Data communication scheduling system based on virtual scheduling actuator

Abstract

The invention provides a data communication scheduling system based on a virtual scheduling executor, and relates to the technical field of data communication scheduling, the virtual scheduling executor is introduced into a traditional network-based data transmission architecture, and the virtual scheduling executor comprises a plurality of function modules. According to the method and the system, a plurality of link scheduling modules such as a strategy scheduling module, a scheduling execution module, a link state sampling and analysis module and a first data transmission server are arranged, so that data packet transmission link selection and link scheduling strategy decoupling control are realized, and the communication flexibility, scheduling intelligence and control configurability of the system in a multi-link environment are improved; the method is suitable for a multi-link concurrent and high-reliability data transmission scene.

Description

Technical Field

[0001] The present invention relates to the technical field of data communication scheduling, and in particular to a data communication scheduling system based on a virtual scheduling executor. Background Art

[0002] With the widespread deployment of multi-link networks, efficient and stable data communication across multiple heterogeneous links has become a critical issue that needs to be addressed in intelligent terminals, edge nodes, and distributed systems. In typical scenarios such as industrial robot cluster control, remote medical data transmission, collaborative edge computing management, and large-scale distributed monitoring, the system must be able to dynamically perceive link status and intelligently distribute data tasks across multiple physical channels to improve communication availability, bandwidth utilization, and system robustness.

[0003] To address these issues, patent application number CN201310162315.1, titled "A Network-Based Data Transmission Method, Apparatus, and System," proposes a multi-link scheduling method. This solution implements a virtual network device and data transmission service module to perform two-level encapsulation of original data packets. It then dynamically distributes data across links and schedules physical links based on performance metrics such as available link bandwidth and RTT (Round-Trip Time), significantly improving throughput and fault recovery capabilities in a multi-link transmission environment.

[0004] However, the above schemes have poor communication flexibility, scheduling intelligence and control configurability in a multi-link environment. Summary of the Invention

[0005] In view of this, a data communication scheduling system based on a virtual scheduling executor is provided to solve the problems of poor communication flexibility, scheduling intelligence and control configurability in the prior art in a multi-link environment.

[0006] The present invention adopts the following technical solutions:

[0007] A data communication scheduling system based on a virtual scheduling executor includes: a sending end component and a receiving end component; the sending end component includes a first virtual network device, a first protocol server, a virtual scheduling executor, and a physical link resource pool; the virtual scheduling executor includes a policy scheduling module, a scheduling execution module, a link status sampling and analysis module, and a first data transmission server;

[0008] The first virtual network device is configured to, upon receiving original data sent by the application layer, encapsulate the original data into a network layer-like message and send the message to the first protocol server;

[0009] The first protocol server is used to encapsulate the network layer-like message based on the TCP and IP protocols, and perform fragmentation according to the MTU rule to obtain a transmittable first data packet, and send it to the first data transmission server;

[0010] The first data transmission server is configured to encapsulate the first data packet, add system-level scheduling header information to the first data packet, obtain a second data packet, and store the second data packet in the task scheduling queue managed by the first data transmission server; the system-level scheduling header information includes a recommended physical link ID, a task identifier, a slice sequence number, a slice flag, and task tag information; the first data transmission server is further configured to send a scheduling request to the policy scheduling module, the scheduling request including context parameters of the current task;

[0011] When receiving the scheduling request, the policy scheduling module generates a candidate physical link set based on the scheduling request by calling the link state sampling and analysis module, and sends the candidate physical link set to the scheduling execution module;

[0012] The scheduling execution module generates a first scheduling instruction according to the candidate physical link set, and sends the first scheduling instruction to the first data transmission server;

[0013] The first data transmission server is further configured to complete the binding between the second data packet and the physical link and actually send it according to the first scheduling instruction, so that the receiving end component receives the second data packet;

[0014] The physical link resource pool includes a resource set of multiple physical links and their state management modules, and provides link state information of the physical links for access by the link state sampling and analysis module.

[0015] Optionally, the task tag information is a priority tag or a service type code;

[0016] The priority tag is used to identify the transmission priority of the second data packet;

[0017] The service type code is used to identify the quality of service requirement of the second data packet.

[0018] Optionally, the context parameters of the current task include a task identifier, a target address, a data length, a service tag, a priority level, and whether fragmented transmission is allowed.

[0019] Optionally, the link state sampling and analysis module includes a link state perception unit, a link state cache unit, a multi-physical link optimization unit and an optimization perception distribution unit;

[0020] When receiving the scheduling request, the policy scheduling module generates a candidate physical link set based on the scheduling request by calling the link status sampling and analysis module, specifically including:

[0021] When receiving the scheduling request, the policy scheduling module obtains the first target link state information of each physical link in the physical link resource pool from the link state cache unit, and selects a target policy from a plurality of preset policies based on the context parameters of the current task and the first target link state information; wherein the link state sensing unit is used to collect the link state information of each physical link in the physical link resource pool in real time, and the link state cache unit is used to perform short-term caching and preprocessing on the link state information collected by the link state sensing unit;

[0022] The policy scheduling module calls the multi-physical link optimization unit to calculate the scores of all the physical links based on the target policy; the multi-physical link optimization unit is used to calculate the scores of all the physical links according to the target policy and the first target link state information;

[0023] The policy scheduling module calls the optimization-aware distribution unit to generate the candidate physical link set; the optimization-aware distribution unit is used to screen all the physical links according to the scores of all the physical links to obtain the candidate physical link set.

[0024] Optionally, the strategy includes:

[0025] Minimum delay priority strategy, used to prioritize the physical link with the lowest link delay for data transmission;

[0026] The maximum bandwidth margin priority strategy is used to prioritize physical links with more surplus bandwidth resources for data transmission;

[0027] The service label and physical link QoS label adaptation strategy is used to match the service type label of the task with the QoS label of the physical link, and schedule the corresponding physical link for data transmission based on the matching result;

[0028] The historical scheduling success rate weighted physical link strategy is used to weight each physical link based on the success rate of historical scheduling tasks, and schedule the corresponding physical link for data transmission according to the weighted score result.

[0029] Optionally, the strategy also includes: a physical link trend pre-selection strategy based on an AI prediction model.

[0030] Optionally, the link status sampling and analysis module further includes: an abnormality scheduling unit;

[0031] The first data transmission server is used to return task confirmation information to the scheduling execution module after completing the sending of all the second data packets of the entire task;

[0032] The scheduling execution module comprehensively detects whether a network anomaly meeting preset conditions occurs during the task transmission process based on the task confirmation information and the link status information collected in real time by the link status perception unit, and forms a detection result;

[0033] When the detection result shows that a network abnormality meeting a preset condition occurs during the task transmission process, the scheduling execution module sends a rerouting instruction including the abnormality type and priority to the abnormality scheduling unit;

[0034] When receiving the rerouting instruction, the abnormal scheduling unit selects the best replacement link from the currently available physical links in combination with the latest link performance score and historical scheduling success rate of each physical link provided by the link status cache unit, and feeds back the optimal replacement link to the scheduling execution module;

[0035] The scheduling execution module regenerates a second scheduling instruction based on the optimal alternative link and sends it to the first data transmission server, while synchronizing the task status and data packet sequence information to ensure the transmission order and integrity of the data;

[0036] After receiving the second scheduling instruction, the first data transmission server switches to the optimal alternative link according to the second scheduling instruction to continue transmitting the remaining data packets;

[0037] When the detection result shows that a network anomaly that meets preset conditions occurs during the task transmission process, the scheduling execution module dynamically adjusts its own parameters based on the task confirmation information and the link status information collected in real time by the link status perception unit.

[0038] Optionally, the receiving end component includes: a second data transmission server, a second protocol server and a second virtual network device;

[0039] The second data transmission server is configured to, upon receiving the second data packets from multiple physical links, perform fragment classification, out-of-order recovery, integrity verification, and data splicing on the second data packets, and send the spliced third data packets to the second protocol server;

[0040] The second protocol server is used to perform a decapsulation operation based on the TCP and IP protocols on the third data packet to obtain original application layer data and send the original application layer data to the second virtual network device;

[0041] The second virtual network device is used to temporarily store the original application layer data.

[0042] Optionally, the virtual scheduling executor further includes: a policy configuration and hot loading interface module;

[0043] The policy configuration and hot loading interface module is used to support the policy scheduling module to communicate with external devices, receive remote or local policy configurations, and implement policy hot plugging, version switching, and link status visualization.

[0044] Optionally, the receiving end component is further configured to feed back a task completion status and a link feedback result to the policy scheduling module after processing all the second data packets;

[0045] The policy scheduling module is further configured to optimize the corresponding policy according to the task completion status and link feedback results.

[0046] The multi-link data transmission system of the prior art generally adopts a fixed policy scheduling or static link allocation scheme, which makes it difficult for the prior art to dynamically adapt to changes in link status in a complex network environment, and lacks a unified scheduling hub and modular expansion capabilities, resulting in limited system flexibility and intelligence. Compared with the prior art, the present invention introduces a virtual scheduling executor into the traditional network-based data transmission architecture. The virtual scheduling executor includes multiple functional modules, such as a policy scheduling module, a scheduling execution module, a link status sampling and analysis module, and a first data transmission server. These functional modules are centrally managed and modularly packaged, so that the present invention realizes the decoupling control of data packet transmission link selection and link scheduling strategy, can independently upgrade and replace strategies, and has modular and plug-in capabilities, thereby improving the system's communication flexibility, scheduling intelligence, and control configurability in a multi-link environment, and is suitable for multi-link concurrent and high-reliability data transmission scenarios.

[0047] Furthermore, this invention proposes and implements for the first time a multi-link scheduling architecture centered around a virtual scheduling executor. This architecture integrates a task-context-driven link selection mechanism, a multi-strategy scheduling system based on link status scoring, an anomaly detection and rerouting mechanism, and a policy management interface that supports hot reloading and remote configuration. This innovative architecture enables the system to decouple link scheduling and control, dynamically optimize, and elastically scale without relying on underlying protocol modifications. This significantly improves data scheduling performance, fault tolerance, and application adaptability in multi-link transmission scenarios, breaking through the bottlenecks of existing technologies. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0049] Figure 1 1 is a structural diagram of a data communication scheduling system based on a virtual scheduling executor provided by an embodiment of the present invention;

[0050] Figure 2 This is a schematic diagram of a data reorganization process provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0051] To make the objectives, technical solutions, and advantages of the present invention more apparent, the technical solutions of the present invention will be described in detail below. Obviously, the embodiments described are only some of the embodiments of the present invention, not all of them. Based on the embodiments of the present invention, all other implementations obtained by those of ordinary skill in the art without inventive effort are within the scope of protection of the present invention.

[0052] Figure 1 This is a structural diagram of a data communication scheduling system based on a virtual scheduling executor provided by an embodiment of the present invention. Figure 1 As shown, the system includes: a sending end component 11 and a receiving end component 12.

[0053] The sending end component 11 includes a first virtual network device 111 , a first protocol server 112 , a virtual scheduling executor 113 and a physical link resource pool 114 .

[0054] The virtual scheduling executor 113 includes a policy scheduling module 1131 , a scheduling execution module 1132 , a link status sampling and analysis module 1133 and a first data transmission server 1134 .

[0055] The sending end component 11 further includes an application layer data processing module (not shown in the figure) for generating original data and sending the original data to the first virtual network device 111 .

[0056] The first virtual network device 111 acts as a bridge between user mode and kernel mode in the system. This device can be implemented using TUN (Network TUNnel, a virtual point-to-point network device) or TAP (Network TAP, a virtual Ethernet network device) technology, creating a virtual network interface within the operating system kernel to support data injection and reading. When the first virtual network device 111 receives raw data from the application layer (i.e., the application layer data processing module), it encapsulates the raw data into a network layer-like message and sends it to the first protocol server 112. The network layer-like message can be IP (Internet Protocol) layer data.

[0057] The first protocol server 112 is configured to encapsulate network layer packets based on TCP (Transmission Control Protocol) and IP protocols, and fragment them according to the MTU (Maximum Transmission Unit) rule to obtain a transmittable first data packet, which is then sent to the first data transmission server 1134. The encapsulation process here includes adding standard transport layer and network layer headers to generate a complete encapsulated data packet that complies with network protocol specifications. For example, the transport layer may be TCP (Transmission Control Protocol) or UDP (User Datagram Protocol), and the network layer may be IPv4 or IPv6.

[0058] The first data transmission server 1134 is set in the VEE (Virtual Execution Environment) and is used to encapsulate the first data packet, add system-level scheduling header information to the first data packet, obtain the second data packet, and store it in the task scheduling queue it manages; the system-level scheduling header information includes the recommended physical link ID, task identifier, fragment sequence number, fragment flag and task tag information. Among them, the recommended physical link ID is used to indicate the physical link recommended for use; the task identifier can be a task number, which is used by the receiving end component 12 to reorganize and classify data; the fragment sequence number and fragment flag are used by the receiving end component 12 to perform out-of-order recovery and integrity detection of data packets; the task tag information is used by the receiving end component 12 to identify and provide policy feedback for data packets. The task tag information can be a priority tag or a service type code. The priority tag is used to identify the transmission priority of the second data packet, and the service type code is used to identify the service quality requirements of the second data packet.

[0059] Furthermore, when a second data packet in the task scheduling queue managed by the first data transmission server 1134 triggers a scheduling request trigger mechanism, the first data transmission server 1134 is further configured to send a scheduling request to the policy scheduling module 1131. The scheduling request includes the context parameters of the current task, including the task identifier, target address, data length, service tag (e.g., video, control, file, etc.), latency tolerance, historical transmission records, priority level, and whether fragmented transmission is permitted. The scheduling request trigger mechanism can be any of four methods: cache data arrival, task queue backlog, reaching a transmission latency threshold, or timed polling trigger.

[0060] When receiving a scheduling request, the policy scheduling module 1131 generates a candidate physical link set based on the scheduling request by calling the link status sampling and analysis module, and sends the candidate physical link set to the scheduling execution module 1132 .

[0061] The scheduling execution module 1132 is configured to generate a first scheduling instruction based on the candidate physical link set and send the first scheduling instruction to the first data transmission server 1134. The first scheduling instruction specifies the physical link corresponding to the relevant second data packet.

[0062] The first data transmission server 1134 is further configured to complete the binding and actual delivery between the second data packet and the corresponding physical link according to the first scheduling instruction, so that the receiving end component 12 receives the second data packet.

[0063] The physical link resource pool 114 contains a resource collection of multiple physical links and their state management modules, and provides the link state information of the physical link for access by the link state sampling and analysis module 1133. Specifically, the physical link resource pool 114 is used to perform unified abstraction, state registration and scheduling control for all available physical links in the system. Each physical link (such as a network card interface or a multi-hop forwarding link) is encapsulated as a resource object in the system and has attributes such as a unique identifier, performance indicator cache and scheduling state tag. It should be noted that the physical link resource pool 114 itself is not responsible for data forwarding. Its responsibility is to maintain link state information for access by the link state sampling and analysis module 1133. The data is actually transmitted by the first data transmission server 1134 through the bound network interface.

[0064] The present invention adopts the above technical solution, by introducing a virtual scheduling executor into the traditional network-based data transmission architecture. The virtual scheduling executor includes multiple functional modules, such as a policy scheduling module, a scheduling execution module, a link state sampling and analysis module, and a first data transmission server. These functional modules are centrally managed and modularly encapsulated, so that the present invention can achieve decoupled control of data packet transmission link selection and link scheduling strategy, and can independently upgrade and replace strategies. It has modular and plug-in capabilities, thereby improving the system's communication flexibility, scheduling intelligence, and control configurability in a multi-link environment, and is suitable for multi-link concurrent and high-reliability data transmission scenarios. In addition, by using the virtual scheduling executor as an independent scheduling control center, link state acquisition and policy execution can be centralized, scheduling behavior is unified, and the system's responsiveness to dynamic network environments and task scheduling efficiency are improved.

[0065] In the embodiment of the present invention, the link state sampling and analysis module 1133 includes a link state sensing unit 11331 , a link state caching unit 11332 , a multi-physical link optimization unit 11333 and an optimization perception distribution unit 11334 .

[0066] When receiving a scheduling request, the policy scheduling module 1131 generates a candidate physical link set based on the scheduling request by calling the link state sampling and analysis module 1133. Specifically, the generation of the candidate physical link set may include:

[0067] (1) Upon receiving a scheduling request, the policy scheduling module 1131 obtains the first target link state information of each physical link in the physical link resource pool 114 from the link state cache unit 11332 and selects a target policy from a plurality of preset policies based on the context parameters of the current task and the first target link state information. The link state sensing unit 11331 is used to collect the link state information of each physical link in the physical link resource pool in real time, and the link state cache unit 11332 is used to perform short-term caching and preprocessing on the link state information collected by the link state sensing unit 11331, thereby providing a time synchronization view and timing comparison capability for the multi-physical link optimization unit. The first target link state information may include bandwidth utilization, RTT (Round-Trip Time), jitter, packet loss rate, I / O occupancy, current load and availability, and historical task scheduling performance, such as success rate and transmission time.

[0068] (2) The policy scheduling module 1131 calls the multi-physical link optimization unit 11333 to calculate the scores of all physical links based on the target policy; the multi-physical link optimization unit is used to weightedly calculate the scores of all physical links according to the target policy and the first target link state information.

[0069] (3) The policy scheduling module 1131 calls the optimization perception distribution unit 11334 to generate a candidate physical link set; the optimization perception distribution unit is used to screen all physical links according to the scores of all physical links to obtain the candidate physical link set.

[0070] The present invention collects the link status information of each physical link in the physical link resource pool in real time through a link status perception unit, the link status cache unit performs short-term cache and preprocessing on the link status information collected by the link status perception unit, the multi-physical link optimization unit calculates the scores of all physical links according to the target strategy and the first target link status information, and the optimization perception distribution unit screens all physical links according to the scores of all physical links to obtain a candidate physical link set, thereby realizing dynamic selection and load balancing of multiple links.

[0071] It should be noted that when this system is actually used, it is first necessary to initialize and register the link. Specifically, after the virtual scheduling executor is started, the network interface enumeration API (Application Programming Interface) provided by this system is called to identify the enabled physical network interfaces in the system (such as Ethernet, Wi-Fi, VPN, 5G modules, etc.). The link state perception unit 11331 performs real-time reachability detection and performance sampling on each link, and the link state cache unit 11332 archives the data collected by the link state perception unit 11331, and the virtual scheduling executor constructs the corresponding link object in the physical link resource pool 114. The system marks the link with communication capability as "schedulable state". The link state sampling of the link state perception unit 11331 supports periodic triggering and event triggering mechanisms, and is compatible with active detection and passive monitoring logic.

[0072] In the embodiment of the present invention, the strategy may include:

[0073] The minimum latency priority strategy prioritizes data transmission over the physical link with the lowest latency. Latency refers to the time it takes for a data packet to travel from the sender to the receiver. Lower latency means higher transmission efficiency. Therefore, this strategy is suitable for tasks with high real-time requirements, such as video conferencing and online gaming.

[0074] The maximum bandwidth margin priority policy prioritizes data transmission over physical links with sufficient bandwidth resources. This ensures sufficient bandwidth for data transmission and avoids congestion and packet loss caused by bandwidth bottlenecks. Bandwidth margin refers to the remaining bandwidth after subtracting the currently used bandwidth from the link's available bandwidth. Prioritizing links with the largest bandwidth margin during scheduling helps improve overall system throughput and transmission stability, making it suitable for high-traffic tasks or services with high bandwidth requirements.

[0075] The service label and physical link QoS label adaptation strategy matches the task's service type label (such as video, voice, or data) with the physical link's QoS (Quality of Service) label, and schedules data transmission over the corresponding physical link based on the matching result. Different types of data transmission have different network performance requirements. For example, voice requires low latency and high reliability, while video requires a larger bandwidth. This strategy ensures that each service type is matched to the most appropriate physical link, improving user experience and service quality.

[0076] The weighted physical link strategy based on historical scheduling success rate assigns a weighted score to each physical link based on the success rate of historically scheduled tasks. Data transmission is scheduled on the corresponding physical link based on this weighted score. The historically scheduled task success rate refers to the proportion of tasks that successfully completed transmissions over the corresponding physical link within a preset historical time period. A link with good historical performance indicates high stability and reliability, thereby improving the overall system transmission success rate and robustness.

[0077] In an embodiment of the present invention, the strategy may further include: a physical link trend pre-selection strategy based on an AI prediction model. The execution process of this strategy includes:

[0078] a) Constructing an input feature vector based on the historical link state information of each physical link and the first target link state information.

[0079] b) Input the input feature vector into a preset time series prediction model to use the time series prediction model to predict the future link status trend of each physical link, and obtain the trend score of the link status information (bandwidth utilization, RTT, jitter, packet loss rate, IO occupancy, current load and availability, and historical task scheduling performance) of each physical link in the future preset time period output by the model (i.e., based on the target strategy, the score of the physical link calculated by calling the multi-physical link optimization unit 11333). This strategy can improve the system's response speed to link status fluctuations, enhance the accuracy of scheduling decisions and the stability of network transmission. It should be noted that the training method of the time series prediction model is a prior art and will not be repeated here.

[0080] In the embodiment of the present invention, the policy scheduling module 1131 selects a target policy from a plurality of preset policies based on the context parameters of the current task and the first target link state information, which may specifically include:

[0081] (1) According to the context parameters of the current task, determine whether the priority of the current task is high priority. If the priority of the current task is high priority, select the service label and physical link QoS label adaptation strategy as the target strategy.

[0082] (2) If the priority of the current task is not high, then based on the first target link status information, determine whether the RTT difference between the currently available physical links reaches a preset threshold. If the RTT difference between the currently available physical links reaches the preset threshold, then select the minimum delay priority strategy as the target strategy.

[0083] (3) If the RTT difference between the currently available physical links does not reach the preset threshold, then based on the context parameters of the current task, determine whether the task volume of the current task exceeds the preset volume threshold. If the task volume of the current task exceeds the preset volume threshold, then select the physical link strategy weighted by the historical scheduling success rate as the target strategy.

[0084] (4) If the task volume of the current task does not exceed the preset volume threshold, then based on the context parameters of the current task, it is determined whether the current task hits the preset AI prediction model. If the current task hits the preset AI prediction model, the physical link trend pre-selection strategy based on the AI prediction model is selected as the target strategy.

[0085] (5) If the current task does not hit the preset AI prediction model, the maximum bandwidth margin priority strategy is selected as the target strategy.

[0086] The policy decision tree supports multi-dimensional policy fusion and custom policy combinations. The policy module supports plug-in deployment and can dynamically load new models or rules to ensure the continuous evolution of the system's scheduling logic. Through a policy-driven link selection mechanism, link scheduling and transmission links are dynamically bound, allowing flexible scheduling based on task attributes and network congestion, significantly reducing the risk of network congestion and improving link utilization and data transmission quality.

[0087] In the embodiment of the present invention, the link status sampling and analysis module 1133 may further include: an abnormality scheduling unit 11335 .

[0088] The first data transmission server 1134 is configured to transmit task confirmation information back to the scheduling execution module 1132 after completing the sending of all second data packets of the entire task.

[0089] Based on the task confirmation information and the link status information collected in real time by the link status sensing unit 11331, the scheduling execution module 1132 comprehensively detects whether network anomalies that meet preset conditions occur during the task transmission process, and generates a detection result. The preset conditions may include indicators in the link status information exceeding the corresponding standard value range and task level anomalies.

[0090] When the detection results indicate that a network anomaly meeting preset conditions occurs during task transmission, the scheduling execution module 1132 sends a rerouting instruction containing the anomaly type and priority to the anomaly scheduling unit 11335. Network anomalies meeting the preset conditions may include link interruption, link congestion, RTT (Round-Trip Time) exceeding a preset duration threshold, and packet loss rate exceeding a preset packet loss rate threshold.

[0091] When receiving the rerouting instruction, the abnormal scheduling unit 11335 selects the best replacement link from the currently available physical links based on the latest link performance score and historical scheduling success rate of each physical link provided by the link status cache unit, and feeds it back to the scheduling execution module 1132.

[0092] The scheduling execution module 1132 regenerates a second scheduling instruction based on the optimal alternative link and sends it to the first data transmission server 1134, while synchronizing the task status and data packet sequence information to ensure the transmission order and integrity of the data.

[0093] After receiving the second scheduling instruction, first data transmission server 1134 switches to the optimal alternative link according to the second scheduling instruction to continue transmitting the remaining data packets. In this way, unfinished data fragments are reallocated to the new link without interrupting the task, and the original link resources are released or marked as "unschedulable." By scheduling link switching for tasks when network anomalies occur, the system ensures high availability and fault tolerance, guarantees data transmission continuity and overall system stability, and supports link disablement and recovery control mechanisms.

[0094] Furthermore, if the detection results indicate that no network anomalies that meet preset conditions have occurred during task transmission, the scheduling execution module dynamically adjusts its own parameters based on task confirmation information and link status information collected in real time by the link status sensing unit. This helps improve the system's ability to adapt to dynamic changes in complex networks.

[0095] In the embodiment of the present invention, the receiving end component 12 includes: a second data transmission server 123 , a second protocol server 122 and a second virtual network device 121 .

[0096] Upon receiving second data packets from multiple physical links, the second data transmission server 123 submits these second data packets to its data reassembly module. The data reassembly module classifies the fragments, performs out-of-order recovery, performs integrity verification, and performs data splicing based on the system-level scheduling header information (including the physical link ID, task identifier, fragment sequence number, fragment flag, and task tag information) of the second data packets. This process supports hash mapping and multi-queue caching mechanisms. Furthermore, the spliced third data packet is sent to the second protocol server 122.

[0097] Specifically, the data reorganization module is the key processing unit for cross-link data fusion and task-level restoration. Figure 2 This is a schematic diagram of a data reorganization process provided by an embodiment of the present invention. Figure 2 As shown, when the data reassembly module receives second data packets from multiple physical links, it identifies and classifies the second data packets (data fragments) based on the system-level scheduling header information of each second data packet. The data reassembly module uses a hash table structure to quickly index the task number (i.e., task identifier), combines the task queue to buffer the task fragments in chronological order, and uses a multi-channel cache mechanism to store the second data packets from different physical links separately. The above structures work together to ensure normalized data storage and efficient scheduling in a multi-link concurrent environment.

[0098] The data reassembly module performs splicing operations based on the task number and shard sequence number, and has the following capabilities:

[0099] 1. Out-of-order recovery: reorder the second data packets according to their fragment sequence numbers.

[0100] 2. Integrity check: Determine whether the reorganization is complete based on the preset number of shards or the task tail identifier.

[0101] 3. Partial Fault Tolerance: When the task allows, the system can be configured to tolerate the loss of non-critical segments. The task definition can use service tags or segment identification fields to mark the importance of segments, such as control packets and image frames (I-frames), to support flexible data fault tolerance strategies.

[0102] The second protocol server 122 is configured to perform a decapsulation operation based on the TCP and IP protocols on the third data packet, strip off the transport layer and network layer headers, obtain the original application layer data, and send the data to the second virtual network device 121 .

[0103] The second virtual network device 121 is used to temporarily store original application layer data and use it as a "quasi-application layer data flow" for the upper layer processing module of the system to call or store.

[0104] In the embodiment of the present invention, the virtual scheduling executor 113 may further include: a policy configuration and hot loading interface module.

[0105] The policy configuration and hot loading interface module is used to support the communication between the policy scheduling module and the external device, receive remote or local policy configuration, remotely load configuration files, realize policy hot plug and rollback capabilities, version switching and link status visualization. Specifically, the external device can be a computer or tablet computer and other smart devices with a display screen. The policy configuration and hot loading interface module can realize the policy scheduling module to dynamically load relevant policies from the local console or remote policy server, and send link status diagrams, task link distribution diagrams, policy hit statistics and other operation and maintenance visual information to the external device in real time, so that R&D personnel can edit the policies stored in the policy scheduling module through the policy configuration and hot loading interface module, as well as perform policy switching and parameter debugging, so that policy replacement and optimization can be completed during system operation without restarting the communication service or interrupting the business flow, thereby improving the system operation and maintenance efficiency and responsiveness, and is particularly suitable for high-availability network environments and distributed systems, facilitating system operation and maintenance and policy evolution deployment. In addition, the policy configuration and hot loading interface module can also realize the integration of virtual scheduling executors with external control systems, such as industrial control systems, application perception mechanisms, QoS control systems and AI model driving mechanisms, so that the present invention has high portability and intelligent expansion capabilities. Industrial control systems such as robot link scheduling and task flow control; application perception mechanisms dynamically allocate links according to business type or task priority; QoS control systems implement on-demand hierarchical scheduling of link resources; and AI model-driven mechanisms have the ability to learn historical behaviors and predict strategies.

[0106] The policy configuration and hot-loading interface module can be deployed on local edge nodes or central management nodes to achieve centralized control and flexible updates of the policy layer.

[0107] In the embodiment of the present invention, the receiving end component 12 is further configured to feed back the task completion status and link feedback result to the policy scheduling module after processing all second data packets.

[0108] The policy scheduling module is also used to optimize the corresponding policy according to the task completion status and link feedback results to achieve adaptive optimization of the policy.

[0109] Specifically, after the task is completed, the second data transmission server 123 will summarize the information such as link utilization status, transmission time, packet loss rate, whether to retransmit and whether to fail over into the first feedback information, and report it asynchronously to the policy scheduling module. In addition, after completing the reassembly of the second data packet, the data reassembly module will also generate second feedback data. The second feedback data is the reassembly status feedback, such as whether the reassembly is successful or failed. In addition, the second feedback data may also include the missing of key segments. After generating the second feedback data, the data reassembly module reports the second feedback data to the policy scheduling module. The task completion status and link feedback results include the first feedback data and the second feedback data. The policy scheduling module can write the task completion status and link feedback results into the learning sample library to update the corresponding policy weights and realize policy evolution and adaptive scheduling optimization.

[0110] In the embodiments of the present invention, the virtual scheduling executor has different deployment structures on different computing platforms. For example, the deployment structure in cloud services and data centers is centralized, while the deployment structure on edge nodes is distributed. This demonstrates that the system can flexibly operate in various network environments and architectural levels, supports centralized or distributed deployment, and can work in conjunction with multi-level scheduling systems.

[0111] It can be understood that the same or similar parts of the above embodiments can be referenced to each other, and the contents not described in detail in some embodiments can refer to the same or similar contents in other embodiments.

[0112] It should be noted that, in the description of the present invention, the terms "first", "second", etc. are used for descriptive purposes only and should not be understood as indicating or implying relative importance. In addition, in the description of the present invention, unless otherwise specified, the meaning of "plurality" is at least two.

[0113] It should be understood that various parts of the present invention can be implemented using hardware, software, firmware, or a combination thereof. In the above-described embodiments, multiple steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, any one or a combination of the following technologies known in the art can be used: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0114] Those skilled in the art will understand that all or part of the steps in the method of the above embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

[0115] In addition, the functional units in the various embodiments of the present invention may be integrated into a single processing module, or each unit may exist physically separately, or two or more units may be integrated into a single module. The aforementioned integrated modules may be implemented in the form of hardware or in the form of software functional modules. If the integrated modules are implemented in the form of software functional modules and sold or used as independent products, they may also be stored in a computer-readable storage medium.

[0116] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0117] Throughout this specification, reference to terms such as "one embodiment," "some embodiments," "examples," "specific examples," or "some examples" means that a specific feature, structure, material, or characteristic described in conjunction with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.

[0118] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A data communication scheduling system based on a virtual scheduling executor, characterized in that: include: A sending end component and a receiving end component; the sending end component includes a first virtual network device, a first protocol server, a virtual scheduling executor and a physical link resource pool; the virtual scheduling executor includes a policy scheduling module, a scheduling execution module, a link state sampling and analysis module and a first data transmission server; The first virtual network device is configured to, upon receiving original data sent by the application layer, encapsulate the original data into a network layer-like message and send the message to the first protocol server; The first protocol server is used to encapsulate the network layer-like message based on the TCP and IP protocols, and perform fragmentation according to the MTU rule to obtain a transmittable first data packet, and send it to the first data transmission server; The first data transmission server is configured to encapsulate the first data packet, add system-level scheduling header information to the first data packet, obtain a second data packet, and store the second data packet in the task scheduling queue managed by the first data transmission server; the system-level scheduling header information includes a recommended physical link ID, a task identifier, a slice sequence number, a slice flag, and task tag information; the first data transmission server is further configured to send a scheduling request to the policy scheduling module, the scheduling request including context parameters of the current task; When receiving the scheduling request, the policy scheduling module generates a candidate physical link set based on the scheduling request by calling the link state sampling and analysis module, and sends the candidate physical link set to the scheduling execution module; The scheduling execution module generates a first scheduling instruction according to the candidate physical link set, and sends the first scheduling instruction to the first data transmission server; The first data transmission server is further configured to complete the binding between the second data packet and the physical link and actually send it according to the first scheduling instruction, so that the receiving end component receives the second data packet; The physical link resource pool includes a resource set of multiple physical links and their state management modules, and provides link state information of the physical links for access by the link state sampling and analysis module.

2. The data communication scheduling system based on the virtual scheduling executor according to claim 1, characterized in that: The task tag information is a priority tag or a service type code; The priority tag is used to identify the transmission priority of the second data packet; The service type code is used to identify the quality of service requirement of the second data packet.

3. The data communication scheduling system based on the virtual scheduling executor according to claim 1, characterized in that: The context parameters of the current task include task identifier, target address, data length, service tag, priority level, and whether fragmented transmission is allowed.

4. The data communication scheduling system based on the virtual scheduling executor according to claim 1, characterized in that: The link state sampling and analysis module includes a link state perception unit, a link state cache unit, a multi-physical link optimization unit and an optimization perception distribution unit; When receiving the scheduling request, the policy scheduling module generates a candidate physical link set based on the scheduling request by calling the link status sampling and analysis module, specifically including: When receiving the scheduling request, the policy scheduling module obtains the first target link state information of each physical link in the physical link resource pool from the link state cache unit, and selects a target policy from a plurality of preset policies based on the context parameters of the current task and the first target link state information; wherein the link state sensing unit is used to collect the link state information of each physical link in the physical link resource pool in real time, and the link state cache unit is used to perform short-term caching and preprocessing on the link state information collected by the link state sensing unit; The policy scheduling module calls the multi-physical link optimization unit to calculate the scores of all the physical links based on the target policy; the multi-physical link optimization unit is used to calculate the scores of all the physical links according to the target policy and the first target link state information; The policy scheduling module calls the optimization-aware distribution unit to generate the candidate physical link set; the optimization-aware distribution unit is used to screen all the physical links according to the scores of all the physical links to obtain the candidate physical link set.

5. The data communication scheduling system based on the virtual scheduling executor according to claim 4 is characterized in that: The strategies include: Minimum delay priority strategy, used to prioritize the physical link with the lowest link delay for data transmission; The maximum bandwidth margin priority strategy is used to prioritize physical links with more surplus bandwidth resources for data transmission; The service label and physical link QoS label adaptation strategy is used to match the service type label of the task with the QoS label of the physical link, and schedule the corresponding physical link for data transmission based on the matching result; The historical scheduling success rate weighted physical link strategy is used to weight each physical link based on the success rate of historical scheduling tasks, and schedule the corresponding physical link for data transmission according to the weighted score result.

6. The data communication scheduling system based on the virtual scheduling executor according to claim 5, characterized in that: The strategy also includes: a physical link trend pre-selection strategy based on an AI prediction model.

7. The data communication scheduling system based on the virtual scheduling executor according to claim 4 is characterized in that: The link status sampling and analysis module further includes: an abnormality scheduling unit; The first data transmission server is used to return task confirmation information to the scheduling execution module after completing the sending of all the second data packets of the entire task; The scheduling execution module comprehensively detects whether a network anomaly meeting preset conditions occurs during the task transmission process based on the task confirmation information and the link status information collected in real time by the link status perception unit, and forms a detection result; When the detection result shows that a network abnormality meeting a preset condition occurs during the task transmission process, the scheduling execution module sends a rerouting instruction including the abnormality type and priority to the abnormality scheduling unit; When receiving the rerouting instruction, the abnormal scheduling unit selects the best replacement link from the currently available physical links in combination with the latest link performance score and historical scheduling success rate of each physical link provided by the link status cache unit, and feeds back the optimal replacement link to the scheduling execution module; The scheduling execution module regenerates a second scheduling instruction based on the optimal alternative link and sends it to the first data transmission server, while synchronizing the task status and data packet sequence information to ensure the transmission order and integrity of the data; After receiving the second scheduling instruction, the first data transmission server switches to the optimal alternative link according to the second scheduling instruction to continue transmitting the remaining data packets; When the detection result shows that a network anomaly that meets preset conditions occurs during the task transmission process, the scheduling execution module dynamically adjusts its own parameters based on the task confirmation information and the link status information collected in real time by the link status perception unit.

8. The data communication scheduling system based on the virtual scheduling executor according to claim 1 is characterized in that: The receiving end components include: a second data transmission server, a second protocol server and a second virtual network device; The second data transmission server is configured to, upon receiving the second data packets from multiple physical links, perform fragment classification, out-of-order recovery, integrity verification, and data splicing on the second data packets, and send the spliced third data packets to the second protocol server; The second protocol server is used to perform a decapsulation operation based on the TCP and IP protocols on the third data packet to obtain original application layer data and send the original application layer data to the second virtual network device; The second virtual network device is used to temporarily store the original application layer data.

9. The data communication scheduling system based on the virtual scheduling executor according to claim 1, characterized in that: The virtual scheduling executor also includes: a policy configuration and hot loading interface module; The policy configuration and hot loading interface module is used to support the policy scheduling module to communicate with external devices, receive remote or local policy configurations, and implement policy hot plugging, version switching, and link status visualization.

10. The data communication scheduling system based on the virtual scheduling executor according to claim 1, characterized in that: The receiving end component is further configured to feed back the task completion status and link feedback result to the policy scheduling module after processing all the second data packets; The policy scheduling module is further configured to optimize the corresponding policy according to the task completion status and link feedback results.

Images