Bandwidth adaptive allocation method based on service priority evaluation and flow prediction

Abstract

The invention relates to a bandwidth adaptive allocation method based on service priority evaluation and flow prediction, and belongs to the technical field of computers. The method comprises the following steps: collecting historical communication data and network environment data of each application system; constructing a business priority evaluation model; constructing a flow prediction model; performing initial bandwidth allocation according to the priority evaluation model; according to the network performance data, the application change condition and the communication quality which are acquired in real time, reallocating and adjusting the bandwidth based on the flow prediction result of the flow prediction model; and sending the dynamically adjusted and optimized bandwidth allocation decision to each application system, executing bandwidth allocation operation, and adjusting the data transmission rate. According to the invention, the most effective utilization of bandwidth resources can be ensured, and the overall performance of network communication is ensured.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of computer, and particularly relates to a bandwidth self-adaptive allocation method based on service priority evaluation and traffic prediction. BACKGROUND

[0002] Application system communication network bandwidth resources are limited in a distributed communication control network environment, and different application systems have different demands for bandwidth resources, and the allocated bandwidth resources are idle after a certain application system is damaged, so limited network bandwidth resources need to be self-adaptively allocated to fully utilize bandwidth resources while guaranteeing stable operation and reliable transmission of key tasks. The project designs a bandwidth self-adaptive allocation method based on service priority evaluation and traffic prediction, dynamically adjusts network bandwidth resources according to network load conditions, application system demands, communication quality and the like, provides key system task efficiency, and reduces energy consumption and resource cost. SUMMARY (I) Technical problems to be solved The technical problem to be solved by the application is how to provide a bandwidth self-adaptive allocation method based on service priority evaluation and traffic prediction to solve the problem of self-adaptively allocating limited network bandwidth resources, fully utilizing bandwidth resources while guaranteeing stable operation and reliable transmission of key tasks.

[0003] (II) Technical scheme In order to solve the above technical problem, the application provides a bandwidth self-adaptive allocation method based on service priority evaluation and traffic prediction, which comprises the following steps: Step 1: Collect historical communication data and network environment data of each application system; Step 2: Data preprocessing is performed on various data collected in step 1 to meet the needs of future model construction; Step 3: Construct a service priority evaluation model; Step 4: Construct a traffic prediction model; Step 5: Perform initial bandwidth allocation according to the model constructed in step 3; Step 6: Reallocate and adjust bandwidth based on the traffic prediction results of the traffic prediction model according to real-time collected network performance data, application change conditions and communication quality; Step 7: Send the bandwidth allocation decision after dynamic adjustment and optimization to each application system to perform bandwidth allocation operation and adjust data transmission rate.

[0004] (III) Beneficial effects The application provides a bandwidth adaptive allocation method based on service priority evaluation and traffic prediction. By using the bandwidth adaptive method provided by the application, future traffic in a period of time can be predicted in advance according to historical bandwidth occupation, application demand, etc., and bandwidth allocation is comprehensively performed according to the current network state, real-time communication data, etc., so that bandwidth resources are most effectively utilized and the overall performance of network communication is ensured. BRIEF DESCRIPTION OF DRAWINGS

[0005] Figure 1 The bandwidth adaptive allocation overall architecture diagram of the application is shown in the figure. Figure 2 The bandwidth adaptive allocation flowchart is shown in the figure. DETAILED DESCRIPTION

[0006] In order to make the purpose, content and advantages of the application clearer, the specific embodiments of the application are described in further detail below in combination with the drawings and examples.

[0007] The technical problem to be solved by the application is how to provide a bandwidth adaptive allocation method based on fixed physical bandwidth, application bandwidth demand, network real-time state and communication quality, so as to ensure that bandwidth resources are most effectively utilized and the overall performance of network communication is ensured.

[0008] In order to solve the above technical problem, the application provides a bandwidth adaptive allocation method based on multiple source information, which comprises the following steps: Step one: collect historical communication data of each application system and network environment data; Step two: data preprocessing is performed on the various data collected in step one, so as to meet the needs of future model construction; Step three: a service priority evaluation model is constructed; Step four: a traffic prediction model is constructed; Step five: bandwidth initial allocation is performed according to the model constructed in step three; Step six: according to the real-time collected network performance data, application change and communication quality, the bandwidth is re-allocated and adjusted based on the traffic prediction result of the traffic prediction model; Step seven: the bandwidth allocation decision after dynamic adjustment and optimization is sent to each application system, and bandwidth allocation operation is performed to adjust the data transmission rate.

[0009] Embodiment 1 Step one: data collection The following data are mainly included: (1) historical communication data of each application system, including communication time, data size, service type information; (2) Collect network signal strength, channel quality, interference situation, node connection situation, and statistics of total bandwidth usage in the network, including allocated bandwidth and remaining available bandwidth.

[0010] Step two: data preprocessing, cleaning and organizing the collected data (1) Data cleaning The purpose of data cleaning is to remove errors, incomplete or abnormal data, and ensure data quality and consistency.

[0011] 1) First, load data, load historical communication data and network status data from the original data source to the database, ensure that data is not lost or damaged during loading; 2) Check data quality, including missing value check, outlier check, duplicate value check, data format check, etc., identify errors, missing values, outliers and duplicates in data; clean data, including missing value processing, such as deleting records containing missing values, filling missing values using linear interpolation method, deleting obvious error outliers, merging duplicate records, repairing fields not conforming to format (unifying time format, unifying units, etc.).

[0012] 3) Verify data The invention uses variance-based statistical indicators to check whether the data is reasonable, ensuring that the cleaning operation is correct.

[0013] (2) Data standardization Convert data to a uniform format and dimension for subsequent analysis and processing.

[0014] 1) Data format uniformity Time format uniformity: unify time fields to standard format (e.g. YYYY-MM-DD HH:MM:SS).

[0015] Value format uniformity: unify numerical fields to integers.

[0016] Text format uniformity: unify text fields to lowercase and remove extra spaces.

[0017] 2) Data dimension uniformity Convert data of different dimensions to the same dimension.

[0018] Normalization: scale data to Min-Max normalization.

[0019] Standardization: convert data to Z-score standardization.

[0020] Unit uniformity: unify data of different units to the same unit (e.g. convert speed from "km / h" to "m / s").

[0021] (3) Data organization Organize the cleaned and standardized data according to certain rules to facilitate subsequent analysis.

[0022] 1) Data grouping and classification Group or classify data according to specific rules.

[0023] 2) Data storage Store the organized data in the database to prepare for future data analysis.

[0024] Step three: business priority evaluation (1) According to the importance and real-time requirements of each system business, a set of priority evaluation index system is prepared in advance, for example, if the priority of business system 1 is higher than that of business system 2, a higher weight is given to business system 1, and a lower weight is given to business system 2.

[0025] (2) Establish a business priority evaluation model, considering business type and urgency, etc., to assign a priority level to each business.

[0026] (3) According to the collected real-time data, use the priority evaluation model to calculate the priority of each system in real time, so as to distinguish in bandwidth allocation.

[0027] Step four: construction of traffic prediction model According to the collected historical communication data, use LSTM model to predict future communication traffic, construct traffic prediction model, and according to the collected real-time data, predict the expected data transmission of each system in the next 1 minute through the traffic prediction model. Specifically as follows: (1) LSTM model construction Input layer: define the shape of input data (time step, feature number), the time step selected by the invention is 10, the feature number is 5, and the input shape is (10, 5).

[0028] LSTM layer: add LSTM layer, set hidden unit number, return sequence, etc. The invention selects hidden unit number as 128 layers, and returns all time step output sequence.

[0029] Fully connected layer: set the fully connected layer to 64 for classification.

[0030] Output layer: set the output layer and activation function, use softmax classifier as the activation function of the output layer.

[0031] (2) Model training The historical communication data is used as training data, the cross-entropy loss function is used as the model loss function, the model is trained using the training data, the batch size is set to 32, and the iteration is 50 times.

[0032] (3) Feature extraction The feature vector output by the LSTM model is extracted, which reflects the time dependence of the traffic.

[0033] (4) Traffic prediction After the model training is completed, the future one minute of each system data transmission can be predicted according to the trained model, which provides a decision basis for subsequent bandwidth allocation.

[0034] Step five: initial bandwidth allocation (1) In the initial stage of the system, a fair allocation strategy is adopted, and the total bandwidth is evenly distributed to each system to ensure that each system has basic communication opportunities and avoid the situation that some systems cannot communicate normally due to insufficient bandwidth; (2) According to the business priority evaluation results, the initial bandwidth is appropriately adjusted, and more initial bandwidth is allocated to high-priority businesses to ensure that critical businesses can be transmitted in time.

[0035] Step six: dynamic adjustment and optimization.

[0036] (1) Continuously monitor network status and business changes, including bandwidth usage, signal quality, business priority changes, etc., and judge whether it meets business needs by measuring the actual data transmission rate of each system in real time; (2) The monitored information is fed back to the business priority allocation model to provide the basis for dynamic adjustment; (3) The monitored information is input into the traffic prediction model, and according to the traffic prediction results, the bandwidth allocation is adjusted in advance. For systems with low traffic demand prediction or have been down, reduce their bandwidth allocation or not allocate, and allocate idle bandwidth to other systems; (4) According to the network performance indicators, data transmission delay, packet loss rate, etc., optimize the bandwidth allocation, use adaptive algorithms to automatically adjust the bandwidth allocation strategy and parameters according to the network performance changes, and realize the dynamic optimization of allocation, such as the data transmission delay of a certain system is too high, which may affect the normal operation of its business, the bandwidth allocation of the system can be dynamically increased to reduce the delay.

[0037] Step seven: bandwidth allocation execution.

[0038] The bandwidth allocation decision after dynamic adjustment and optimization is sent to each network device to execute the bandwidth allocation operation and adjust the data transmission rate.

[0039] Beneficial effects: By using the bandwidth self-adapting method provided by the application, the future traffic in a period of time can be predicted in advance according to the historical bandwidth occupation, application demand, etc., and the bandwidth is allocated comprehensively according to the current network state, real-time communication data, etc., so that the bandwidth resources are used most effectively, and the overall performance of network communication is ensured.

[0040] The above only describes the preferred embodiments of the present application, and it should be noted that those skilled in the art can make some improvements and modifications without departing from the technical principles of the present application, and these improvements and modifications should also be considered as the protection scope of the present application.

Claims

1. A bandwidth adaptive allocation method based on service priority assessment and traffic prediction, characterized in that, The method includes the following steps: Step 1: Collect historical communication data and network environment data from each application system; Step 2: Perform data preprocessing on the various types of data collected in Step 1 to meet the needs of future model construction; Step 3: Construct a business priority assessment model; Step 4: Build a traffic prediction model; Step 5: Perform initial bandwidth allocation based on the model constructed in Step 3; Step Six: Based on the real-time collected network performance data, application changes, and communication quality, and using the traffic prediction results from the traffic prediction model, reallocate and adjust the bandwidth. Step 7: Send the dynamically adjusted and optimized bandwidth allocation decision to each application system, execute the bandwidth allocation operation, and adjust the data transmission rate.

2. The bandwidth adaptive allocation method based on service priority assessment and traffic prediction as described in claim 1, characterized in that, The data collected in step one includes: Historical communication data for each application system, including communication time, data volume, and service type information; Collect data on network signal strength, channel quality, interference, and node connectivity; and statistically analyze the total bandwidth usage in the network, including allocated bandwidth and remaining available bandwidth.

3. The bandwidth adaptive allocation method based on service priority assessment and traffic prediction as described in claim 2, characterized in that, The preprocessing in step two includes data cleaning, specifically: First, load the data by loading historical communication data and network status data from the original data source into the database, ensuring that no data is lost or corrupted during the data loading process. Data quality checks include checking for missing values, outliers, duplicates, and data format; identifying errors, missing values, outliers, and duplicates in the data. Data cleaning includes handling missing values, such as deleting records containing missing values, filling missing values ​​with linear interpolation, deleting obviously erroneous outliers, merging duplicate records, and repairing fields that do not conform to the format. Validate the data by using variance-based statistical indicators to check whether the data is reasonable.

4. The bandwidth adaptive allocation method based on service priority assessment and traffic prediction as described in claim 1, characterized in that, The preprocessing in step two includes data standardization, specifically including: Data format standardization includes: standardization of time format, standardization of data format, and standardization of file format; among which, The time format is standardized by using the same format for all time fields. The numerical format should be standardized to: all numeric fields should be set to integers; The text format should be standardized as follows: all text fields should be lowercase and extra spaces should be removed. Data dimension unification: converting data with different dimensions to the same dimension, including: normalization, standardization, and unit unification; among which, Normalization is defined as scaling the data to Min-Max standardization. Standardization refers to converting the data into Z-score standardized data. Unit unification: unify data from different units into the same unit.

5. The bandwidth adaptive allocation method based on service priority assessment and traffic prediction as described in claim 1, characterized in that, The preprocessing in step two includes data cleaning, specifically including: Data grouping and classification: grouping or classifying data according to specific rules; Data storage: The organized data is stored in a database to prepare for future data analysis.

6. The bandwidth adaptive allocation method based on service priority assessment and traffic prediction as described in any one of claims 1-5, characterized in that, Step three includes: Based on the importance and real-time requirements of each system's business, a set of priority evaluation indicators should be developed in advance; Establish a business priority assessment model that comprehensively considers business type and urgency, and assign a priority level to each business. Based on the collected real-time data, the priority of each system is calculated in real time using a priority evaluation model so that it can be differentiated during bandwidth allocation.

7. The bandwidth adaptive allocation method based on service priority assessment and traffic prediction as described in claim 6, characterized in that, Step four includes: Based on collected historical communication data, a traffic prediction model is constructed using an LSTM model. Based on collected real-time data, the traffic prediction model predicts the expected data transmission volume of each system within the next minute. Specifically, this includes: (1) LSTM model construction Input layer: Defines the shape of the input data; LSTM layer: Add an LSTM layer, set the number of hidden units and the return sequence parameters. The number of hidden units is 128, and the output sequence of all time steps is returned. Fully connected layer: The fully connected layer is set to 64 for classification; Output layer: Set the output layer and activation function, using the softmax classifier as the activation function for the output layer; (2) Model training Historical communication data is used as training data, cross-entropy loss function is used as model loss function, and training data is used to train the model. (3) Feature extraction Extract the feature vector output by the LSTM model, which reflects the time dependence of the flow; (4) Flow forecasting After the model is trained, the data transmission volume of each system is predicted in the next minute based on the trained model, providing a basis for subsequent bandwidth allocation decisions.

8. The bandwidth adaptive allocation method based on service priority assessment and traffic prediction as described in claim 7, characterized in that, Step five includes: In the initial stage of the system, a fair allocation strategy is adopted to distribute the total bandwidth equally to each system, ensuring that each system has a basic communication opportunity and avoiding situations where some systems cannot communicate normally due to insufficient bandwidth. Based on the service priority assessment results, the initial bandwidth is adjusted appropriately, allocating more initial bandwidth to high-priority services to ensure that critical services can be transmitted in a timely manner.

9. The bandwidth adaptive allocation method based on service priority assessment and traffic prediction as described in claim 8, characterized in that, Step six includes: Continuously monitor network status and service changes, including bandwidth usage, signal quality, and changes in service priority. Determine whether the actual data transmission rate of each system meets service requirements by measuring the actual data transmission rate in real time. The monitored information is promptly fed back to the business priority allocation model to provide a basis for dynamic adjustments; Based on traffic forecast results, adjust bandwidth allocation in advance. For systems with low traffic demand forecasts or those that have already crashed, reduce or eliminate bandwidth allocation and allocate idle bandwidth to other systems.

10. The bandwidth adaptive allocation method based on service priority assessment and traffic prediction as described in claim 9, characterized in that, Step six also includes: optimizing bandwidth allocation based on network performance indicators, using an adaptive algorithm to automatically adjust bandwidth allocation strategies and parameters according to changes in network performance, thereby achieving dynamic optimization of allocation.

Images