Communication information processing method based on edge computing

Abstract

The invention discloses a communication information processing method based on edge computing, which belongs to the technical field of communication network intelligent management and optimization, and comprises the following steps: collecting communication data in real time by using edge equipment, and carrying out data preprocessing on the communication data to obtain a feature vector; inputting the feature vector into an analysis model constructed based on a decision tree algorithm to identify the abnormal mode of the communication network, and outputting an identification result; carrying out adaptive channel allocation and load balancing processing according to the identification result; and optimizing the communication parameters after the adaptive channel allocation and load balancing processing by using a cold ice algorithm, and reconfiguring the signal frequency and power of the transmitter according to the optimized communication parameters. The method improves the accuracy and real-time performance of anomaly detection, optimizes the configuration and utilization of network resources, reduces the network delay and packet loss rate, and improves the overall performance of a communication system and the user experience.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent management and optimization of communication networks, and in particular to a communication information processing method based on edge computing. Background Art

[0002] The traditional cloud computing model transmits large amounts of data to central servers for processing, resulting in a sharp increase in network bandwidth pressure and a significant increase in latency, which cannot meet the requirements of real-time and high reliability. To meet this challenge, edge computing, as an emerging distributed computing architecture, has gradually received widespread attention.

[0003] Edge computing sinks data processing to the edge of the network, using the computing power of edge devices to process data nearby, thereby effectively reducing network latency and improving data processing efficiency. However, existing edge computing technology still has some shortcomings in communication information processing:

[0004] First, the accuracy of abnormal pattern recognition needs to be improved. Traditional methods mainly rely on simple threshold settings or statistical analysis, which are difficult to adapt to complex and changing network environments, resulting in high rates of missed reports and false positives in anomaly detection.

[0005] Secondly, in terms of allocation and scheduling of network resources, there is a lack of adaptability and intelligent means, making it difficult to respond to changes in network status in a timely manner and achieve dynamic channel allocation and load balancing, which not only affects the overall performance of the network, but may also lead to waste of resources or congestion;

[0006] Finally, the optimization of communication parameters usually adopts a fixed strategy and fails to make full use of intelligent optimization algorithms, which makes it difficult to achieve optimal system performance. Summary of the invention

[0007] The purpose of this section is to summarize some aspects of the embodiments of the present invention and briefly introduce some preferred embodiments. Some simplifications or omissions may be made in this section and the abstract of the specification and the title of the invention of this application to avoid blurring the purpose of this section, the abstract of the specification and the title of the invention, and such simplifications or omissions cannot be used to limit the scope of the present invention.

[0008] In view of the above existing problems, the present invention is proposed.

[0009] Therefore, the technical problem solved by the present invention is that the existing communication information processing technology has the problems of inaccurate anomaly detection, unintelligent resource allocation, and insufficient optimization of communication parameters.

[0010] In order to solve the above technical problems, the present invention provides the following technical solutions: using edge devices to collect communication data in real time, and preprocessing the data to obtain feature vectors;

[0011] Inputting the feature vector into an analysis model constructed based on a decision tree algorithm to identify abnormal patterns in the communication network, and outputting the identification result;

[0012] Performing adaptive channel allocation and load balancing processing according to the identification result;

[0013] The communication parameters processed by the adaptive channel allocation and load balancing are optimized using the Ice algorithm, and the signal frequency and power of the transmitter are reconfigured according to the optimized communication parameters.

[0014] As a preferred solution of the communication information processing method based on edge computing described in the present invention, the collected communication data at least includes signal strength, network delay, packet loss rate, number of user connections and data transmission rate.

[0015] As a preferred solution of the communication information processing method based on edge computing described in the present invention, the preprocessing removes noise through a moving average method, standardizes all data to have the same dimension, converts them into proportional values, and outputs a feature vector.

[0016] As a preferred solution of the communication information processing method based on edge computing described in the present invention, based on the decision tree algorithm, the feature vector is used as input to build an analysis model, and then the information gain IG is used to divide the nodes. The mathematical expression formula is:

[0017]

[0018] Among them, v is a value of the feature attribute A, D v is the data subset when the feature attribute A takes the value v, D is the current data set, A is the feature attribute, and H(D) is the entropy of the data set D:

[0019]

[0020] C represents all categories, and p(c) represents the probability of category c.

[0021] As a preferred solution of the communication information processing method based on edge computing described in the present invention, a decision tree is applied to the input feature vector, wherein:

[0022] In the leaf nodes of the decision tree, each node represents an output category. When the feature vector is sent to the decision tree, the decision tree will make judgments on different paths based on the value of the feature vector, and finally reach the designated leaf node to obtain the classification result of the feature vector and output an abnormal identification, that is, the recognition result. The abnormal identification includes normal, high delay, high packet loss rate, and overload.

[0023] As a preferred solution of the communication information processing method based on edge computing described in the present invention, the current load is calculated according to the identification result, and adaptive channel allocation is performed according to the identified abnormal mode, and the abnormal mode includes a high delay mode, a high packet loss rate mode and an overload mode.

[0024] As a preferred solution of the communication information processing method based on edge computing described in the present invention, the adaptive channel allocation according to the identified abnormal mode includes:

[0025] When a high-latency mode is identified, the channel bandwidth is increased;

[0026] When a high packet loss rate mode is identified, the retransmission mechanism is adjusted to reduce the transmission rate;

[0027] When an overload condition is detected, the current is limited to reduce the number of user connections.

[0028] As a preferred solution of the communication information processing method based on edge computing described in the present invention, the communication parameters are optimized using the ice algorithm, including:

[0029] Set the initial signal frequency F r and power P w ;

[0030] Use the ice algorithm for search and optimization, the objective function is:

[0031] Z=w 1 ·L+w 2 ·P

[0032] Randomly generate solutions in the search space and evaluate their fitness;

[0033] Update the current best solution until the termination condition is met;

[0034] Output optimized communication parameters F′ r and P′ w :

[0035] F′ r =F r ×k

[0036] P′ w =P w ×m

[0037] Among them, k and m are optimization factors, w 1 、w 2 is the weight, reflecting the importance of network delay and packet loss rate, L is the network delay, and P is the packet loss rate.

[0038] Beneficial effects of the present invention:

[0039] 1. Improve data quality, reduce noise interference, and make subsequent analysis more accurate and efficient;

[0040] 2. Accurately identify abnormal modes in the communication network (such as high latency, high packet loss rate, and overload), improving network reliability and service quality;

[0041] 3. It realizes the dynamic optimization configuration of network resources, effectively alleviates network congestion, improves data transmission efficiency, and improves user experience and overall network performance;

[0042] 4. Reduce network delay and packet loss rate, and improve the overall performance and stability of the communication system. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:

[0044] Figure 1 It is a flow chart of the communication information processing method based on edge computing shown in the present invention;

[0045] Figure 2 A schematic diagram of network delay comparison shown in the present invention;

[0046] Figure 3 A schematic diagram of packet loss rate comparison shown in the present invention;

[0047] Figure 4 A schematic diagram showing a comparison of data throughputs shown in the present invention;

[0048] Figure 5 This is a schematic diagram of comparing the accuracy of anomaly detection shown in the present invention;

[0049] Figure 6 It is a schematic diagram of response time comparison shown in the present invention. DETAILED DESCRIPTION

[0050] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0051] Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without making any creative work should fall within the scope of protection of the present invention.

[0052] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0053] According to an embodiment of the present invention, Figure 1 The flowchart shown is a communication information processing method based on edge computing, which specifically includes the following steps:

[0054] S1. Use edge devices to collect communication data in real time and preprocess the data to obtain feature vectors. The following points need to be explained in this step:

[0055] The collected communication data includes at least signal strength, network delay, packet loss rate, number of user connections and data transmission rate;

[0056] The preprocessing uses the moving average method to remove noise, standardize all data to make them have the same dimension, convert them into proportional values, and output the feature vector.

[0057] As an example, noise removal: Applying the moving average method

[0058]

[0059] Format conversion: Normalize all data to have the same dimension and convert to proportional values:

[0060]

[0061] The preprocessed data constitutes a feature vector: F = [S avg ,L norm ,P norm ,N,R]

[0062] Among them, n is the window size, which is selected as 5, S avg It represents the average signal strength after noise removal, reflecting the stability of the communication signal. i is the signal strength, L i is the network delay, P i is the packet loss rate, N is the number of user connections, R is the data transmission rate, L norm Indicates the standardized network delay, with a value range of 0 to 1. norm Indicates the standardized packet loss rate, ranging from 0 to 1, reflecting the stability of network quality;

[0063] Assume that the signal strength data collected in real time is: S = [10, 12, 11, 9, 13] (unit: dBm);

[0064] Calculate S avg :

[0065]

[0066] Standardization:

[0067] Assume that the network delay L = 100ms and the maximum delay L max =200ms, minimum delay L min =50ms:

[0068]

[0069] Assume that the packet loss rate P = 0.03 (i.e. 3%) and the maximum packet loss rate is 5%:

[0070]

[0071] The final output feature vector: F = [11, 0.25, 0.6].

[0072] S2. Input the feature vector into the analysis model built based on the decision tree algorithm to identify abnormal patterns in the communication network, and output the identification results. Among them, what needs to be explained in this step is:

[0073] Based on the decision tree algorithm, the feature vector is used as input to build an analysis model, and then the information gain IG is used to divide the nodes. The mathematical expression formula is:

[0074]

[0075] Among them, v is a value of the feature attribute A, D v is the data subset when the feature attribute A takes the value v, D is the current data set, A is the feature attribute, and H(D) is the entropy of the data set D:

[0076]

[0077] C represents all categories, and p(c) represents the probability of category c.

[0078] Specifically, a decision tree is applied to the input feature vector, including:

[0079] In the leaf nodes of the decision tree, each node represents an output category. When the feature vector is sent to the decision tree, the decision tree will make judgments on different paths according to the value of the feature vector, and finally reach the specified leaf node to obtain the classification result of the feature vector and output the abnormal identification, that is, the recognition result. The abnormal identification includes normal, high delay, high packet loss rate, and overload; among them:

[0080] In high latency mode, Lnorm >0.8;

[0081] In high packet loss rate mode, P norm >0.05;

[0082] In overload mode, N>N threshold , N thrreshold The threshold value of the number of user connections;

[0083] As an example, assume that the feature vector collected in real time is:

[0084] F=[S avg =11,L norm =0.85,P norm =0.06,N=10,R=16]

[0085] Input features: S avg =11(signal strength), L norm =0.85(delay), P norm =0.06 (packet loss rate), N = 10 (number of connections), R = 16 (transmission rate);

[0086] Assume that the existing decision tree model is as follows:

[0087] Root node: judge L norm , if L norm >0.8, then judge P norm , if P norm >0.05, the output is "high latency, high packet loss" abnormality. If P norm ≤0.05, output "high delay, normal", if L norm ≤0.8, then judge N, if N>8, output "normal", if N≤8, output "low load";

[0088] Feed the input feature F into the decision tree: Check L norm =0.85(satisfies the condition L norm >0.8), enter the left subtree and check P norm =0.06 (satisfies the condition P norm >0.05), enter the corresponding leaf node;

[0089] Output result: The judgment result is "high latency, high packet loss";

[0090] Record abnormal patterns and their characteristic parameters:

[0091] Abnormal mode = "high latency, high packet loss"

[0092] Characteristic parameter = [S avg =11,L norm =0.85,Pnorm =0.06,N=10,R=16].

[0093] S3, perform adaptive channel allocation and load balancing processing according to the identification result. Among them, what needs to be explained in this step is:

[0094] Based on the identification results, the current load is calculated, and adaptive channel allocation is performed based on the identified abnormal modes, including high latency mode, high packet loss rate mode, and overload mode.

[0095] Adaptive channel allocation based on identified abnormal patterns, including:

[0096] When a high-latency mode is identified, the channel bandwidth is increased;

[0097] When a high packet loss rate mode is identified, the retransmission mechanism is adjusted to reduce the transmission rate;

[0098] When an overload mode is identified, the current is limited to reduce the number of user connections;

[0099] Furthermore, set the current load L c :

[0100]

[0101] Among them, R k is the transmission rate of the kth user;

[0102] The allocation process for exception mode processing is as follows:

[0103] High latency mode: When L norm >0.8, increase bandwidth C:

[0104] C new =C old +ΔC (e.g., ΔC = 10 Mbps)

[0105] High packet loss rate mode: When P norm >0.05, reduce the transmission rate R:

[0106] R new =R old ×0.8

[0107] Overload mode: When N>N threshold , current limiting reduces the number of connections N:

[0108] N new =N old -ΔN (for example, ΔN=2)

[0109] Completion standard threshold determination:

[0110] Delay standard: L norm ≤0.5;

[0111] Packet loss standard: P norm ≤0.02;

[0112] Connection number standard: N≤N threshold ;

[0113] Among them, C new represents the adjusted channel bandwidth, C old Represents the current channel bandwidth, R new Represents the data transmission rate adjusted according to the abnormal situation, R old Indicates the current data transmission rate, N new Indicates the number of user connections after current limiting, N old Indicates the current number of user connections.

[0114] As an example, assume that during real-time monitoring, the current feature vector is detected as:

[0115] F=[S avg =11,L norm =0.9,P norm =0.06,N=10,R=20]

[0116] Based on the preset threshold:

[0117] High latency threshold: L norm >0.8;

[0118] High packet loss rate threshold: P norm >0.05;

[0119] User connection number threshold: N>N threshold (such as N threshold =8);

[0120] Channel allocation process:

[0121] Due to L norm =0.9 (high latency), P norm = 0.06 (high packet loss rate), and N = 10 (exceeding the threshold), the system determines that high latency, high packet loss, and overload modes have occurred;

[0122] Increase bandwidth: Current bandwidth C old =50Mbps, increase ΔC = 10Mbps:

[0123] C new =C old +ΔC=50+10=60Mbps

[0124] Reduce the transmission rate: Current transmission speed Rold =20Mbps, reduced to:

[0125] R new =R old ×0.8=20×0.8=16Mbps

[0126] Limit the number of connections to reduce the current number of connections: N old =10, reduce ΔN=2:

[0127] N new =N old -ΔN=10-2=8

[0128] After making the above adjustments, continue to monitor the performance indicators. If in the next monitoring cycle:

[0129] L norm ≤0.5

[0130] P norm ≤0.02

[0131] N≤8

[0132] The adjustment is marked as successful.

[0133] S4. Optimizing the communication parameters after adaptive channel allocation and load balancing using the Ice algorithm, and reconfiguring the signal frequency and power of the transmitter according to the optimized communication parameters. It should be noted that optimizing the communication parameters using the Ice algorithm includes:

[0134] Set the initial signal frequency F r and power P w ;

[0135] Use the ice algorithm for search and optimization, the objective function is:

[0136] Z=w 1 ·L+w 2 ·P

[0137] Randomly generate solutions in the search space and evaluate their fitness;

[0138] Update the current best solution until the termination condition is met;

[0139] Output optimized communication parameters F′ r and P′ w :

[0140] F′ r =F r ×k

[0141] P′ w =P w ×m

[0142] Among them, k and m are optimization factors, w 1 、w 2 is the weight, reflecting the importance of network delay and packet loss rate, L is the network delay, and P is the packet loss rate.

[0143] As an example, in step S3, the adjusted parameters are:

[0144] F′ r =1.0(signal frequency),P′ w =5(transmitting power),C new =60Mbps,R new =16Mbps

[0145] Ice algorithm optimization process:

[0146] Setting the objective function: Assume that the weights of the delay L and packet loss rate P that we want to minimize are set to w 1 =0.7, w 2 =0.3:

[0147] Z=w 1 ·L+w 2 ·P

[0148] Initialization parameters: initial signal frequency F r =1.0 and power P w =5;

[0149] Run the Ice Algorithm: randomly generate new parameter combinations in the search space, such as F ′ r =1.1,P ′ w =5.5, and evaluate its fitness:

[0150] Assume that the calculated delay is L = 0.4ms and the packet loss rate is P = 0.01;

[0151] Calculate fitness:

[0152] Z=0.7·0.4+0.3·0.01=0.28+0.003=0.283

[0153] Update parameters: If the fitness of the new combination is better than the old combination, update it to the new parameters;

[0154] According to the optimization results, reconfigure the parameters of the signal transmitter:

[0155] F ′ r =1.1(new signal frequency),P ′ w=5.5(new transmission power)

[0156] Continue to monitor the optimized delay L ′ and packet loss rate P ′ :

[0157] If L ′ <0.5 and P ′ <0.02, marker optimization was successful and returned to the conventional monitoring mode;

[0158] If the criteria are not met, return to the Ice Algorithm optimization step to re-evaluate and adjust the parameters.

[0159] Preferably, in order to better verify the technical effect adopted in the present method, it is explained that this embodiment selects the traditional method and adopts the method of the present invention for comparative testing, and compares the test results by means of scientific demonstration to verify the real effect of the present method.

[0160] Traditional communication information processing methods mainly rely on centralized data processing architecture, uploading all communication data to cloud servers for unified processing. This method has the defects of high processing delay, large bandwidth consumption, and inability to respond in real time when faced with large-scale data. In addition, the accuracy of anomaly detection is low, making it difficult to detect network anomalies in a timely manner, resulting in affected network performance and user experience.

[0161] In order to verify that the method of the present invention has higher anomaly detection accuracy, faster response speed and better network performance than the traditional method, this embodiment will adopt the traditional centralized processing method and the communication information processing method based on edge computing provided by the embodiment of the present invention to perform real-time measurement and comparison of the network delay, packet loss rate and data throughput of the Internet of Things communication information system.

[0162] (1) Experimental conditions

[0163] Test environment: Build a simulated network containing 100 IoT devices, of which 50 devices generate normal communication data and 50 devices randomly generate abnormal communication data (such as high latency and packet loss);

[0164] Hardware configuration: The edge device uses an intelligent gateway with computing power, and the central server has the same configuration;

[0165] Test indicators: network delay (ms), packet loss rate (%), data throughput (Mbps), anomaly detection accuracy (%), response time (ms);

[0166] Test time: The test was run continuously for 24 hours, and the data of various indicators under the traditional method and the method of the present invention were recorded respectively;

[0167] (2) Experimental process

[0168] Traditional method:

[0169] Data collection: The communication data of all IoT devices are directly uploaded to the central server;

[0170] Data processing: The central server processes the collected data in a unified manner, including data cleaning and feature extraction;

[0171] Anomaly detection: A simple threshold judgment method is used to mark data with a delay exceeding 200ms or a packet loss rate exceeding 5% as abnormal;

[0172] Network optimization: Manual intervention to adjust network parameters based on abnormal markers, without adaptive mechanism;

[0173] The method of the present invention:

[0174] Data collection and preprocessing: Edge devices collect communication data in real time, including signal strength, network delay, packet loss rate, number of user connections, and data transmission rate. Moving average method is used to remove noise, standardize data, and generate feature vectors.

[0175] Anomaly detection: Input feature vectors into an analysis model built based on a decision tree algorithm, use information gain to perform node partitioning, and accurately identify abnormal patterns (high latency, high packet loss rate, overload);

[0176] Adaptive optimization: Based on the recognition results, the edge device automatically performs adaptive channel allocation and load balancing. When high latency is detected, the channel bandwidth is increased; when the packet loss rate is high, the retransmission mechanism is adjusted; when overloaded, the flow is limited to reduce the number of user connections;

[0177] Parameter optimization: Use the Ice algorithm to optimize the processed communication parameters, reconfigure the signal frequency and power of the transmitter, and achieve the optimal configuration of network performance.

[0178] Reference Figure 2 It can be seen intuitively that during the entire test cycle, the network delay of the traditional method is high and unstable, while the network delay of the method of the present invention is low and stable, which improves the real-time performance.

[0179] Reference Figure 3 The packet loss rate of the method of the present invention is stable at about 2%, which is significantly lower than the 6.5% to 7.2% of the traditional method. This shows that the method of the present invention has significantly improved the reliability of data transmission, reduced data loss, and ensured communication quality.

[0180] Reference Figure 4The throughput of the method of the present invention is always maintained above 109 Mbps, while the throughput of the traditional method is between 82 Mbps and 86 Mbps. The method of the present invention increases the throughput by about 30 Mbps, improves the data transmission efficiency, and can support a larger network load and data flow.

[0181] parameter Figure 5 The accuracy of the method of the present invention is between 94% and 96%, which is much higher than the 67% to 70% of the traditional method. Since the present invention adopts an intelligent analysis model based on a decision tree, it can identify anomalies more accurately and improve the efficiency and effect of network maintenance.

[0182] Reference Figure 6 The response time of the method of the present invention is stable at about 50ms, while the response time of the traditional method is more than 500ms. The method of the present invention realizes rapid response to network anomalies through edge computing and adaptive mechanism, significantly shortens the processing time, and improves the real-time performance of the network.

[0183] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A communication information processing method based on edge computing, characterized in that: include: Use edge devices to collect communication data in real time and preprocess the data to obtain feature vectors; Inputting the feature vector into an analysis model constructed based on a decision tree algorithm to identify abnormal patterns in the communication network, and outputting the identification result; Performing adaptive channel allocation and load balancing processing according to the identification result; The communication parameters processed by the adaptive channel allocation and load balancing are optimized using the Ice algorithm, and the signal frequency and power of the transmitter are reconfigured according to the optimized communication parameters.

2. The communication information processing method based on edge computing according to claim 1, characterized in that: The collected communication data at least includes signal strength, network delay, packet loss rate, number of user connections and data transmission rate.

3. The communication information processing method based on edge computing according to claim 1 or 2, characterized in that: The preprocessing removes noise by moving average method, standardizes all data to make them have the same dimension, converts them into proportional values, and outputs feature vectors.

4. The communication information processing method based on edge computing according to claim 3 is characterized in that: Based on the decision tree algorithm, the feature vector is used as input to build an analysis model, and then the information gain IG is used to divide the nodes. The mathematical expression formula is: Among them, v is a value of the feature attribute A, D v is the data subset when the feature attribute A takes the value v, D is the current data set, A is the feature attribute, and H(D) is the entropy of the data set D: C represents all categories, and p(c) represents the probability of category c.

5. The communication information processing method based on edge computing according to claim 4 is characterized in that: Apply a decision tree to the input feature vector, where: In the leaf nodes of the decision tree, each node represents an output category. When the feature vector is sent to the decision tree, the decision tree will make judgments on different paths based on the value of the feature vector, and finally reach the designated leaf node to obtain the classification result of the feature vector and output an abnormal identification, that is, the recognition result. The abnormal identification includes normal, high delay, high packet loss rate, and overload.

6. The communication information processing method based on edge computing according to claim 5 is characterized in that: According to the identification result, the current load is calculated, and adaptive channel allocation is performed according to the identified abnormal mode, wherein the abnormal mode includes a high delay mode, a high packet loss rate mode, and an overload mode.

7. The communication information processing method based on edge computing according to claim 6 is characterized in that: The adaptive channel allocation according to the identified abnormal pattern includes: When a high-latency mode is identified, the channel bandwidth is increased; When a high packet loss rate mode is identified, the retransmission mechanism is adjusted to reduce the transmission rate; When an overload condition is detected, the current is limited to reduce the number of user connections.

8. The communication information processing method based on edge computing according to claim 1, characterized in that: The communication parameters are optimized using the Ice algorithm, including: Set the initial signal frequency F r and power P w ; Use the ice algorithm for search and optimization, the objective function is: Z=w1·L+w2·P Randomly generate solutions in the search space and evaluate their fitness; Update the current best solution until the termination condition is met; Output optimized communication parameters F′ r and P′ w : F′ r =F r ×k P′ w =P w ×m Among them, k and m are optimization factors, w1 and w2 are weights that reflect the importance of network delay and packet loss rate, L is the network delay, and P is the packet loss rate.

Images