A communication processing method based on high-quality channel routing and scheduling

Abstract

The present invention relates to the technical field of communication processing, and discloses a communication processing method based on high-quality channel routing scheduling, including the following steps: Initialization and preparation: Set a group of input parameters, import a pre-trained machine learning model, and output an initialized parameter set. Traffic monitoring and warning: Collect real-time traffic statistics data from all network elements through the southbound interface, and then extract key features based on historical data and the collected real-time traffic statistics data, and define upper and lower limit thresholds within the normal traffic range. The present invention gives an optimal path recommendation for different application types and traffic patterns, especially an effective solution for burst traffic. Thanks to more intelligent traffic management and resource allocation, users enjoy a smoother and more reliable network service, and can maintain a good experience even during peak network traffic periods, avoiding the adverse effects caused by service interruptions due to congestion.

Description

Technical Field

[0001] The present invention relates to the field of communication processing, and more specifically, it relates to a communication processing method based on high-quality channel routing and scheduling. Background Art

[0002] In the current era of information explosion, with the rapid development of Internet technology, traditional network architectures are facing unprecedented challenges. On the one hand, the increasing data traffic places higher demands on network bandwidth; on the other hand, diverse application requirements also make network management complex and intractable. Traditional hardware-based network devices are difficult to adapt to the rapidly changing application environment, and their fixed functions and closed architectures limit network flexibility and service innovation capabilities.

[0003] In the context of the combination of software-defined networking (SDN) and network function virtualization (NFV), certain events (such as live broadcasts of large-scale events and reports of emergencies) can cause a surge in traffic during a specific period. Links on the conventional path may not be able to withstand the suddenly increased load, resulting in congestion or even service interruption. Summary of the Invention

[0004] The present invention provides a communication processing method based on high-quality channel routing and scheduling to solve the technical problems in the related art.

[0005] The present invention provides a communication processing method based on high-quality channel routing and scheduling, including the following steps:

[0006] S100, Initialization and Preparation: Set a group of input parameters, import a pre-trained machine learning model, and output an initialized parameter set;

[0007] S200, Traffic Monitoring and Warning: Collect real-time traffic statistical data from all network elements through the southbound interface, then extract key features based on historical data and the collected real-time traffic statistical data, and define the upper and lower threshold values within the normal traffic range;

[0008] S300, Traffic Surge Detection and Response: Once it is found that the link utilization rate in the key features of a certain link or area deviates from the set threshold, it is immediately marked as a potential traffic surge point, notify relevant edge computing nodes and other network devices, and quickly calculate one or more emergency paths according to the current network conditions. Important traffic bypasses the congested area and is directed to the emergency path;

[0009] S400, Intelligent Routing Adjustment: Predict the traffic for the next moment, and give an adjustment to the cost function to find the path combination that minimizes the total cost;

[0010] S500, Instruction Issuance and Implementation: Convert the path combination that minimizes the calculated total cost into specific routing rules and send them to the corresponding network devices through the southbound interface. The edge computing node further optimizes the traffic allocation according to the local situation.

[0011] Furthermore, the input parameters are , and the expression is as follows:

[0012] ;

[0013] where represents the network topology matrix, is the link state vector at the t-th moment, is the application demand vector;

[0014] where the network topology matrix represents the connection relationship between nodes in the network;

[0015] where the application demand vector represents the requirements of different applications for network performance;

[0016] where the link state vector represents the state information of each link at the current moment, including bandwidth capacity, utilization rate, delay, jitter, and packet loss rate;

[0017] where the initialized parameter set is .

[0018] Furthermore, the pre-trained machine learning model uses LSTM. LSTM includes a forget gate, an input gate, a cell state, and an output gate. The t-th output of the forget gate is:

[0019] ;

[0020] where, is the weight matrix of the forget gate, is the bias term of the forget gate, is the concatenation of the (t - 1)-th intermediate representation data and , is the sigmoid function;

[0021] ;

[0022] The t-th output of the input gate and the t-th candidate cell state are calculated as follows:

[0023] ;

[0024] ;

[0025] Among them, is the weight matrix of the input gate, is the bias term of the input gate, is the hyperbolic tangent activation function, is the weight matrix for calculating the candidate cell state, is the bias term of the candidate cell state;

[0026] The t-th cell state is composed of the (t - 1)-th cell state and weighted synthesis, and the calculation method is:

[0027] ;

[0028] Among them, represents the element-wise multiplication operation;

[0029] The t-th activation value of the output gate and the t-th initialized parameter set The calculation method is:

[0030] ;

[0031] ;

[0032] Among them, is the weight matrix of the output gate, is the bias term of the output gate.

[0033] Furthermore, historical data refers to the network activity information recorded in a certain past time period, and the historical data is stored in the network management system or a dedicated log system, including traffic statistics, error reports, and configuration changes;

[0034] Real-time traffic statistical data includes bandwidth utilization, throughput, latency, jitter, packet loss rate, and error rate.

[0035] Furthermore, the establishment process of the emergency path is as follows:

[0036] S310, use the algorithm to find all possible paths from the source to the destination;

[0037] Evaluate each path, considering the following factors: path length, path capacity, path latency;

[0038] S320, obtain the candidate path set , through this candidate path set and the corresponding weight vector, calculate the comprehensive score of each candidate path:

[0039] ;

[0040] Among them, is the path length, is the remaining bandwidth of the path, is the total delay of the path, where , and are the corresponding weight parameters;

[0041] S330, and select the optimal set of emergency paths according to the comprehensive score of the selected path ; Select the optimal set of emergency paths ; Update the routing table according to the optimal set of emergency paths Redirect the affected traffic to the newly selected optimal emergency path and output the updated routing rules .

[0042] Furthermore, the calculation formula for traffic prediction at the next moment is as follows:

[0043] ;

[0044] Use the input key features at the current time point and the model parameters , combined with the traffic surge increment to predict the traffic distribution at the next moment t + 1, represents the traffic prediction at the next moment t + 1, represents the traffic surge increment, represents the impact factor of the traffic surge increment, represents the prediction function.

[0045] Furthermore, the calculation formula for cost function adjustment is as follows:

[0046] ;

[0047] Among them represents the peak load factor, where represents the peak load of the link , represents the adjusted cost to reflect the additional cost brought by the traffic surge, represents the traffic weight factor used to balance the impact of the base cost and the predicted traffic, is the initial cost of the link , is the maximum capacity of the link .

[0048] Furthermore, the solution steps for the path combination that minimizes the total cost are as follows:

[0049] 1) Through the emergency path set and the adjusted cost function , for each emergency path , calculate its total cost :

[0050] ;

[0051] where represents the cost of link in the emergency path;

[0052] 2) Use a heuristic algorithm or a linear programming method to generate new candidate paths ;

[0053] The optimization problem is as follows:

[0054] Minimize the total cost:

[0055] ;

[0056] where represents the decision variable;

[0057] Flow conservation:

[0058] ;

[0059] where represents the traffic source node, represents the destination node; represents the amount of data sent, represents the amount of data received;

[0060] Link capacity constraint:

[0061] ;

[0062] where represents the traffic flowing through link ;

[0063] Use a heuristic algorithm or a linear programming solver to solve the above optimization problem to obtain a new set of candidate paths ;

[0064] 3) Input the new set of candidate paths and the adjusted cost function , for each candidate path , calculate its total cost ;

[0065] ;

[0066] Among them, is the cost of the link in the new candidate path ;

[0067] 4) Compare the total costs of the paths in the emergency path set and the total costs of the paths in the new candidate path set , replace the paths with higher total costs in the new candidate path set, and form a candidate path set after replacement ; ;

[0068] 5) Sort the emergency paths in the optimal emergency path set in ascending order of cost , and select the paths with costs lower than the threshold as the near-optimal path set ;

[0069] 6) Merge the near-optimal paths with the candidate path set after replacement to form a total path set . Select the path combination with the minimum total cost in the total path set ; ;

[0070] ;

[0071] Among them is the total cost of the paths in the total path set;

[0072] 7) According to the formulated progressive adjustment plan, gradually migrate the real-time traffic data to the paths in the path combination with the minimum total cost .

[0073] Furthermore, the progressive adjustment plan is as follows:

[0074] Set the migration time window and the proportion of each migration ;

[0075] In each time step , migrate of the current traffic to the optimal path:

[0076] ;

[0077] Among them represents the current traffic distribution, represents the traffic on the optimal path currently.

[0078] The present invention also provides a storage medium storing non - transient computer - readable instructions for executing one or more steps in the foregoing communication processing method based on high - quality channel routing and scheduling.

[0079] The beneficial effects of the present invention are as follows:

[0080] The present invention gives optimal path suggestions for different application types and traffic patterns, especially an effective solution for burst traffic. Thanks to more intelligent traffic management and resource allocation, users enjoy a smoother and more reliable network service, and can maintain a good experience even during peak network traffic periods, avoiding the adverse effects caused by service interruptions due to congestion. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] Figure 1 is a flowchart of a communication processing method based on high - quality channel routing and scheduling proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0082] Now, the subject matter described herein will be discussed with reference to example embodiments. It should be understood that discussing these embodiments is only to enable those skilled in the art to better understand and thus implement the subject matter described herein. Without departing from the scope of protection of the content of this specification, changes can be made to the functions and arrangements of the elements discussed. Each example can omit, substitute, or add various processes or components as needed. Additionally, the features described relative to some examples can also be combined in other examples.

[0083] As Figure 1 shown, a communication processing method based on high - quality channel routing and scheduling includes the following steps:

[0084] S100, Initialization and Preparation:

[0085] Set a group of input parameters:

[0086] ;

[0087] Where represents the input parameters, represents the network topology matrix, is the link - state vector at the t - th moment, is the application - demand vector;

[0088] Among them, the network topology matrix represents the connection relationship between nodes (such as routers, switches) in the network;

[0089] Among them, the application - demand vector represents the requirements of different applications for network performance, such as minimum bandwidth, maximum latency, etc.;

[0090] Among them, the link state vector represents the state information of each link at the current moment, including bandwidth capacity, utilization rate, delay, jitter, packet loss rate, etc.

[0091] Load the model: Import the pre-trained machine learning model and output the initialized parameter set ;

[0092] In an embodiment of the present invention, the pre-trained machine learning model adopts LSTM (Long Short-Term Memory Network). The Long Short-Term Memory Network includes a forget gate, an input gate, a cell state, and an output gate. The output of the forget gate at the t-th moment is:

[0093] ;

[0094] Among them, is the weight matrix of the forget gate, is the bias term of the forget gate, is the concatenation of the (t - 1)-th intermediate representation data and , is the sigmoid function;

[0095] ;

[0096] The output of the input gate at the t-th moment and the calculation method of the t-th candidate cell state are:

[0097] ;

[0098] ;

[0099] Among them, is the weight matrix of the input gate, is the bias term of the input gate, is the hyperbolic tangent activation function, is the weight matrix for calculating the candidate cell state, is the bias term of the candidate cell state;

[0100] The t-th cell state is weighted and synthesized by the (t - 1)-th cell state and , and the calculation method is:

[0101] ;

[0102] Among them, represents the element-wise multiplication operation;

[0103] The t-th activation value of the output gate and the t-th initialized parameter set are calculated as follows:

[0104] ;

[0105] ;

[0106] where is the weight matrix of the output gate and

[0107] S200, Traffic Monitoring and Warning:

[0108] Collect real-time traffic statistical data from all network elements through the southbound interface, and then extract key features based on historical data and the collected real-time traffic statistical data The key features are used to predict future traffic;

[0109] where historical data refers to network activity information recorded within a certain past time period, and historical data is stored in a network management system (NMS) or a dedicated logging system, including traffic statistics, error reports, and configuration changes;

[0110] where real-time traffic statistical data includes bandwidth utilization, throughput, latency, jitter, packet loss rate, and error rate;

[0111] Threshold Setting: Define the upper and lower threshold values within the normal traffic range and ;

[0112] In an embodiment of the present invention, the southbound interface refers to the interface for communication between the controller and underlying network devices (such as switches, routers, etc.) in a network architecture, especially in a software-defined network (SDN) environment. It allows the SDN controller to directly program and configure these devices, thereby achieving centralized management and control of the entire network.

[0113] The main functions of the southbound interface include:

[0114] Configuration Management: Through the southbound interface, the SDN controller can set and modify rules or flow table entries on network devices. For example, under the OpenFlow protocol, the controller can specify how packets are forwarded, that is, the matching conditions (such as source IP address, destination IP address, port number, etc.) and corresponding actions (such as forwarding to a specific port, discarding packets, etc.).

[0115] Status Acquisition: The controller can collect real-time information from network devices through the southbound interface, such as link status, traffic statistics, error counts, etc. This helps monitor network performance and make corresponding routing decisions.

[0116] Event Notification: When a network device detects certain specific events (such as port status changes, new packets arriving without matching rules), it can use the southbound interface to send notifications to the controller, enabling the controller to adjust policies according to the latest situation.

[0117] Fault Handling: In case of hardware failures or other anomalies, the southbound interface allows the controller to respond quickly, take measures to restore services or reconfigure paths to ensure the high availability and reliability of the network.

[0118] S300, Traffic Surge Detection and Response: Once it is found that the key feature

[0119] (link utilization) of a certain link or area deviates from the set threshold 、 , it is immediately marked as a potential traffic surge point, notifying relevant edge computing nodes and other network devices, preparing to take emergency measures, and quickly calculating one or more emergency paths based on the current network conditions (including the status information of all links), redirecting important traffic to bypass the congested area and directing it to the emergency path;

[0120] In one embodiment of the present invention, the process of establishing an emergency path is as follows:

[0121] S310, Use the breadth-first search (BFS) or depth-first search (DFS) algorithm to find all possible paths from the source to the destination;

[0122] Evaluate each path, considering the following factors:

[0123] Path length: Shorter paths are usually more optimal.

[0124] Path capacity: Select paths with sufficient remaining bandwidth.

[0125] Path delay: Select paths with lower delay to ensure service quality.

[0126] S320, Obtain the set of candidate paths , through this set of candidate paths and the corresponding weight vector (defining the weights of each evaluation factor), calculate the comprehensive score of each candidate path :

[0127] ;

[0128] Among them, is the path length, is the remaining bandwidth of the path, is the total delay of the path, where , and are the corresponding weight parameters.

[0129] S330, and select the optimal set of emergency paths according to the comprehensive score of the selected path ; Update the routing table according to the optimal set of emergency paths ; Redirect the affected traffic to the newly selected optimal emergency path and output the updated routing rules ; ;

[0130] S400, intelligent routing adjustment:

[0131] S410, traffic prediction update, the calculation formula is as follows:

[0132] ;

[0133] Use the input key features at the current time point and model parameters to predict the traffic distribution at the next moment t + 1 in combination with the traffic surge increment , where represents the traffic prediction at the next moment t + 1, represents the traffic surge increment, represents the impact factor of the traffic surge increment, represents the prediction function; ;

[0134] S420, cost function adjustment, the calculation formula is as follows:

[0135]

[0136] where represents the peak load factor, where represents the peak load of link , represents the adjusted cost to reflect the additional cost brought by the traffic surge, represents the traffic weight factor used to balance the basic cost and the impact of the predicted traffic, is the initial cost of link , is the maximum capacity of link ;

[0137] S430, solve the optimal path: Use heuristic algorithms or linear programming methods to find the path combination that minimizes the total cost 。

[0138] 1) Calculate the total cost for each emergency path through the emergency path set and the adjusted cost function : ; :

[0139] ;

[0140] where represents the cost of link in the emergency path;

[0141] 2) Generate new candidate paths using a heuristic algorithm or a linear programming method ;

[0142] The optimization problem is as follows:

[0143] Minimize the total cost:

[0144] ;

[0145] where represents the decision variable;

[0146] Flow conservation:

[0147] ;

[0148] where represents the source node of the flow, represents the destination node; represents the amount of data sent, represents the amount of data received;

[0149] Link capacity constraint:

[0150] ;

[0151] where represents the flow through link ;

[0152] Solve the above optimization problem using a heuristic algorithm (such as a genetic algorithm, simulated annealing) or a linear programming solver (such as the Simplex algorithm, interior point method) to obtain a new set of candidate paths ;

[0153] 3) Input the new set of candidate paths and the adjusted cost function and calculate the total cost for each candidate path ; ;

[0154] ;

[0155] Wherein, is the cost of the link in the new candidate path ;

[0156] 4) Compare the total cost of the paths in the emergency path set and the total cost of the paths in the new candidate path set , replace the path with the higher total cost in the new candidate path set, and form the replaced candidate path set ; ; ;

[0157] 5) Sort the emergency paths in the optimal emergency path set in ascending order of cost , and select the paths with a cost lower than the threshold as the near-optimal path set ;

[0158] 6) Combine the near-optimal paths with the replaced candidate path set to form the total path set , and select the path combination with the minimum total cost in the total path set ; ;

[0159] ;

[0160] Wherein is the total cost of the paths in the total path set;

[0161] 7) According to the formulated progressive adjustment plan, gradually migrate the real-time traffic data to the paths in the path combination with the minimum total cost ;

[0162] The progressive adjustment plan is as follows:

[0163] 1. Set the migration time window and the proportion of each migration ;

[0164] 2. Within each time step , migrate of the current traffic to the optimal path:

[0165] ;

[0166] Wherein represents the current traffic distribution, Represents the traffic currently on the optimal path;

[0167] S500, Instruction Issuance and Implementation: The path combination that minimizes the calculated total cost is converted into specific routing rules and sent to the corresponding network devices through the southbound interface. The edge computing nodes can further optimize the traffic distribution according to local conditions to relieve the core network pressure.

[0168] In an embodiment of the present invention, it specifically includes the following steps:

[0169] S510, Routing Rule Generation:

[0170] Convert each optimal path into specific routing rules, generate , verify the routing rules , ensure their correctness and security, and form ;

[0171] S520, Instruction Issuance:

[0172] Construct a control instruction , and send it to the corresponding network devices through the southbound interface, collect the transmission status report , and confirm that the instruction is successfully executed;

[0173] S530, Edge Computing Node Optimization:

[0174] The edge computing node analyzes the local traffic data , generates an analysis result , formulates and implements an optimization strategy based on the analysis result , adjusts the local traffic distribution; monitors the optimization effect to ensure that the core network pressure is relieved, and forms an optimized local traffic distribution .

[0175] Formulate an optimization strategy based on the analysis result , such as caching hot content, adjusting the service deployment location, enabling load balancing, etc. For critical business traffic, consider setting higher priorities or dedicated channels.

[0176] An embodiment of the present invention also proposes a storage medium storing non - transient computer - readable instructions for executing one or more steps in the foregoing communication processing method based on high - quality channel routing scheduling.

[0177] The computer program can be stored / distributed on a suitable medium, such as an optical storage medium or a solid - state medium supplied together with other hardware or as part of other hardware, but can also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems.

[0178] The above embodiments of the present embodiment have been described, but the present embodiment is not limited to the above specific implementation manners. The above specific implementation manners are merely illustrative rather than restrictive. Under the inspiration of the present embodiment, those of ordinary skill in the art can also make many forms, all of which fall within the protection scope of the present embodiment.

Claims

1. A communication processing method based on high-quality channel routing scheduling, characterized in that It includes the following steps: S100, set a set of input parameters, import a pre-trained machine learning model, and output an initialized parameter set; S200, collect real-time traffic statistics data from all network elements through the southbound interface, extract key features based on historical data and the collected real-time traffic statistics data, and define upper and lower threshold values within the normal traffic range; S300, when the link utilization rate in the key features of a certain link or area deviates from the set threshold, immediately mark it as a potential traffic surge point, notify relevant edge computing nodes and network devices, calculate at least one emergency path according to the current network status, detour important traffic to avoid the congestion area, and direct it to the emergency path; S400, predict the traffic for the next moment, and give an adjustment to the cost function to find the path combination that minimizes the total cost; The calculation formula for predicting the traffic for the next moment is as follows: ； Using the current time point of the input key features and model parameters , combined with the sudden increase in traffic increment to predict the traffic distribution at the next moment t+1, denotes the traffic prediction at the next moment t+1, denotes the sudden increase in traffic increment, denotes the impact factor of the sudden increase in traffic increment, denotes the prediction function; Among them, the calculation formula for adjusting the cost function is as follows: ； wherein represents the peak load factor, wherein represents the peak load of the link , represents the adjusted cost to reflect the additional cost brought by traffic surges, represents the traffic weight factor used to balance the impact of the base cost and the predicted traffic, is the initial cost of the link , is the maximum capacity of the link ; S500, convert the calculated path combination that minimizes the total cost into a routing rule, and send it to the corresponding network device through the southbound interface. The edge computing node further optimizes the traffic distribution according to the local situation.

2. The communication processing method based on high-quality channel routing scheduling according to claim 1, wherein The input parameters include a network topology matrix, a link state vector at the t-th moment, and an application demand vector; Among them, the network topology matrix represents the connection relationship between nodes in the network; Among them, the application demand vector represents the requirements of different applications for network performance; Among them, the link state vector represents the state information of each link at the current moment, including bandwidth capacity, utilization rate, delay, jitter, and packet loss rate.

3. The communication processing method based on high-quality channel routing scheduling according to claim 2, characterized in that The pre-trained machine learning model uses LSTM, and LSTM includes a forget gate, an input gate, a cell state, and an output gate.

4. A communication processing method based on high-quality channel routing scheduling according to claim 3, characterized in that, Historical data refers to the network activity information recorded during a past period of time. The historical data is stored in a network management system and a dedicated log system, including traffic statistics, error reports, and configuration changes; The real-time traffic statistics data includes bandwidth utilization rate, throughput, delay, jitter, packet loss rate, and error rate.

5. A communication processing method based on high-quality channel routing scheduling according to claim 4, characterized in that The establishment process of the emergency path is as follows: S310, use an algorithm to find all possible paths from the source to the destination; Evaluate each path, considering the following factors: path length, path capacity, and path delay; S320, obtain a candidate path set, and calculate the comprehensive score of each candidate path through this candidate path set and the corresponding weight vector; S330, and select the optimal emergency path set according to the comprehensive score of the selected path; update the routing table according to the optimal emergency path set, redirect the affected traffic to the newly selected optimal emergency path, and output the updated routing rule.

6. The communication processing method based on high-quality channel routing scheduling according to claim 5, wherein The solution steps for the path combination that minimizes the total cost are as follows: Through the emergency path set and the adjusted cost function, calculate the total cost of each emergency path; Use a heuristic algorithm and a linear programming method to generate new candidate paths; The optimization problem includes minimizing the total cost, traffic conservation, and link capacity limitation; Use a heuristic algorithm and a linear programming solver to solve the above optimization problem to obtain a new candidate path set; Input the new set of candidate paths and the adjusted cost function, and for each candidate path, calculate its total cost; Compare the total costs of the paths in the emergency path set with the total costs of the paths in the new candidate path set, and replace the paths with higher total costs in the new candidate path set to form the replaced candidate path set; Sort the emergency paths in the optimal emergency path set in ascending order of cost, and select the paths with costs lower than the threshold as the near-optimal path set; Merge the near-optimal paths with the replaced candidate path set to form the total path set, and select the path combination with the minimum total cost in the total path set; According to the formulated progressive adjustment plan, gradually migrate the real-time traffic data to the paths in the path combination with the minimum total cost.

7. A communication processing method based on high-quality channel routing scheduling according to claim 6, characterized in that The progressive adjustment plan is as follows: Set the migration time window and the proportion of each migration; At each time step, migrate the current flow to the optimal path: ； Among them represents the current traffic distribution represents the traffic on the current optimal path 8. A storage medium stores non-transitory computer-readable instructions for performing the steps in a communication processing method based on high-quality channel routing scheduling as described in any one of claims 1-7.

Images