling scheme is used as the fitness function, and the fitness value is calculated by fusing the weighted overall scheduling score, transmission delay satisfaction rate, and link resource utilization of all in-service links with weights of 0.5:0.3:0.2. The inertia weight is set to 0.7, the individual learning factor to 1.5, the global learning factor to 1.5, the population size to 20, and the maximum number of iterations to 50. Half of the particles in the initial population are generated based on the basic scheme of descending order of the overall scheduling score, while the remaining particles are randomly generated, balancing the feasibility of the initial solution with population diversity. When the change in the optimal fitness value of the population over five consecutive generations is less than 0.01, the algorithm is considered converged, the iteration is terminated, and the globally optimal scheduling parameter combination is output.

[0166] The optimal link role allocation follows these constraints: The candidate link ranked first with a task adaptability value of not less than 0.6 is determined as the primary transmission link. This 0.6 threshold serves as the qualification benchmark for the primary link's carrying capacity, obtained through engineering calibration in industrial communication scenarios. The corresponding link can stably meet the core bandwidth, latency, and packet loss constraints of the task. If all candidate links have a task adaptability value below 0.6, the candidate link with the highest comprehensive scheduling score is selected as the primary transmission link, and its transmission performance risk is marked. The candidate link ranked second and with a link type different from the primary transmission link is determined as the backup transmission link. If only two candidate links exist, the second candidate link is directly determined as the backup transmission link.

[0167] If the continuous transmission demand level in the transmission task profile is high, or the available bandwidth of the main transmission link cannot support the peak transmission rate requirement of the task, then the link with the highest comprehensive scheduling score from the remaining candidate links is selected as the collaborative diversion link; the peak transmission rate requirement of the task is derived from the task data volume and the maximum allowable end-to-end one-way delay threshold, and is used to determine whether a single link can independently support the transmission task.

[0168] S402: Calculation of data carrying order and splitting ratio.

[0169] Based on the security level and latency requirement level in the transmission task profile, the priority of the data to be transmitted is divided. The higher the security level and the higher the latency requirement level, the higher the priority of the data. The data is divided into two categories: high priority data and low priority data.

[0170] The data carrying order follows the rule that the main link carries high-priority data first: the main transmission link carries all high-priority data, while the cooperative diversion link carries only low-priority data; the backup transmission link does not carry service data under normal transmission conditions, but only maintains periodic heartbeat message interaction and link status monitoring.

[0171] If a coordinated offloading link exists, the data offloading ratio is calculated based on the proportion of available bandwidth on that link. The low-priority data offloading ratio of the primary transmission link is calculated using the following formula:

[0172] ,

[0173] in, The proportion of low-priority data offloading to the primary transmission link. The available bandwidth for the current observation time slice of the main transmission link. The available bandwidth for the current observation time slice of the coordinated offloading link. The offloading ratio for low-priority data in the coordinated offloading link is... .when hour, If the value is 1, all data is carried by the main transmission link to avoid calculation errors caused by a denominator of zero.

[0174] The final traffic splitting ratio is determined by the particle swarm optimization algorithm based on the above benchmark value, with a fluctuation range not exceeding ±15%, and meets the link security constraints: for tasks with a security level of 3 or above, the collaborative traffic splitting link must be a private network link, and public network links are prohibited from carrying any core data.

[0175] S403: Structured generation of communication path scheduling schemes.

[0176] The link identifiers and role definitions of the main transmission link, backup transmission link, and collaborative diversion link are structurally integrated with the data carrying order, diversion ratio, and pre-switching trigger threshold to generate a communication path scheduling scheme for the corresponding task.

[0177] The communication path scheduling scheme includes seven structured fields: unique task identifier, primary transmission link information, backup transmission link information, collaborative diversion link information, data bearer priority rules, diversion ratio parameters, and pre-switching judgment rules. The scheme is bound to the unique task identifier, serving as the direct basis for subsequent transmission execution and link control.

[0178] S404: Data transmission and link splitting execution.

[0179] According to the communication path scheduling scheme, the data of the task to be transmitted is split according to the carrying priority and mapped to the sending queue of the corresponding transmission link. The main transmission link carries all high-priority data and the corresponding proportion of low-priority data according to the diversion ratio, while the cooperative diversion link carries the remaining proportion of low-priority data.

[0180] During transmission, the core parameters of each link's operational status are updated in real time at the observation time slice granularity, and the link connectivity and performance indicators are verified synchronously. When the link degradation value of a collaborative traffic splitting link exceeds 0.5, its splitting ratio is dynamically reduced, and the corresponding traffic is gradually switched back to the main transmission link to avoid the performance degradation of the splitting link affecting the overall transmission quality. A link degradation threshold of 0.5 corresponds to a link performance drop to half of its steady-state performance, making it unsuitable for continuously bearing the splitting load.

[0181] S405: Link pre-switching triggering and control.

[0182] Two levels of handover thresholds are preset, with default initial values ​​of 0.6 for the pre-handover trigger threshold and 0.8 for the handover execution threshold. The final thresholds are determined by particle swarm optimization algorithm combined with link degradation characteristics. The threshold gradient is calibrated based on the engineering statistical law of industrial communication link degradation. 0.6 corresponds to nodes where the link performance shows a continuous deterioration trend but still has sufficient pre-connection processing margin, while 0.8 corresponds to the critical state where the link is close to the failure boundary and service migration needs to be completed immediately. The two-level thresholds can complete pre-connection and service migration before the link completely fails, effectively reducing the handover interruption time.

[0183] The pre-switching process is triggered when the link degradation value of the main transmission link reaches the pre-switching trigger threshold, or when the transmission task has mobile accompanying characteristics and the received signal strength of the main transmission link shows a downward trend for three consecutive observation time slots with a cumulative decrease of more than 15dB.

[0184] The pre-switching process involves the following steps: activating the backup transmission link, completing link layer negotiation, transmission parameter configuration, and end-to-end path pre-connection; gradually migrating low-priority data from the primary transmission link to the backup transmission link, while high-priority data remains transmitted on the primary transmission link; and simultaneously verifying the connectivity and transmission performance of the backup transmission link to confirm its availability.

[0185] When the degradation value of the primary transmission link reaches the switching execution threshold, or when the primary transmission link becomes disconnected, a formal switch is executed: all remaining high-priority data is switched to the backup transmission link for transmission, completing a seamless handover of services; after the switch is completed, the link roles are updated, the original backup transmission link becomes the new primary transmission link, the original primary transmission link becomes the backup transmission link, the link scheduling factor set is updated synchronously, and the next scheduling cycle begins.

[0186] refer to Figure 4 , Figure 4 This is a schematic diagram of communication path scheduling and pre-switching control.

[0187] Figure 4 The process of calculating a comprehensive scheduling score and selecting link roles based on a set of link scheduling factors is shown to determine the primary transmission link, backup transmission link, and cooperative diversion link, and further to carry out data carrying and allocation and pre-switching control.

[0188] It should be noted that, Figure 4 This is for illustrative purposes only, intended to help understand the process structure and data organization, and does not represent the precise working state in actual operation.

[0189] This embodiment collects and standardizes the operational status and transmission task attributes of image data transmission links, fiber optic links, and 4G / 5G links to form multi-link observation data. Based on the multi-link observation data, task constraints are analyzed to generate transmission task profiles. Based on the transmission task profiles, a comprehensive evaluation of task adaptation, link degradation, and handover impact is performed on each communication link to form a link scheduling factor set. Finally, a communication path scheduling scheme is generated based on the link scheduling factor set, and data transmission, link splitting, and pre-handover control are executed, thereby improving the adaptability, continuity, and transmission reliability of path scheduling in multi-link communication scenarios.

[0190] Example 2

[0191] This invention discloses a multi-link communication joint scheduling system based on task adaptation, including a link acquisition module, a task profiling module, a link evaluation module, and a path scheduling module:

[0192] The link acquisition module obtains the link operation status and pending transmission task attributes of image data transmission links, fiber optic links, and 4G / 5G links, performs time stamp alignment, link identification, and anomaly removal, and forms multi-link observation data.

[0193] The task profiling module analyzes the data volume, latency requirements, security level, continuous transmission needs, and mobile accompanying characteristics of the task to be transmitted based on multi-link observation data to generate a transmission task profile.

[0194] The link evaluation module calculates the task adaptation value, link degradation value, and handover impact value of each communication link based on the transmission task profile, eliminates links that do not meet connectivity and security requirements, and forms a set of link scheduling factors.

[0195] The path scheduling module uses a particle swarm optimization algorithm based on the link scheduling factor set to globally optimize the link role allocation, data diversion ratio and pre-switching trigger threshold. It determines the main transmission link, backup transmission link, cooperative diversion link and data carrying order, generates a communication path scheduling scheme, and executes data transmission, link diversion and pre-switching control according to the communication path scheduling scheme.

[0196] The specific functional implementation of each module is described in the relevant content of the multi-link communication joint scheduling method based on task adaptation in Example 1, and will not be repeated here.

[0197] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A multi-link communication joint scheduling method based on task adaptation, characterized in that, Includes the following steps: The system acquires the link operation status and pending transmission task attributes of image data transmission links, fiber optic links, and 4G / 5G links, performs time stamp alignment, link identification, and anomaly removal, and generates multi-link observation data. Based on multi-link observation data analysis, the data volume, latency requirements, security level, continuous transmission requirements, and mobile accompanying characteristics of the task to be transmitted are analyzed to generate a transmission task profile; Based on the transmission task profile, calculate the task adaptation value, link degradation value, and handover impact value of each communication link, eliminate links that do not meet connectivity and security requirements, and form a set of link scheduling factors; Based on the link scheduling factor set, the particle swarm optimization algorithm is used to globally optimize the link role allocation, data diversion ratio and pre-switching trigger threshold, determine the main transmission link, backup transmission link, cooperative diversion link and data carrying order, generate a communication path scheduling scheme and execute data transmission, link diversion and pre-switching control.

2. The multi-link communication joint scheduling method based on task adaptation as described in claim 1, characterized in that, The formation of the link scheduling factor set includes: Extract link security constraints and latency constraints from the transmission task profile, and extract link operation status parameters for the current observation time slice from multi-link observation data; Based on the link connectivity status and link security constraints, links that do not meet the connectivity and security requirements are eliminated to obtain candidate links; The task adaptation value is calculated based on the available bandwidth, end-to-end one-way delay, and packet loss count of the candidate links; the link degradation value is calculated based on the received signal strength, end-to-end one-way delay, packet loss count, and anomaly markers of the candidate links; and the handover impact value is calculated based on the currently used main link and candidate links. The link identifier, task adaptation value, link degradation value, and handover impact value of the candidate links are associated and sorted to form a set of link scheduling factors.

3. The multi-link communication joint scheduling method based on task adaptation as described in claim 1, characterized in that, The generation of the communication path scheduling scheme includes: The comprehensive scheduling score is calculated based on the task adaptation value, link degradation value, and handover impact value of each candidate link in the link scheduling factor set. The primary and backup transmission links are determined based on the comprehensive scheduling score ranking, and the collaborative diversion links are determined based on the continuous transmission requirements and the carrying capacity of the primary links; Based on security level and latency requirements, the priority of data to be transmitted is determined, and the data transmission order and splitting ratio are determined. The main transmission link, backup transmission link, collaborative diversion link, data carrying order, diversion ratio and pre-switching judgment rules are integrated to generate a communication path scheduling scheme.

4. The multi-link communication joint scheduling method based on task adaptation as described in claim 1, characterized in that, The generation of the transmission task profile includes: Extract the attributes of the task to be transmitted from multi-link observation data and generate a standardized dataset of task attributes; Based on the standardized dataset of task attributes, the data volume level, latency constraint parameters and link security constraints are analyzed. The continuous transmission requirement level is determined based on the data volume level and latency constraint parameters, and the estimated transmission duration of a single link is calculated. At the same time, the mobility-related characteristics are determined based on the change pattern of the received signal strength of the corresponding link at the transmission start and end nodes. By binding the data volume level, estimated single-link transmission duration, latency constraint parameters, link security constraints, continuous transmission requirement level, mobile accompanying characteristic identifier, and task unique identifier, a transmission task profile is generated.

5. The multi-link communication joint scheduling method based on task adaptation as described in claim 1, characterized in that, The formation of multi-link observation data includes: Collect raw data on the operational status of image and data transmission links, fiber optic links, and 4G / 5G links, and obtain attribute data of tasks to be transmitted from the upper-layer task scheduling queue; Timestamp compensation and time axis resampling are performed on the raw acquired data based on the global reference clock; the link identifier mapping table is queried according to the acquisition port number and the device number to which it belongs, and the link identifier is bound; Abnormal data are removed based on physical range, historical operating benchmarks, mutation judgment thresholds, and logical consistency. The data is then compiled according to observation time slices and link identifiers to form multi-link observation data.

6. The multi-link communication joint scheduling method based on task adaptation as described in claim 2, characterized in that, The calculation of task adaptation values ​​includes: Extract available bandwidth, end-to-end one-way latency, and packet loss count from the core parameters of the candidate link's operational status; Bandwidth suitability is determined based on available bandwidth and candidate link available bandwidth benchmarks; latency satisfaction is determined based on end-to-end one-way latency and the maximum allowable end-to-end one-way latency threshold; and packet loss suitability is determined based on packet loss count. The bandwidth adaptability, latency satisfaction, and packet loss adaptability are fused according to preset weights to obtain the task adaptability value.

7. The multi-link communication joint scheduling method based on task adaptation as described in claim 2, characterized in that, The calculation of link degradation value includes: Retrieve the core parameters and anomaly markers of the candidate link's operational status within the preset observation time slot; The signal strength degradation rate is obtained based on the decreasing state of the received signal strength relative to the historical average, the delay degradation rate is obtained based on the increasing state of the end-to-end one-way delay relative to the historical average, and the packet loss degradation rate is obtained based on the increasing state of the packet loss rate relative to the historical average. Determine the anomaly labeling correction coefficient based on the number of anomaly labels; The link degradation value is obtained by fusing the signal strength degradation rate, latency degradation rate, packet loss degradation rate, and anomaly labeling correction coefficient according to preset weights.

8. The multi-link communication joint scheduling method based on task adaptation as described in claim 2, characterized in that, The calculation of the switching impact value includes: If there is no currently active primary link, set the switching impact value of the candidate link to the initial value; When a currently used primary link exists, the inherent switching loss is determined based on the inherent switching delay benchmark of the candidate link and the maximum allowable end-to-end one-way delay threshold in the transmission task profile. Performance difference loss is determined based on the end-to-end one-way latency difference between candidate links and currently used main links; The handover impact value is obtained by integrating the inherent handover loss and performance difference loss according to the preset weights.

9. The multi-link communication joint scheduling method based on task adaptation as described in claim 1, characterized in that, The execution of data transmission and link splitting includes: According to the communication path scheduling scheme, the data to be transmitted is divided into transmission queues based on the bearer priority; Control the main transmission link to carry high-priority data and the corresponding proportion of low-priority data, and control the cooperative diversion link to carry the remaining low-priority data; During transmission, update the core parameters of the link's operating status and verify the link's connectivity and performance indicators. When the link degradation value of the collaborative offloading link reaches the offloading adjustment condition, the offloading ratio of the collaborative offloading link is reduced, and the corresponding data is switched back to the main transmission link.

10. A task-adaptive multi-link communication joint scheduling system, used to implement the task-adaptive multi-link communication joint scheduling method according to any one of claims 1-9, characterized in that, include: The link acquisition module obtains the link operation status and transmission task attributes of the image data transmission link, fiber optic link and 4G / 5G link, performs time stamp alignment, link identification and anomaly removal, and forms multi-link observation data; The task profiling module analyzes the data volume, latency requirements, security level, continuous transmission needs, and mobile accompanying characteristics of the task to be transmitted based on multi-link observation data to generate a transmission task profile; The link evaluation module calculates the task adaptation value, link degradation value and handover impact value of each communication link based on the transmission task profile, eliminates links that do not meet connectivity and security requirements, and forms a set of link scheduling factors. The path scheduling module uses a particle swarm optimization algorithm based on the link scheduling factor set to globally optimize the link role allocation, data diversion ratio and pre-switching trigger threshold. It determines the main transmission link, backup transmission link, cooperative diversion link and data carrying order, generates a communication path scheduling scheme and executes data transmission, link diversion and pre-switching control.

Images