Network flow intelligent scheduling control method and system

Abstract

The invention relates to the technical field of network communication, and particularly discloses an intelligent scheduling control method and system for network traffic, and the method comprises the steps: obtaining first-class network state information and second-class network state information of each NDN node, the first-class network state information comprises a node topology connection state, data throughput and cache utility information, and the second-class network state information comprises a node topology connection state, a data throughput and cache utility information; the second type of network state information comprises node link transmission delay fluctuation, port buffer queue depth and a message loss rate instantaneous value; and acquiring a real-time request intensity change rate according to the node topology connection state, the data throughput and the cache utility information. Closed-loop control is realized by acquiring the network state information, calculating the scheduling parameter, issuing the scheduling parameter to the forwarding node and updating the scheduling parameter based on feedback, and the method has the advantages of being capable of sensing the network state change in real time, realizing distributed adaptive adjustment and effectively improving the network flow scheduling efficiency and the resource utilization rate.

Description

Technical Field

[0001] This invention relates to the field of network communication technology, and in particular to a method and system for intelligent scheduling and control of network traffic. Background Technology

[0002] In the field of network communication, especially with the rapid development of the Internet of Things, high-definition video streaming, cloud computing and the industrial Internet, the scale, complexity and dynamism of network traffic are growing exponentially.

[0003] Existing technical solutions generally employ a centralized controller architecture, generating scheduling strategies by periodically collecting network statistics. This approach has significant drawbacks when dealing with rapid fluctuations in network conditions. Decision-making and execution delays exist between the centralized controller and forwarding nodes, causing scheduling strategies to fail to synchronously reflect instantaneous network changes, such as sudden traffic congestion or rapid migration of data object hotspots. Furthermore, forwarding nodes can only passively execute pre-set strategies, lacking the ability to adaptively fine-tune based on local real-time conditions. This mechanism disconnects the global optimization objective from the distributed execution environment, resulting in link utilization imbalances, data throughput fluctuations, and increased control signaling overhead. Summary of the Invention

[0004] The purpose of this invention is to provide a network traffic intelligent scheduling and control method and system to solve the technical problems mentioned in the background.

[0005] To achieve the above objectives, the present invention provides the following technical solution: A network traffic intelligent scheduling and control method, comprising: Obtain first-class network status information and second-class network status information for each NDN node. The first-class network status information includes node topology connection status, data throughput and buffer utility information. The second-class network status information includes node link transmission latency fluctuation, port buffer queue depth and instantaneous packet loss rate. The real-time request density change rate is obtained based on the node topology connection status, data throughput, and cache utility information, and the instantaneous popularity of node data objects is obtained based on the real-time request density change rate. The dynamic quality score of each node's transmission channel is obtained based on the node link transmission delay fluctuation, port buffer queue depth, and instantaneous packet loss rate. The scheduling parameter set for node data object traffic is obtained based on the dynamic quality score and the instantaneous popularity. The set of scheduling parameters is distributed to the forwarding nodes related to the traffic of the node data object. The scheduling parameter set is obtained based on the forwarding node, and the actual load ratio adjustment value of each node is obtained according to the scheduling parameter set. Based on the actual load ratio adjustment value, obtain the actual load ratio statistics and decision basis information within a preset time period, and obtain feedback data based on the actual load ratio statistics and decision basis information; The updated set of scheduling parameters is obtained based on the feedback data and used for traffic scheduling in the next time period.

[0006] Preferably, the steps of obtaining the real-time request density change rate based on the node topology connection status, data throughput, and cache utility information, and obtaining the instantaneous popularity of node data objects based on the real-time request density change rate, include: Based on the node topology connection status, obtain the node connection activity, which represents the connection activity between nodes; Based on the data throughput, obtain the data flow rate feature that characterizes the data flow rate per unit time. Based on the cache utility information, obtain the cache reuse performance that represents the probability of cached content being requested repeatedly; Based on the node connection activity, data flow rate characteristics, and cache reuse efficiency, feature fusion is performed to obtain a comprehensive load characterization of the nodes. Based on the temporal changes in the comprehensive load characterization of the nodes within a preset time window, obtain the trajectory of request intensity changes within a unit time window; Based on the request intensity change trajectory and combined with historical access records, the node data object is dynamically mapped to obtain the dynamic popularity mapping of the node data object, which is the instantaneous popularity of the node data object.

[0007] Preferably, the step of obtaining the dynamic quality score of each node's transmission channel based on the node link transmission delay fluctuation, port buffer queue depth, and instantaneous packet loss rate includes: Based on the transmission delay fluctuation of the node link, obtain the delay deviation degree, which characterizes the degree of deviation of the transmission delay from the baseline, and obtain the intensity of delay change based on the change sequence of the delay deviation degree within a preset time window; Based on the port buffer queue depth, an instantaneous buffer saturation value representing the port buffer space occupancy is obtained, and based on the peak characteristics of the instantaneous buffer saturation value within a preset time window, the port congestion accumulation status is obtained. Based on the instantaneous value of the message loss rate, obtain the instantaneous signal loss degree that characterizes the reliability of message transmission; Multidimensional feature normalization is performed on the intensity of the time delay change, the cumulative state of port congestion, and the instantaneous signal loss to obtain a performance fusion vector that reflects the current comprehensive performance of the channel. Obtain the set of performance fusion vectors of the channels in the historical period, and extract the steady-state baseline features of the channel performance from them; The performance fusion vector is compared with the steady-state baseline feature to obtain a dynamic offset that characterizes the degree of deviation of the channel performance from its historical steady-state level; Based on the dynamic offset and the performance fusion vector, the dynamic quality score of the node transmission channel is obtained through a weighted fusion operation.

[0008] Preferably, the step of obtaining the scheduling parameter set of node data object traffic based on the dynamic quality score and the instantaneous popularity includes: Based on the dynamic quality score, obtain the channel selection weight of each node's transmission channel; Based on the instantaneous popularity, obtain the popularity adaptation coefficient of the node data object; The channel selection weights and the heat matching coefficients are collaboratively mapped to generate an initial candidate route sequence for the traffic of the node data object; Based on the dynamic quality score of each channel in the initial candidate routing sequence and the preset channel capacity benchmark, the initial load allocation benchmark of each channel is obtained. The initial load allocation benchmark is dynamically corrected based on the heat adaptation coefficient to obtain the benchmark load ratio of each channel; Based on the temporal variation characteristics of the dynamic quality score and the prediction trend of the heat adaptation coefficient, the optimization target of the node data object traffic is set. Based on the initial candidate routing sequence, the baseline load ratio, and the optimization objective, the autonomous adjustment range of the nodes is comprehensively determined. The initial candidate routing sequence, the baseline load ratio, the optimization target, and the node's autonomous adjustment range are integrated into a set of scheduling parameters for the node data object traffic.

[0009] Preferably, the step of distributing the scheduling parameter set to the forwarding nodes related to the traffic of the node data object includes: The estimated path coverage of node data object traffic is obtained based on the initial candidate route sequence; Based on the estimated path coverage and in conjunction with the node topology connection status, obtain the set of target forwarding nodes related to the traffic of the node data object; Based on the node identifier and node performance attributes of each forwarding node in the target forwarding node set, the scheduling parameter set is adapted and format encapsulated to generate a node-specific policy delivery data packet. Based on the current network control channel load status and the priority identifier of the scheduling parameter set, obtain the optimal execution window for policy issuance; Within the optimal execution window, the node-specific policy distribution data packets are distributed to each forwarding node in the target forwarding node set via a reliable transmission protocol.

[0010] Preferably, the step of obtaining a set of scheduling parameters based on the forwarding nodes and obtaining the actual load ratio adjustment value of each node according to the set of scheduling parameters includes: The received set of scheduling parameters is parsed to obtain the parsed data set corresponding to the initial candidate route sequence, baseline load ratio, optimization target, and node autonomous adjustment range; Obtain real-time status monitoring data locally on the forwarding node, and obtain the current link quality metric and current buffer pressure index associated with each candidate route in the initial candidate route sequence; Based on the optimization objective, the current link quality metric, and the current buffer pressure index, a route adaptability evaluation is performed on each candidate route to obtain the current adaptability evaluation value of each candidate route. Based on the parsed data set and the current adaptability evaluation value of each candidate route, load adjustment decision calculation is performed within the autonomous adjustment range of the node to obtain a load ratio adjustment vector that satisfies the optimization objective; The adjustment range and continuity constraints of the load ratio adjustment vector are verified to generate the actual load ratio adjustment value for each node.

[0011] Preferably, the step of obtaining statistical information on the actual load ratio and decision-making basis information within a preset time period based on the actual load ratio adjustment value, and obtaining feedback data based on the statistical information on the actual load ratio and decision-making basis information, includes: Within a preset time period, continuously record the execution status of traffic scheduling by each node based on the actual load ratio adjustment value, and generate an adjustment trajectory sequence containing time sequence, node identifier, and load ratio value; Multi-dimensional time series analysis is performed on the adjustment trajectory sequence to extract the distribution concentration, change volatility and trend direction of the load ratio values ​​of each node within a preset time period, and to obtain the statistical profile characteristics of the actual load ratio. Based on the current link quality metric, current buffer pressure index, and adaptive evaluation criteria on which each actual load ratio adjustment value is generated, the decision triggering conditions and logical relationships of each adjustment are traced and recorded in a structured manner to generate a decision logic graph. The statistical profile features of the actual load ratio are correlated and fused with the decision logic graph to generate structured feedback metadata. The feedback data is generated based on the structured feedback metadata, with the addition of a preset time period identifier, a version identifier of the scheduling parameter set, and the overall optimization target.

[0012] The present invention also provides a network traffic intelligent scheduling and control system, comprising: The network status acquisition module is used to acquire first-type network status information and second-type network status information for each NDN node. The first-type network status information includes node topology connection status, data throughput and cache utility information, and the second-type network status information includes node link transmission delay fluctuation, port buffer queue depth and instantaneous packet loss rate. The instantaneous popularity acquisition module is used to acquire the real-time request density change rate based on the node topology connection status, data throughput and cache utility information, and to acquire the instantaneous popularity of the node data object based on the real-time request density change rate. The dynamic quality score acquisition module is used to obtain the dynamic quality score of each node's transmission channel based on the node link transmission delay fluctuation, port buffer queue depth, and instantaneous packet loss rate. The scheduling parameter set acquisition module is used to acquire the scheduling parameter set of node data object traffic based on the dynamic quality score and the instantaneous popularity. The forwarding node acquisition module is used to send the scheduling parameter set to the forwarding nodes related to the traffic of the node data object; The actual load ratio adjustment value acquisition module is used to acquire a set of scheduling parameters based on the forwarding nodes, and to acquire the actual load ratio adjustment value of each node according to the set of scheduling parameters. The feedback data acquisition module is used to acquire statistical information on the actual load ratio and decision-making basis information within a preset time period based on the actual load ratio adjustment value, and to acquire feedback data based on the statistical information on the actual load ratio and decision-making basis information. The traffic scheduling acquisition module is used to acquire an updated set of scheduling parameters based on the feedback data for traffic scheduling in the next time period.

[0013] The present invention also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps of an intelligent scheduling method for PCBA assembly processes.

[0014] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the steps of an intelligent scheduling method for PCBA assembly processes. The beneficial effects of this application are as follows: This invention achieves closed-loop control by acquiring network status information, calculating scheduling parameters, sending them to forwarding nodes, and updating scheduling parameters based on feedback. It has the advantages of being able to perceive network status changes in real time and achieve distributed adaptive adjustment, effectively improving the efficiency of network traffic scheduling and resource utilization. Attached Figure Description

[0015] Figure 1 This is a schematic diagram of a method flow according to an embodiment of this application.

[0016] Figure 2 This is a schematic diagram of the system structure according to an embodiment of this application.

[0017] Figure 3 This is a schematic diagram of the internal structure of a computer device according to an embodiment of this application.

[0018] The realization of the purpose, functional features and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0019] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0020] like Figure 1 As shown, this application provides a network traffic intelligent scheduling and control method, applied to a network with multiple NDN nodes, including: S1. Obtain the first type of network status information and the second type of network status information for each NDN node, wherein the first type of network status information includes node topology connection status, data throughput and cache utility information, and the second type of network status information includes node link transmission delay fluctuation, port buffer queue depth and instantaneous packet loss rate. S2. Obtain the real-time request density change rate based on the node topology connection status, data throughput, and cache utility information, and obtain the instantaneous popularity of node data objects based on the real-time request density change rate. S3. Obtain the dynamic quality score of each node's transmission channel based on the node link transmission delay fluctuation, port buffer queue depth, and instantaneous packet loss rate. S4. Obtain the set of scheduling parameters for node data object traffic based on the dynamic quality score and the instantaneous popularity. S5. The set of scheduling parameters is sent to the forwarding nodes related to the traffic of the node data object; S6. Obtain a set of scheduling parameters based on the forwarding nodes, and obtain the actual load ratio adjustment value of each node according to the set of scheduling parameters; S7. Obtain statistical information on the actual load ratio and decision basis information within a preset time period based on the actual load ratio adjustment value, and obtain feedback data based on the statistical information on the actual load ratio and decision basis information; S8. Obtain the updated set of scheduling parameters based on the feedback data for traffic scheduling in the next time period.

[0021] As described in steps S1-S8 above, this invention synchronously collects two types of network status information for each NDN node by deploying a monitoring agent or invoking a management protocol: the first type includes node topology connection status, data throughput, and cache utility information, used to characterize the macro-level network situation and content storage efficiency; the second type includes node link transmission latency fluctuations, port buffer queue depth, and instantaneous packet loss rate values, used to reflect the real-time micro-quality of the transmission channel. Based on the first type of information, the system dynamically calculates the real-time request density change rate through weighted fusion and short-time window analysis, thereby deducing the instantaneous popularity of node data objects to accurately capture the formation and migration of content hotspots. Simultaneously, using the second type of information, a dynamic quality score for each node's transmission channel is generated through comprehensive weighting or grading of multiple indicators such as latency, queue size, and packet loss, to assess the link health in real time. Subsequently, the dynamic quality score is combined with the instantaneous popularity, and through collaborative mapping and dynamic correction, a set of scheduling parameters is generated, including candidate routing sequences, baseline load ratios, optimization objectives, and the node's autonomous adjustment range. This set of data is precisely distributed to relevant forwarding nodes via the control plane. Upon receiving the data, the nodes not only execute the distributed policy framework but also adaptively fine-tune within permissible adjustment ranges based on real-time monitoring of link quality and buffer pressure, generating actual load balancing adjustment values. This enables a distributed, agile response under centralized guidance. Nodes continuously record actual load balancing statistics and the decision-making basis for each adjustment within a preset time period, integrating this data into structured feedback data for reporting. The control center evaluates the policy execution effectiveness and network status changes based on the feedback data, driving iterative updates to the scheduling parameter set. This forms a complete intelligent scheduling loop from global awareness, dynamic decision-making, elastic execution to closed-loop optimization, significantly improving the accuracy of traffic scheduling, resource utilization, and overall stability of the NDN network in dynamic environments.

[0022] In one embodiment, the steps of obtaining the real-time request density change rate based on the node topology connection status, data throughput, and cache utility information, and obtaining the instantaneous popularity of node data objects based on the real-time request density change rate, include: S201. Based on the node topology connection status, obtain the node connection activity, which represents the connection activity between nodes; S202. Based on the data throughput, obtain the data flow rate characteristics that characterize the data flow rate per unit time. S203. Based on the cache utility information, obtain the cache reuse performance that represents the probability of cached content being requested repeatedly; S204. Based on the node connection activity, data flow rate characteristics and cache reuse efficiency, feature fusion is performed to obtain a comprehensive load characterization of the node. S205. Based on the temporal changes of the node's comprehensive load characterization within the preset time window, obtain the trajectory of request intensity changes within a unit time window. S206. Based on the request intensity change trajectory and combined with historical access records, perform dynamic heat mapping on the node data object to obtain the dynamic heat mapping of the node data object, which is the instantaneous popularity of the node data object.

[0023] As described in steps S201-S206 above, this invention accurately acquires the instantaneous popularity of node data objects through a multi-dimensional real-time perception and dynamic mapping mechanism, overcoming the lag of traditional solutions in dynamic networks. Its core lies in extracting node connection activity (quantifying connection dynamics by monitoring packet exchange frequency or route update stability), data flow rate characteristics (reflecting request intensity by real-time monitoring of interface throughput), and cache reuse efficiency (assessing the probability of repeated content requests by statistically analyzing cache hit rate or content access patterns) from node topology connection status, data throughput, and cache utility information. Subsequently, these heterogeneous features are fused using a weighted average or machine learning model to generate a comprehensive node load characterization that fully reflects the node load status. The system continuously monitors the temporal changes of this characterization within a preset or adaptive short time window, accurately capturing the trajectory of request intensity changes to track the dynamic fluctuation trend of network load in real time. Finally, by combining this real-time trajectory with a dataset storing historical access records, a dynamic popularity mapping is performed using a dynamic scoring function or prediction model. This combines the current network dynamics of data objects with long-term access patterns, outputting a dynamic popularity mapping quantified as a numerical value or classification label, which serves as the instantaneous popularity basis for scheduling. This method, by integrating real-time micro-indicators with historical macro-patterns, achieves agile and accurate perception of the formation and migration of content hotspots, significantly improving the response speed and decision-making accuracy of traffic scheduling strategies in dynamic environments.

[0024] By acquiring node connection activity, data flow rate characteristics, and cache reuse efficiency from multiple dimensions and performing feature fusion, this solution can comprehensively and in real-time characterize the overall load status of NDN nodes, overcoming the limitations of single indicators. Based on this, by analyzing the temporal changes in the overall node load characterization, it can accurately capture the trajectory of request intensity changes within a unit time window, thereby tracking the dynamic evolution trend of request intensity in real time. Finally, by combining these real-time trajectories with historical access records for dynamic popularity mapping, this solution can accurately calculate the instantaneous popularity of node data objects, providing a highly dynamic and reliable basis for subsequent traffic scheduling.

[0025] In one embodiment, the step of obtaining the dynamic quality score of each node's transmission channel based on the node link transmission delay fluctuation, port buffer queue depth, and instantaneous packet loss rate includes: S301. Based on the transmission delay fluctuation of the node link, obtain the delay deviation degree, which characterizes the degree of deviation of the transmission delay from the baseline, and obtain the intensity of delay change based on the change sequence of the delay deviation degree within a preset time window. S302. Based on the port buffer queue depth, obtain the instantaneous buffer saturation that characterizes the port buffer space occupancy, and based on the peak characteristics of the instantaneous buffer saturation within a preset time window, obtain the port congestion accumulation status. S303. Based on the instantaneous value of the message loss rate, obtain the instantaneous signal loss degree that characterizes the reliability of message transmission; S304. Perform multi-dimensional feature normalization on the intensity of the time delay change, the cumulative state of port congestion and the instantaneous signal loss to obtain a performance fusion vector that reflects the current comprehensive performance of the channel. S305. Obtain the set of performance fusion vectors of the channels in the historical period, and extract the steady-state baseline features of the channel performance from them; S306. Compare the performance fusion vector with the steady-state baseline feature to obtain a dynamic offset that characterizes the degree of deviation of the channel performance from its historical steady-state level. S307. Based on the dynamic offset and the performance fusion vector, obtain the dynamic quality score of the node transmission channel through a weighted fusion operation.

[0026] As described in steps S301-S307 above, this invention achieves accurate dynamic quality assessment through multi-dimensional real-time monitoring of the transmission channel and comparison with historical baselines. Specifically, it collects three core indicators in real time: latency, buffering, and packet loss. The latency deviation is calculated based on the difference between the current latency and the baseline value, and its fluctuations are analyzed within a preset time window to obtain the intensity of latency changes reflecting stability. Instantaneous buffer saturation is obtained by reading the port buffer utilization rate, and its peak value and duration within the time window are further analyzed to form a port congestion accumulation trend reflecting the degree and development trend of congestion. Simultaneously, the instantaneous packet loss rate is obtained by statistically analyzing the packet loss ratio within a very short time and mapped to an instantaneous signal loss degree characterizing the degree of link reliability degradation. Subsequently, these three indicators with different dimensions are standardized or normalized to eliminate the influence of dimensions and then combined into a performance fusion vector to comprehensively quantify the real-time performance status of the channel. To provide an evaluation benchmark, the system extracts steady-state baseline features representing the healthy operating state of the channel from performance data of historical periods (e.g., 24 hours). By calculating the dynamic offset between the current performance fusion vector and this baseline feature, the deviation of the channel from its historical normal level is quantified. Finally, the dynamic offset and the performance fusion vector are weighted and fused (e.g., by linear summation with different weights or by model calculation) to generate a comprehensive dynamic quality score. This score not only integrates the real-time impact of latency fluctuations, congestion accumulation, and packet loss, but also highlights abnormal deviations by comparing with historical steady-state data. Therefore, it can more comprehensively, robustly, and timely reflect the true quality of the transmission channel, providing accurate and reliable decision-making basis for upper-layer traffic scheduling, and significantly improving adaptability and scheduling accuracy in dynamic network environments.

[0027] In one embodiment, the step of obtaining the scheduling parameter set of node data object traffic based on the dynamic quality score and the instantaneous popularity includes: S401. Based on the dynamic quality score, obtain the channel selection weight of each node's transmission channel; S402. Based on the instantaneous popularity, obtain the popularity adaptation coefficient of the node data object; S403. Perform a collaborative mapping between the channel selection weight and the heat adaptation coefficient to generate an initial candidate route sequence for the traffic of the node data object; S404. Based on the dynamic quality score of each channel in the initial candidate routing sequence and the preset channel capacity benchmark, obtain the initial load allocation benchmark for each channel. S405. Dynamically correct the initial load allocation benchmark according to the heat adaptation coefficient to obtain the benchmark load ratio of each channel; S406. Based on the temporal variation characteristics of the dynamic quality score and the predicted trend of the heat adaptation coefficient, set the optimization target for the node data object traffic. S407. Based on the initial candidate route sequence, the baseline load ratio, and the optimization objective, comprehensively determine the node's autonomous adjustment range; S408. The initial candidate routing sequence, the baseline load ratio, the optimization target, and the node autonomous adjustment range are integrated into a set of scheduling parameters for the node data object traffic.

[0028] As described in steps S401-S408 above, this invention transforms the two core dynamic indicators obtained from the aforementioned calculations—the dynamic quality score reflecting the real-time health status of the channel and the instantaneous popularity reflecting the instantaneous demand for content—into quantitative parameters that can be directly used for decision-making. The dynamic quality score is generated through fuzzy logic or machine learning models, and the channel selection weight is obtained based on this score through normalization or classification mapping, which quantifies the "performance priority" of each channel. Simultaneously, the instantaneous popularity is obtained through real-time statistics or time decay algorithms, and the popularity adaptation coefficient is determined accordingly, quantifying the "scheduling urgency" of the data object. Secondly, the significance of step S402 (cooperative mapping to generate initial candidate route sequences) lies in fusing the two different dimensions of "channel performance priority" and "data scheduling urgency" into a unified decision. Through multi-objective optimization algorithms or rule-based expert systems, this step outputs a set of initial candidate route sequences. Its beneficial effect is that it realizes the initial route screening of "quality-driven" and "popularity-aware", ensuring that the selected path set can theoretically utilize high-quality channels and effectively serve popular data streams, avoiding the one-sidedness of route selection based on only a single dimension (such as the shortest path).

[0029] Steps S403 and S404 constitute a two-step method for load allocation. S403 obtains a preliminary load allocation benchmark based on capacity and quality, using the dynamic quality score of each channel in the candidate routes and a preset channel capacity benchmark (such as physical bandwidth or historical recommended values). S404 then introduces a popularity adaptation coefficient to dynamically adjust this benchmark, for example, by using a multiplication factor to amplify the allocation ratio of popular data on high-quality channels. The initial benchmark ensures that the allocation conforms to network physical constraints and basic quality principles; while the popularity-based dynamic adjustment allows the allocation strategy to sensitively follow the instantaneous fluctuations in content popularity, preventing load imbalances caused by hotspot migration and significantly improving the real-time rationality and efficiency of resource allocation.

[0030] Step S405 provides a clear performance orientation for the entire scheduling process. Based on the temporal trend of dynamic quality scores (e.g., whether latency is deteriorating) and the prediction of heat adaptation coefficients (e.g., whether heat will continue), optimization objectives such as "minimizing latency" or "maximizing throughput" are dynamically set. Its beneficial effect is that it enables the scheduling strategy to possess "objective adaptability," flexibly adjusting the primary optimization direction according to the main network challenges at different times (whether it is severe congestion or sudden hotspots), rather than adhering to a single objective.

[0031] Step S406 is a key design element bridging centralized decision-making and distributed execution. Based on the initial routing sequence, baseline load ratio, and optimization objectives, it comprehensively determines the autonomous adjustment range of a node. It clarifies the guiding framework issued by the centralized controller while delegating real-time fine-tuning authority within certain limits to the forwarding nodes. This resolves the fundamental contradiction between the slow response of fully centralized control and the chaotic nature of fully distributed decision-making, providing institutional guarantees for achieving a hybrid intelligent model of "centralized guidance and distributed execution."

[0032] Step S407 encapsulates the outputs of all the above steps—initial candidate route sequence, baseline load ratio, optimization target, and node autonomous adjustment range—into a structured set of scheduling parameters.

[0033] In one embodiment, the step of sending the scheduling parameter set to the forwarding nodes related to the node data object traffic includes: S501. Obtain the estimated path coverage of node data object traffic based on the initial candidate routing sequence; S502. Based on the estimated path coverage and in conjunction with the node topology connection status, obtain the set of target forwarding nodes related to the traffic of the node data object; S503. Based on the node identifier and node performance attributes of each forwarding node in the target forwarding node set, perform parameter adaptation and format encapsulation on the scheduling parameter set to generate a node-specific policy delivery data packet. S504. Based on the current network control channel load status and the priority identifier of the scheduling parameter set, obtain the optimal execution window for policy issuance; S505. Within the optimal execution window, the node-specific policy distribution data packet is distributed to each forwarding node in the target forwarding node set via a reliable transmission protocol.

[0034] As described in steps S501-S505 above, obtaining the estimated path coverage of node data object traffic in this invention refers to identifying all nodes and links that the node data object traffic may pass through in the network based on the initial candidate route sequence contained in the scheduling parameter set. This can be achieved by analyzing the source node, destination node, and intermediate hop count in the route sequence, and combining it with path deduction using a network topology database. Obtaining the target forwarding node set related to the node data object traffic refers to further filtering out the nodes actually participating in the forwarding of the node data object traffic after determining the estimated path coverage and combining it with the current node topology connection status of the network. This can be achieved by comparing the nodes within the estimated path coverage with the current active topology connection status, eliminating those nodes that are currently offline or do not have forwarding capabilities, thereby obtaining an accurate and real-time target forwarding node set.

[0035] Parameter adaptation and format encapsulation of the scheduling parameter set to generate node-specific policy delivery data packets refers to converting the general scheduling parameter set into a specific instruction format that the node can understand and execute, based on its unique node identifier (such as device ID, IP address) and node performance attributes (such as hardware model, operating system, supported protocol version) for each node in the target forwarding node set. This can include converting abstract routing policies into specific routing table entries, load balancing into traffic shaping rules, or caching policies into cache management instructions. Obtaining the optimal execution window for policy delivery refers to dynamically selecting the most suitable time period for policy delivery based on the current network control channel load status and the priority identifier of the scheduling parameter set. The control channel load status can be evaluated by monitoring indicators such as control plane traffic, channel bandwidth utilization, and control packet queue depth. The priority identifier of the scheduling parameter set indicates the urgency of the policy. Distributing node-specific policy data packets to each forwarding node in the target forwarding node set via a reliable transmission protocol means ensuring that policy data packets arrive completely, accurately, and promptly at each target forwarding node within the optimal execution window, using protocols with error detection, retransmission, and acknowledgment mechanisms.

[0036] By accurately acquiring the estimated path coverage and target forwarding node set of node data object traffic, policies are avoided from being sent to irrelevant nodes, significantly reducing unnecessary control overhead and network redundancy. Secondly, parameter adaptation and format encapsulation are performed for each target forwarding node to generate node-specific policy delivery data packets, solving the problem of policy mismatch with node execution capabilities in traditional solutions and ensuring the executability and effectiveness of policies in a distributed environment. Furthermore, based on the current network control channel load status and the priority identifier of the scheduling parameter set, the optimal execution window is dynamically obtained, effectively avoiding control channel congestion and ensuring the timeliness of policy delivery, thus overcoming the problems caused by the lag in centralized decision-making. Finally, distribution is performed through a reliable transmission protocol, ensuring the integrity and reliability of policy data packets, preventing policy loss or failure, and further improving the stability of the overall scheduling process.

[0037] In one embodiment, the step of obtaining a set of scheduling parameters based on the forwarding nodes and obtaining the actual load ratio adjustment value of each node according to the set of scheduling parameters includes: S601. Perform policy data parsing on the received set of scheduling parameters to obtain the parsed data set corresponding to the initial candidate route sequence, baseline load ratio, optimization target, and node autonomous adjustment range; S602. Obtain the real-time status monitoring data of the forwarding node, and obtain the current link quality metric and current buffer pressure index associated with each candidate route in the initial candidate route sequence; S603. Based on the optimization objective, the current link quality metric, and the current buffer pressure index, perform a route adaptability evaluation on each candidate route and obtain the current adaptability evaluation value of each candidate route. S604. Based on the parsed data set and the current adaptability evaluation value of each candidate route, perform load adjustment decision calculation within the autonomous adjustment range of the node to obtain a load ratio adjustment vector that satisfies the optimization objective. S605. Verify the adjustment range and continuity constraints of the load ratio adjustment vector to generate the actual load ratio adjustment value for each node.

[0038] As described in steps S601-S605 above, this invention uses forwarding nodes to parse the received scheduling parameter set into policy data, thereby accurately understanding the global policy framework issued by the controller, including candidate paths, baseline allocation, optimization direction, and adjustment boundaries. This provides a unified basis for subsequent local decision-making and avoids execution deviations due to policy misunderstandings. Simultaneously, by acquiring real-time status monitoring data from the forwarding nodes and obtaining the current link quality metric and current buffer pressure index associated with each candidate route in the initial candidate route sequence, local real-time network indicators such as link quality and buffer pressure are directly collected. Nodes can perceive the immediate performance status of each candidate route, compensating for the deficiency of the centralized controller in not being able to obtain instantaneous status data in a timely manner. Based on this, and based on the optimization objective and the current link quality metric and current buffer pressure index, a route adaptability evaluation is performed on each candidate route, obtaining the current adaptability evaluation value of each candidate route. Combining the optimization objective and real-time metric data, the applicability of each route in the current environment is evaluated, its adaptability is quantified, and the evaluation results are ensured to be consistent with the global optimization direction. Furthermore, based on the baseline load ratio, the node's autonomous adjustment range, and the current adaptability evaluation value of each candidate route, load adjustment decision calculations are performed within the node's autonomous adjustment range to obtain a load ratio adjustment vector that meets the optimization objective. Within the global baseline and preset adjustment range, the adjustment vector is dynamically calculated based on the route adaptability value, ensuring that the decision respects the controller's intent while flexibly adapting to local changes, preventing load allocation from deviating from the optimization objective. Finally, the adjustment magnitude and continuity constraints of the load ratio adjustment vector are verified, generating the actual load ratio adjustment value for each node. This verifies the rationality and continuity of the adjustment vector's magnitude, avoiding drastic fluctuations that could cause network instability, and generating safely executable adjustment values ​​to ensure the smoothness and reliability of adaptive execution.

[0039] The received set of scheduling parameters undergoes policy data parsing, aiming to convert the set of scheduling parameters issued by the controller into internal instructions that the forwarding nodes can understand and execute. This can be accomplished, for example, by using a predefined protocol parser to extract fields and transform values ​​according to the data structure of the scheduling parameter set (such as JSON, XML, or binary format), or by loading a pre-compiled policy script. The initial candidate route sequence refers to a set of potential paths pre-determined by the controller that can be used for traffic forwarding of specific data objects, which can exist in the form of a list of route identifiers or a set of path descriptors. The baseline load balancing ratio is the initial traffic allocation ratio suggested by the controller for each candidate route based on the global network state and optimization objectives, which can be expressed as the traffic weight or percentage of each route. The optimization objective is the performance indicator expected to be achieved in this traffic scheduling, such as minimizing end-to-end latency, maximizing throughput, minimizing packet loss rate, or balancing network load, which can be encoded in the form of an enumeration type, objective function expression, or priority list. The node autonomous adjustment range limits the maximum deviation of the forwarding node when adjusting the load balancing locally to prevent over-adaptation from causing network instability, which can be expressed as a percentage or absolute value relative to the upper and lower limits of the baseline load balancing ratio. Local decision-making provides real-time and accurate network status information. Real-time status monitoring data can be periodically collected, for example, through network interface card (NIC) performance counters, network statistics interfaces provided by the operating system, or embedded sensors, to obtain data including, but not limited to, link bandwidth utilization, packet transmission / reception rates, and the number of erroneous packets. Current link quality metrics are key indicators for evaluating the performance of a specific link. These can be calculated using weighted averaging or fuzzy logic algorithms based on real-time monitored parameters such as link bandwidth, latency, jitter, and packet loss rate, or obtained through active probing. Then, the contribution of each candidate route to achieving the preset optimization goal under the current local network conditions is quantified. Route adaptability evaluation can be achieved, for example, by constructing a multi-factor evaluation model. This model takes the optimization goal (such as minimizing latency) as the core consideration and combines it with current link quality metrics (such as high scores for low-latency links) and current buffer pressure indices (such as high scores for low-buffer pressure) for weighted calculation to generate a comprehensive score. Alternatively, a machine learning-based predictive model can be used to predict the performance of each route under current conditions based on historical data and real-time status, and use this prediction as the adaptability evaluation value.

[0040] Determine how to optimally adjust traffic allocation based on the controller's recommended baseline and local real-time conditions. Load balancing decision calculations can be achieved, for example, by solving a constrained optimization problem where the objective function is to maximize the overall adaptive performance value or minimize a specific performance metric. Constraints include total traffic conservation, load balancing within the node's autonomous adjustment range for each route, and deviation limits from the baseline load balancing. The load balancing adjustment vector is validated for adjustment magnitude and continuity constraints to generate actual load balancing adjustment values ​​for each node, aiming to ensure the smoothness of local adaptive adjustments and network stability. Adjustment magnitude constraint validation can be, for example, by setting a maximum single adjustment percentage threshold and checking whether each component in the load balancing adjustment vector exceeds this threshold; if it does, it is truncated or reduced proportionally. Continuity constraint validation can be, for example, by comparing the direction and magnitude of the current adjustment with the previous adjustment; if two consecutive adjustments are found to be in opposite directions and with large magnitudes, the current adjustment is smoothed, such as by using an exponentially weighted moving average or a low-pass filter, to avoid frequent traffic oscillations between different routes.

[0041] In one embodiment, the step of obtaining statistical information on the actual load ratio and decision-making basis information within a preset time period based on the actual load ratio adjustment value, and obtaining feedback data based on the statistical information on the actual load ratio and decision-making basis information, includes: S701. Within a preset time period, continuously record the execution status of traffic scheduling by each node based on the actual load ratio adjustment value, and generate an adjustment trajectory sequence containing time sequence, node identifier, and load ratio value. S702. Perform multi-dimensional time series analysis on the adjustment trajectory sequence, extract the distribution concentration, change volatility and trend direction of the load ratio values ​​of each node within the preset time period, and obtain the statistical profile characteristics of the actual load ratio. S703. Based on the current link quality metric, current buffer pressure index and adaptive evaluation criteria on which each actual load ratio adjustment value is based, trace and record the decision triggering conditions and logical relationships for each adjustment in a structured manner, and generate a decision logic graph. S704. The statistical profile features of the actual load ratio are correlated and fused with the decision logic graph to generate structured feedback metadata. S705. Based on the structured feedback metadata, and with the addition of the identifier of the preset time period, the version identifier of the scheduling parameter set, and the overall optimization target, the feedback data is generated.

[0042] As described in steps S701-S705 above, the step of continuously recording the execution status of traffic scheduling by each node according to the actual load ratio adjustment value, and generating an adjustment trajectory sequence containing time sequence, node identifier, and load ratio value, aims to capture the actual traffic scheduling situation of each forwarding node in the network after receiving the scheduling instruction. Through continuous recording, a detailed time series data can be obtained, which includes the time point of each adjustment, the identity of the node performing the adjustment, and the specific load ratio value after the adjustment.

[0043] Distribution concentration reflects the main distribution range or preference of load allocation values ​​within a preset time period, for example, by calculating the mean, median, or mode; volatility quantifies the degree of change in load allocation values ​​over time, for example, by measuring standard deviation, variance, or mean absolute deviation; trend direction reveals whether the load allocation values ​​rise, fall, or remain stable throughout the period, for example, by using the slope of linear regression or a moving average. These statistical characteristics collectively constitute the statistical profile of the actual load allocation, providing a macro-level perspective for evaluating scheduling effectiveness and network status.

[0044] When a forwarding node makes local adaptive adjustments based on a set of scheduling parameters, its decisions are based on real-time link quality metrics, buffer pressure indices, and preset adaptive evaluation criteria. By tracing these input parameters and evaluation processes, a decision logic graph can be constructed, clearly showing the causal relationship of "under what network conditions, based on what evaluation criteria, what adjustments were made." For example, the forwarding node can record a snapshot of the input state (such as link latency, packet loss rate, queue depth, etc.) and the evaluation function used and its output results at each decision; or, by maintaining a decision log in the control plane, the local information and decision-making process used by the forwarding node when making local adaptive adjustments after the scheduling parameter set is issued can be recorded.

[0045] By integrating macroscopic statistical effects with microscopic decision-making logic, a more comprehensive and interpretable feedback data is formed. Statistical profile features provide a quantitative description of "what happened," while the decision logic graph explains "why it happened." Through information correlation, the statistical characteristics of load distribution within a specific time period can be matched with the decision logic that occurred during the same period, thereby understanding which specific decision logics drive a certain statistical trend. Fusion encoding unifies the representation of this heterogeneous information; for example, key-value pairs, JSON objects, or XML structures can be used to encapsulate this correlation information, making it easy for machines to parse and process.

[0046] Structured feedback metadata contains core statistical and decision-making information. The addition of a preset time period identifier clearly identifies the time window to which the feedback data corresponds, ensuring time-series accuracy. The version identifier of the scheduling parameter set indicates which version of the scheduling strategy the feedback data reflects, which is crucial for evaluating the merits of different strategies. The overall optimization objective provides a benchmark for evaluating the quality of the feedback data; for example, is the goal low latency, high throughput, or load balancing? This additional information makes the feedback data traceable, comparable, and goal-oriented. like Figure 2 As shown, the present invention also provides a network traffic intelligent scheduling and control system, comprising: The network status acquisition module is used to acquire first-type network status information and second-type network status information for each NDN node. The first-type network status information includes node topology connection status, data throughput and cache utility information, and the second-type network status information includes node link transmission delay fluctuation, port buffer queue depth and instantaneous packet loss rate. The instantaneous popularity acquisition module is used to acquire the real-time request density change rate based on the node topology connection status, data throughput and cache utility information, and to acquire the instantaneous popularity of the node data object based on the real-time request density change rate. The dynamic quality score acquisition module is used to obtain the dynamic quality score of each node's transmission channel based on the node link transmission delay fluctuation, port buffer queue depth, and instantaneous packet loss rate. The scheduling parameter set acquisition module is used to acquire the scheduling parameter set of node data object traffic based on the dynamic quality score and the instantaneous popularity. The forwarding node acquisition module is used to send the scheduling parameter set to the forwarding nodes related to the traffic of the node data object; The actual load ratio adjustment value acquisition module is used to acquire a set of scheduling parameters based on the forwarding nodes, and to acquire the actual load ratio adjustment value of each node according to the set of scheduling parameters. The feedback data acquisition module is used to acquire statistical information on the actual load ratio and decision-making basis information within a preset time period based on the actual load ratio adjustment value, and to acquire feedback data based on the statistical information on the actual load ratio and decision-making basis information. The traffic scheduling acquisition module is used to acquire an updated set of scheduling parameters based on the feedback data for traffic scheduling in the next time period.

[0047] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium. When executed, the computer program can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in this application and in the embodiments can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual-speed SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0048] It should be noted that, in this document, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, apparatus, article, or method that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, apparatus, article, or method. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, apparatus, article, or method that includes that element.

[0049] The above description is only a preferred embodiment of the present invention and does not limit the patent scope of the present invention. Any equivalent results or equivalent process transformations made based on the content of the present invention specification and drawings, or direct or indirect applications in other related technical fields, are similarly included within the patent protection scope of the present invention.

Claims

1. A network traffic intelligent scheduling control method, applied to a network of a plurality of NDN nodes, characterized in that, The method comprises the following steps: acquiring first type network state information and second type network state information of each NDN node, wherein the first type network state information comprises node topology connection state, data throughput and cache utility information, and the second type network state information comprises node link transmission delay fluctuation, port buffer queue depth and packet loss rate instantaneous value; acquiring real-time request intensity change rate according to the node topology connection state, data throughput and cache utility information, and acquiring instantaneous popularity of node data object according to the real-time request intensity change rate; acquiring dynamic quality score of each node transmission channel according to node link transmission delay fluctuation, port buffer queue depth and packet loss rate instantaneous value; acquiring a set of scheduling parameters of node data object flow according to the dynamic quality score and the instantaneous popularity; downloading the set of scheduling parameters to a forwarding node related to the node data object flow; acquiring a set of scheduling parameters based on the forwarding node, and acquiring actual load ratio adjustment value of each node according to the set of scheduling parameters; acquiring actual load ratio statistical information and decision basis information in a preset period according to the actual load ratio adjustment value, and acquiring feedback data based on the actual load ratio statistical information and the decision basis information; acquiring an updated set of scheduling parameters for flow scheduling in the next period according to the feedback data.

2. The network traffic intelligent scheduling control method of claim 1, wherein, The step of acquiring real-time request intensity change rate according to the node topology connection state, data throughput and cache utility information, and acquiring instantaneous popularity of node data object according to the real-time request intensity change rate comprises: acquiring node connection activity representing connection activity between nodes according to the node topology connection state; acquiring data flow speed feature representing data flow rate feature in unit time according to the data throughput; acquiring cache reuse efficiency representing probability of repeated request of cache content according to the cache utility information; performing feature fusion according to the node connection activity, data flow speed feature and cache reuse efficiency to acquire node comprehensive load representation; acquiring request intensity change trajectory in a unit time window according to time sequence change of the node comprehensive load representation in a preset time window; performing dynamic heat mapping of node data object according to the request intensity change trajectory and combining historical access records to acquire dynamic heat mapping of node data object, and the dynamic heat mapping is the instantaneous popularity of the node data object.

3. The network traffic intelligent scheduling control method of claim 1, wherein, The step of acquiring dynamic quality score of each node transmission channel according to node link transmission delay fluctuation, port buffer queue depth and packet loss rate instantaneous value comprises: acquiring delay deviation degree representing deviation degree of transmission delay from a reference according to the node link transmission delay fluctuation, and acquiring delay change intensity based on change sequence of the delay deviation degree in a preset time window; acquiring instantaneous buffer saturation representing port cache space occupation according to the port buffer queue depth, and acquiring port congestion accumulation situation based on peak value feature of the instantaneous buffer saturation in a preset time window; Based on the instantaneous value of the message loss rate, obtain the instantaneous signal loss degree that characterizes the reliability of message transmission; Multidimensional feature normalization is performed on the intensity of the time delay change, the cumulative state of port congestion, and the instantaneous signal loss to obtain a performance fusion vector that reflects the current comprehensive performance of the channel. Obtain the set of performance fusion vectors of the channels in the historical period, and extract the steady-state baseline features of the channel performance from them; The performance fusion vector is compared with the steady-state baseline feature to obtain a dynamic offset that characterizes the degree of deviation of the channel performance from its historical steady-state level; Based on the dynamic offset and the performance fusion vector, the dynamic quality score of the node transmission channel is obtained through a weighted fusion operation.

4. The network traffic intelligent scheduling control method of claim 1, wherein, The step of obtaining the scheduling parameter set for node data object traffic based on the dynamic quality score and the instantaneous popularity includes: Based on the dynamic quality score, obtain the channel selection weight of each node's transmission channel; Based on the instantaneous popularity, obtain the popularity adaptation coefficient of the node data object; The channel selection weights and the heat matching coefficients are collaboratively mapped to generate an initial candidate route sequence for the traffic of the node data object; Based on the dynamic quality score of each channel in the initial candidate routing sequence and the preset channel capacity benchmark, the initial load allocation benchmark of each channel is obtained. The initial load allocation benchmark is dynamically corrected based on the heat adaptation coefficient to obtain the benchmark load ratio of each channel; Based on the temporal variation characteristics of the dynamic quality score and the prediction trend of the heat adaptation coefficient, the optimization target of the node data object traffic is set. Based on the initial candidate routing sequence, the baseline load ratio, and the optimization objective, the autonomous adjustment range of the nodes is comprehensively determined. The initial candidate routing sequence, the baseline load ratio, the optimization target, and the node's autonomous adjustment range are integrated into a set of scheduling parameters for the node data object traffic.

5. The network traffic intelligent scheduling control method of claim 4, wherein, The step of distributing the scheduling parameter set to the forwarding nodes related to the traffic of the node data object includes: The estimated path coverage of node data object traffic is obtained based on the initial candidate route sequence; Based on the estimated path coverage and in conjunction with the node topology connection status, obtain the set of target forwarding nodes related to the traffic of the node data object; Based on the node identifier and node performance attributes of each forwarding node in the target forwarding node set, the scheduling parameter set is adapted and format encapsulated to generate a node-specific policy delivery data packet. Based on the current network control channel load status and the priority identifier of the scheduling parameter set, obtain the optimal execution window for policy issuance; Within the optimal execution window, the node-specific policy distribution data packets are distributed to each forwarding node in the target forwarding node set via a reliable transmission protocol.

6. The network traffic intelligent scheduling control method of claim 4, wherein, The steps of obtaining a set of scheduling parameters based on forwarding nodes and obtaining the actual load ratio adjustment value of each node according to the set of scheduling parameters include: The received set of scheduling parameters is parsed to obtain the parsed data set corresponding to the initial candidate route sequence, baseline load ratio, optimization target, and node autonomous adjustment range; Obtain real-time status monitoring data locally on the forwarding node, and obtain the current link quality metric and current buffer pressure index associated with each candidate route in the initial candidate route sequence; Based on the optimization objective, the current link quality metric, and the current buffer pressure index, a route adaptability evaluation is performed on each candidate route to obtain the current adaptability evaluation value of each candidate route. Based on the parsed data set and the current adaptability evaluation value of each candidate route, load adjustment decision calculation is performed within the autonomous adjustment range of the node to obtain a load ratio adjustment vector that satisfies the optimization objective; The adjustment range and continuity constraints of the load ratio adjustment vector are verified to generate the actual load ratio adjustment value for each node.

7. The network traffic intelligent scheduling control method of claim 1, wherein, The steps of obtaining statistical information on the actual load ratio and decision-making basis information within a preset time period based on the actual load ratio adjustment value, and obtaining feedback data based on the statistical information on the actual load ratio and decision-making basis information, include: Within a preset time period, continuously record the execution status of traffic scheduling by each node based on the actual load ratio adjustment value, and generate an adjustment trajectory sequence containing time sequence, node identifier, and load ratio value; Multi-dimensional time series analysis is performed on the adjustment trajectory sequence to extract the distribution concentration, change volatility and trend direction of the load ratio values ​​of each node within a preset time period, and to obtain the statistical profile characteristics of the actual load ratio. Based on the current link quality metric, current buffer pressure index, and adaptive evaluation criteria on which each actual load ratio adjustment value is generated, the decision triggering conditions and logical relationships of each adjustment are traced and recorded in a structured manner to generate a decision logic graph. The statistical profile features of the actual load ratio are correlated and fused with the decision logic graph to generate structured feedback metadata. The feedback data is generated based on the structured feedback metadata, with the addition of a preset time period identifier, a version identifier of the scheduling parameter set, and the overall optimization target.

8. A network traffic intelligent scheduling control system, characterized in that, include: The network status acquisition module is used to acquire first-type network status information and second-type network status information for each NDN node. The first-type network status information includes node topology connection status, data throughput and cache utility information, and the second-type network status information includes node link transmission delay fluctuation, port buffer queue depth and instantaneous packet loss rate. The instantaneous popularity acquisition module is used to acquire the real-time request density change rate based on the node topology connection status, data throughput and cache utility information, and to acquire the instantaneous popularity of the node data object based on the real-time request density change rate. The dynamic quality score acquisition module is used to obtain the dynamic quality score of each node's transmission channel based on the node link transmission delay fluctuation, port buffer queue depth, and instantaneous packet loss rate. The scheduling parameter set acquisition module is used to acquire the scheduling parameter set of node data object traffic based on the dynamic quality score and the instantaneous popularity. The forwarding node acquisition module is used to send the scheduling parameter set to the forwarding nodes related to the traffic of the node data object; The actual load ratio adjustment value acquisition module is used to acquire a set of scheduling parameters based on the forwarding nodes, and to acquire the actual load ratio adjustment value of each node according to the set of scheduling parameters. The feedback data acquisition module is used to acquire statistical information on the actual load ratio and decision-making basis information within a preset time period based on the actual load ratio adjustment value, and to acquire feedback data based on the statistical information on the actual load ratio and decision-making basis information. The traffic scheduling acquisition module is used to acquire an updated set of scheduling parameters based on the feedback data for traffic scheduling in the next time period. 9.A computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the computer device is configured to perform the method according to any one of claims 1-8 when the computer program is executed by the processor. When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium having stored thereon a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images