User satisfaction evaluation index method based on industrial 50GPON large connection

Abstract

The invention discloses an industrial 50GPON large connection-based user satisfaction evaluation method, which comprises the following steps of: establishing and configuring a data acquisition environment, and recording an access request service number of a wireless AP (Access Point); extracting request service wireless AP log information, and establishing a multi-granularity demand matrix access condition; classifying accessed wireless AP request services, and calculating and allocating bandwidths by using a multi-granularity bandwidth allocation algorithm; providing user satisfaction evaluation indexes based on reinforcement learning by using the mathematical model, and respectively evaluating the indexes; and establishing a large-connection user satisfaction evaluation function by adopting a linear weighted combination mode. The invention provides a large-connection user satisfaction evaluation index based on industrial 50GPON, maximization of instantaneous throughput and minimization of average weighted switching time are realized, and fairness constraint is satisfied at the same time. Therefore, the user satisfaction is effectively improved, the utilization efficiency of network resources is optimized, and the problem of waste of bandwidth resources is avoided.

Description

Technical Field

[0001] The present invention relates to the field of optical access network, and in particular to a user satisfaction evaluation index method based on industrial 50GPON large connection. Background Art

[0002] With the rapid development of optical communication technology, major industrial manufacturing entities around the world have proposed a new generation of industrial development strategies. Data communication systems play an important role in connecting the upper and lower levels of the intelligent manufacturing system. Passive optical networks have become a promising solution for providing high-speed broadband services to end users. Among various PON standards, gigabit passive optical networks stand out for their cost-effectiveness and high efficiency. In particular, 50GPON, as an evolution of the GPON standard, promises to provide higher bandwidth capacity to meet the growing demand for bandwidth-intensive applications and become an important foundation for realizing Industry 4.0 and smart factory architecture.

[0003] The goal of Industry 4.0 is to connect and integrate traditional industries, especially manufacturing, to achieve flexibility, adaptability and efficiency, and to increase effective communication between producers and consumers. In addition, Industry 4.0 refers to the cooperation between different factories. Therefore, communication and network play an important role in Industry 4.0. 50GPON technology has the characteristics of anti-interference and dense coverage, and is an important development direction of the Industrial Internet.

[0004] At present, in the relevant technologies of large-connection user satisfaction evaluation indicators, the ultimate success measure for any telecommunications network still lies in the satisfaction of a single indicator of its end users, without considering the background of smart factories under Industry 4.0. Faced with complex and diverse industrial 50GPON specific scenarios, network traffic characteristics and services of different scenario applications are different. The challenges that need to be solved when integrating 50GPON into industrial networks include the conflicting requirements between accommodating large-scale connections and maintaining low latency. In traditional architectures, adding access points often leads to higher latency, which poses an important challenge to meeting the strict latency requirements of industrial applications, resulting in traditional PON (passive optical network) evaluation indicators being insufficient to evaluate the performance of 50GPON networks. Summary of the invention

[0005] In view of the above-mentioned defects of the prior art, the technical problem to be solved by the present invention is that the existing PON (passive optical network) evaluation index is not based on the actual needs of users and does not take into account the complex and diverse industrial 50GPON specific scenarios, which makes it insufficient to evaluate the performance of the 50GPON network, resulting in the inability to accommodate the conflicting requirements between large-scale connections and maintaining low latency, resulting in a decrease in user satisfaction, and there are problems such as low network efficiency and poor user experience. Therefore, the present invention provides a method for evaluating user satisfaction indicators based on industrial 50GPON large connections, while considering multiple factors in complex and changeable scenarios, using a weighted combination method with utility weights to balance the trade-offs between individual indicators, and through a reasonable design of a user satisfaction function, effectively improving user satisfaction and optimizing the utilization efficiency of network resources.

[0006] To achieve the above object, the present invention provides a method for evaluating user satisfaction index based on industrial 50GPON large connection, comprising the following steps:

[0007] Build and configure the data collection environment to record the number of access request services of wireless APs;

[0008] Extract the log information of the eigenvalue parameters, establish a multi-granularity bandwidth requirement matrix, and use the log information of the eigenvalue parameters as the elements of the matrix; further standardize the multi-granularity bandwidth requirement matrix to obtain the corrected wireless AP bandwidth requirement;

[0009] Classify the corrected wireless AP bandwidth requirements and put them into corresponding buffer pools according to different categories;

[0010] The multi-granularity bandwidth allocation algorithm is used to calculate the remaining bandwidth after the previous round of allocation, and after the calculation is completed, bandwidth is allocated to the newly connected wireless AP in the buffer pool;

[0011] Propose a user satisfaction evaluation index based on reinforcement learning;

[0012] Define the user satisfaction evaluation indicators based on reinforcement learning respectively, and calculate the weight coefficient of each indicator after definition;

[0013] A linear weighted combination method is introduced to perform a linear weighted combination of each evaluation indicator and the corresponding weight coefficient, and a large-connection user satisfaction evaluation function is established to achieve a comprehensive evaluation of the user satisfaction based on industrial 50GPON large connections.

[0014] Furthermore, a data collection environment is built and configured to record the number of access request services of wireless APs, including:

[0015] Use OPNET to build a large-connection industrial simulation system model, extract log information containing characteristic value parameters of each wireless AP service type, wavelength, time slot and subcarrier; standardize the extracted characteristic value parameter log information to obtain the corrected request service;

[0016] The wireless AP request services that have been corrected by the standardization process are classified and put into corresponding buffer pools according to different categories of request services.

[0017] Furthermore, the characteristic value parameters include each wireless AP service type, wavelength, time slot and subcarrier.

[0018] Furthermore, the log information of the eigenvalue parameters is extracted to establish a multi-granularity bandwidth demand matrix, and the log information of the eigenvalue parameters is used as the element of the matrix; the log information of the wireless AP service is extracted to establish the access conditions of the multi-granularity demand matrix, and the mathematical model is used to establish the AP i Connection requirements ij :

[0019] d ij =[d i1 d i2 d i3 d i4 ]

[0020] where d ij (j=1, 2, 3, 4) are the bandwidth requirements of 50GPONi for service type, wavelength, time slot and subcarrier respectively;

[0021] In order to avoid frequent bandwidth reconnection allocation, two thresholds (i.e., low threshold and high threshold η1) η2, and a connection demand change matrix are defined, which is shown as follows:

[0022] a ij =[a i1 a i2 a i3 a i4 ]

[0023] where a ij Indicates d ij If the change of d ij If the ratio of the previous round is lower than η1, then a ij will be set to -1, which means that d ij There is a large bandwidth reduction; if d ij If the ratio of a to its previous round is higher than η2, then a ij will be set to 1, which means that d ij With a larger bandwidth increase, otherwise, we think d ij There is no large change in bandwidth demand, i.e. aij will be set to 0.

[0024] Furthermore, the multi-granularity bandwidth requirement matrix is ​​further standardized to obtain the corrected wireless AP bandwidth requirement. Specifically, the "maximum-minimum" standard deviation method is used to further standardize the evaluation indicators with different variation ranges:

[0025]

[0026] Among them, r ij Represents the bandwidth demand utility description function value of the i-th indicator of the j-th wireless AP.

[0027] Furthermore, the corrected wireless AP bandwidth requirements are classified and placed into corresponding buffer pools according to different categories. After completing the standardization of the demand matrix, the characteristic values ​​such as the number of requested services and flow data of the wireless AP are classified by using the method of calculating the sum of the sub-matrices, and the classified wireless APs are placed into corresponding buffer pools. Specifically, the connection demand matrix a is used to classify the wireless AP bandwidth requirements. ij The value of d ij It can be divided into 3 parts: EF, AF and BF, demand matrix d ij can be interpreted as the sum of the following three sub-matrices:

[0028]

[0029] Each sub-matrix corresponds to EF, AF and BF respectively.

[0030] Furthermore, the multi-granularity bandwidth allocation algorithm is used to calculate the remaining bandwidth allocated in the previous round, and after the calculation is completed, bandwidth is allocated to the newly connected wireless AP in the buffer pool, specifically including:

[0031] Two thresholds, namely, the low threshold η1 and the high threshold η2, are defined to avoid frequent bandwidth reconnection allocation;

[0032] According to the bandwidth information to be allocated by the newly connected wireless AP in the buffer pool, a connection demand change matrix a is defined. i ;

[0033] According to the defined wireless AP connection demand change matrix a i , define a variable v i To indicate wireless AP i Whether bandwidth reallocation is required;

[0034] According to the variable v i , define a matrix v that represents the bandwidth allocation of all wireless APs.

[0035] Furthermore, user satisfaction evaluation indicators based on reinforcement learning are defined respectively, and the weight coefficient of each indicator is calculated after the definition; specifically, a user satisfaction evaluation indicator based on reinforcement learning is proposed by using a mathematical model, and the user satisfaction evaluation indicator based on reinforcement learning includes defining calculation formulas for instantaneous throughput, average weighted switching time and fairness constraints respectively, and calculating the weight coefficient of each indicator after the definition.

[0036] Furthermore, a linear weighted combination method is introduced to perform a linear weighted combination of each evaluation index and the corresponding weight coefficient, and a large-connection user satisfaction evaluation function is established to achieve a comprehensive evaluation of the user satisfaction based on industrial 50GPON large connections. The expression of the user satisfaction evaluation function is:

[0037] C=e1·max(US mt )+e2·min(US switch_t )+e3·exp(Z).

[0038] Among them, US mt is the instantaneous throughput satisfaction, US swi_t is the switching time satisfaction, Z is the network fairness satisfaction, e1, e2, e2 are the utility weights of each indicator respectively, and the reward strategy is to maximize the instantaneous throughput, minimize the user switching time, and satisfy the user fairness.

[0039] Technical Effects

[0040] The present invention provides a method for evaluating the user satisfaction index based on industrial 50GPON large connections. Before evaluating the user satisfaction, the demand information of each wireless AP is extracted by using a reinforcement learning reward mechanism model, and the log information of the characteristic value parameters is extracted to perform multi-granularity demand matrix modeling on the demand of the wireless AP. The calculation formulas of the instantaneous throughput, the average weighted switching time and the fairness constraint in the user satisfaction evaluation index based on reinforcement learning are defined respectively, and the weight coefficient of each index is calculated. By introducing a linear weighted combination method, each evaluation index and the corresponding weight coefficient are linearly weighted combined, and a large connection user satisfaction evaluation function is established to achieve a comprehensive evaluation of the satisfaction of the accessed wireless AP, so as to fully utilize the bandwidth and improve the bandwidth utilization rate.

[0041] The concept, specific structure and technical effects of the present invention will be further described below in conjunction with the accompanying drawings to fully understand the purpose, characteristics and effects of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] Figure 1 It is a flow chart of a method for evaluating the user satisfaction index of industrial 50GPON large connection based on a preferred embodiment of the present invention;

[0043] Figure 2 The present invention is a schematic diagram of a multi-granularity bandwidth allocation algorithm process based on an industrial 50GPON large-connection user satisfaction evaluation index method according to a preferred embodiment of the present invention. DETAILED DESCRIPTION

[0044] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0045] In the following description, specific details such as specific internal procedures and techniques are provided for the purpose of illustration rather than limitation, so as to provide a thorough understanding of the embodiments of the present invention. However, it should be clear to those skilled in the art that the present invention may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to prevent unnecessary details from obstructing the description of the present invention.

[0046] like Figure 1 As shown, the embodiment of the present invention provides a method for evaluating user satisfaction index based on industrial 50GPON large connection, which specifically includes the following steps:

[0047] Step S101, build and configure a data collection environment, record the number of access request services of wireless APs; and include log information of characteristic value parameters such as service type, wavelength, time slot and subcarrier of each wireless AP.

[0048] Step S102, extracting the log information of the eigenvalue parameters, establishing a multi-granularity bandwidth requirement matrix, and using the log information of the eigenvalue parameters as elements of the matrix; further standardizing the multi-granularity bandwidth requirement matrix to obtain a corrected wireless AP bandwidth requirement.

[0049] The characteristic value parameters include each wireless AP service type, wavelength, time slot and subcarrier.

[0050] During network operation, as user demand changes, the service connection bandwidth demand will also change, which will lead to service reconnection allocation. However, frequent service reconnection allocation is undesirable for any 50GPON. Therefore, a reasonable 50GPON service connection satisfaction access condition is necessary, which can filter out some unnecessary bandwidth reallocation requirements. Therefore, in this embodiment, the demand matrix modeling criterion is used to represent the AP bandwidth demand. i Connection requirements ij :

[0051] dij =[d i1 d i2 d i3 d i4 ]

[0052] where d ij (j=1,2,3,4) are 50GPON respectively i Bandwidth requirements for service type, wavelength, time slot and subcarrier.

[0053] Since the variation range of each evaluation index is different, it is necessary to further standardize the multi-granularity demand matrix to obtain the corrected wireless AP bandwidth demand. Specifically, the "maximum-minimum" standard deviation method is used to further standardize the individual evaluation indexes with different variation ranges:

[0054]

[0055] Among them, r ij Represents the bandwidth demand utility description function value of the i-th indicator of the j-th wireless AP.

[0056] Step S103, classify the corrected wireless AP bandwidth requirements and put them into corresponding buffer pools according to different categories.

[0057] The method of this embodiment fully considers the characteristics of the network services of the intelligent development of industrial 4.0 power plants, and divides the services transmitted by ONU into three categories: accelerated forwarding (EF) services, guaranteed forwarding (AF) services, and best effort (BE) services; in order to distinguish the changes in the connection requirements of different levels of services, according to a ij The value of d i It can be divided into three parts: EF (Expedited-Forwarding Part, EFP), AF (Assured-Forwarding Part, AFP) and BF (Best efficient-Forwarding Part, BFP). i can be interpreted as the sum of the following three sub-matrices:

[0058]

[0059] Each sub-matrix corresponds to EF, AF and BF respectively.

[0060] The specific operations of putting the classified wireless APs into the corresponding buffer pool are as follows:

[0061] Request classification: First, the requests received by the wireless AP need to be classified. This involves identifying the type of request, priority, target device, and other information;

[0062] Buffer pool management: The classified requests are placed into the corresponding buffer pool as needed. The management of the buffer pool involves the storage, scheduling and processing of requests;

[0063] Performance optimization: The management of the buffer pool takes performance optimization factors into consideration. By adjusting the size of the buffer pool and using appropriate scheduling algorithms, system performance can be optimized to reduce the waiting time of requests and improve the response speed of the system.

[0064] Step S104, using a multi-granularity bandwidth allocation algorithm to calculate the remaining bandwidth allocated in the previous round, and after the calculation is completed, allocate bandwidth to the newly connected wireless AP in the buffer pool; the algorithm flow is as follows: Figure 2 As shown, the specific process is as follows:

[0065] First, two thresholds (i.e., low threshold ηη1 and high threshold ηη2) are defined to avoid frequent bandwidth reconnection allocation;

[0066] According to the wireless AP requirements for bandwidth information to be allocated, a connection demand change matrix is ​​defined; this matrix is ​​as follows:

[0067] a i =[a i1 a i2 a i3 a i4 ]

[0068] where a ij Indicates d ij If d ij If the ratio of its previous round is lower than η1, then a ij will be set to -1. This means that d ij There is a large bandwidth reduction. If d ij If the ratio of a to its previous round is higher than η2, then a ij will be set to 1. This means that d ij There is a large bandwidth increase. Otherwise, it is considered that d ij There is no large change in bandwidth demand, i.e. a ij will be set to 0;

[0069] According to the defined wireless AP connection demand change matrix, define a variable v i To indicate wireless AP i Whether bandwidth needs to be reallocated. Variable v i The values ​​are as follows:

[0070]

[0071] where v i =1 means wireless AP iThe bandwidth needs to be reallocated. According to matrix a i The modulus value indicates that the wireless AP i The bandwidth has changed significantly;

[0072] According to the variable v i , define a matrix representing the bandwidth allocation of all wireless APs v .

[0073] v=[v1v2...v q ]

[0074] where v i (1≤i≤q) is defined above, q is the total number of wireless AP service connections. When the modulus of the matrix v is 0, no bandwidth reallocation will be performed. Otherwise, bandwidth allocation will be performed and the bandwidth will be reallocated to v i The wireless AP whose value is equal to 1. At this point, the wireless APs that need bandwidth allocation can be filtered out.

[0075] Step S105: Propose a user satisfaction evaluation index based on reinforcement learning.

[0076] Step S106, respectively defining user satisfaction evaluation indicators based on reinforcement learning, and calculating the weight coefficient of each indicator after definition;

[0077] In steps S105-S106, a user satisfaction evaluation index based on reinforcement learning is proposed using a mathematical model to maximize instantaneous throughput, minimize average weighted switching time, and meet fairness constraints. Specifically, the following steps are included:

[0078] In order to maximize the use of time resources, the instantaneous throughput should be considered in the scheduling scheme. The instantaneous throughput calculation formula F is defined as:

[0079]

[0080] Among them, n x represents the modulation bandwidth, U x It indicates the total number of users accessing at time T, in units of per second, R is the number of requests sent by each user, MAG j，x Indicates connection to AP x The MAG received by user j, T is the time taken to process the request;

[0081] Here, user satisfaction will be quantified into three dimensions: instantaneous throughput satisfaction (US mt ), switching time satisfaction (US switch_t ) and user fairness satisfaction (US u_fair ). For US mt , which can be expressed as:

[0082]

[0083] This paper studies the mobile switching problem, which is described as the problem of finding the optimal network communication delay between the starting point and the end point of the mobile node's trajectory. According to the traditional mobile network switching strategy, the Nth switching time of the mobile node during its travel within T time can be expressed as:

[0084]

[0085] Where N is the total number of network switches of the mobile node during the entire trajectory movement, t a , t b , t c , t d They are the clustering subnet scanning, switching selection, switching decision, and switching execution time of the mobile node during the Nth switching process;

[0086] For US switch_t , which can be expressed as:

[0087]

[0088] There are many criteria for judging whether a resource scheduling algorithm is fair. This invention adopts a linear weighted combination method based on utility weights to transform the above two objectives into a single optimization objective. Based on this, the following evaluation function is established:

[0089] Z=α*US mt +1 / β*US switc_t

[0090] Since the satisfaction ranges from 0 to 1, and the value of instantaneous throughput is between 30-200Mbps, and the switching time is not less than 10μm, the proportional factors α and 1 / β are added to increase the satisfaction weight, which satisfies

[0091] α+β=1; the evaluation function Z is mainly used to measure the long-term fairness of the system. The larger the F value, the more evenly the resources allocated to users are distributed, the better the fairness to users, and the higher the user satisfaction.

[0092] Step S107, introduce a linear weighted combination method, perform a linear weighted combination on each evaluation index and the corresponding weight coefficient, establish a large-connection user satisfaction evaluation function, and achieve a comprehensive evaluation of the industrial 50GPON large-connection user satisfaction. The details are as follows:

[0093] In view of user satisfaction, the embodiment of the present invention uses a reinforcement learning model. Reinforcement learning is also called evaluation learning or enhanced learning. It is used to solve the problem of maximizing the return of intelligent agents. The embodiment of the present invention defines the following parameters:

[0094] Agent: User satisfaction is considered as an agent, whose goal is to perform the allocation of user access points;

[0095] Strategy: Learning strategies through interaction with the environment;

[0096] Reward function: maximize instantaneous throughput, minimize user switching time, and meet user fairness;

[0097] The reinforcement learning model looks like this:

[0098] R=max(US mt )min(US switch_t )exp(Z)

[0099] When designing a reward function, it is necessary to comprehensively consider various factors and use a weighted combination method that assigns utility weight coefficients to balance the trade-offs between different objectives. By reasonably designing the reward function, reinforcement learning can effectively improve user satisfaction and optimize the utilization efficiency of network resources;

[0100] S i Represents the standard deviation of the i-th evaluation index in different wireless APs:

[0101]

[0102] The utility weight formula of the present invention is: i as follows:

[0103]

[0104] So far, the final expression of the utility function proposed by the present invention can be given:

[0105] C=e1·max(US mt )+e2·min(US switch_t )+e3·exp(Z).

[0106] The following is an example of a mobile network provider using an industrial 50 GPON large connection user satisfaction evaluation index method to calculate user satisfaction according to an embodiment of the present invention. Assuming that the instantaneous throughput value of the user within 60 seconds is 100 Mbps, the switching time is 25 μm, and the proportional factor α=0.7 and 1 / β=0.3 are specified, then substitute into the evaluation function formula:

[0107] Z=α*US mt +1 / β*US switch_t

[0108] Obtain user fairness: Z = 0.7*100+0.3*25

[0109] =70+7.5

[0110] =77.5

[0111] Put US mt =100, US swi_t =25, Z=77.5 and substitute into the reinforcement learning model formula:

[0112] R=max(US mt )min(US switch_t )exp(Z)

[0113] Obtain the reinforcement learning model: R = max(100)min(25)exp(77.5);

[0114] Assume that the standard deviations of instantaneous throughput, switching time, and user fairness are: S1 = 0.78, S2 = 0.56, S3 = 0.85:

[0115] Then substitute S1=0.78, S2=0.56, S3=0.85 into the utility weight formula:

[0116]

[0117] The utility weights are: e1 = 0.35, e2 = 0.25, e3 = 0.38;

[0118] Finally, substitute e1=0.35, e2=0.25, e3=0.38 into the final expression of the utility function proposed by the present invention:

[0119] C=e1·max(US mt )+e2·min(US switch_t )+e3·exp(Z)

[0120] Calculate user satisfaction: C = 0.35*100+0.25*25+0.38*77.5

[0121] =35+6.25+29.45

[0122] =70.7

[0123] Therefore, within 60 seconds, when the user's instantaneous throughput value is 100Mbps, the switching time is 25μm, the proportional factor α=0.7, 1 / β=0.3, the standard deviations of instantaneous throughput, switching time, and user fairness are S1=0.78, S2=0.56, and S3=0.85 respectively, the user satisfaction calculated based on the provided parameters and utility function is 70.7. This score can help mobile network providers evaluate the quality of their network services and take measures to optimize user experience.

[0124] The present invention provides a method for evaluating the user satisfaction index based on industrial 50GPON large connections. Before evaluating the user satisfaction, the demand information of each wireless AP is extracted by using a reinforcement learning reward mechanism model, and the log information of the characteristic value parameters is extracted to establish a multi-granularity demand matrix for the demand of the wireless AP, and the calculation formulas of the instantaneous throughput, average weighted switching time and fairness constraint in the user satisfaction evaluation index based on reinforcement learning are defined respectively, and the weight coefficient of each index is calculated. By introducing a linear weighted combination method, each evaluation index and the corresponding weight coefficient are linearly weighted combined, and a large-connection user satisfaction evaluation function based on a utility function is established, so as to achieve a comprehensive evaluation of the satisfaction of the accessed wireless AP, so as to fully utilize the bandwidth and improve the bandwidth utilization rate.

[0125] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A method for evaluating user satisfaction based on industrial 50GPON large connections, characterized in that: The following steps are involved: Build and configure the data collection environment to record the number of access request services of wireless APs; Extract the log information of the eigenvalue parameters and establish a multi-granularity bandwidth demand matrix, with the log information of the eigenvalue parameters as the elements of the matrix; The multi-granularity bandwidth requirement matrix is ​​further standardized to obtain the corrected wireless AP bandwidth requirement; Classify the corrected wireless AP bandwidth requirements and put them into corresponding buffer pools according to different categories; The multi-granularity bandwidth allocation algorithm is used to calculate the remaining bandwidth after the previous round of allocation, and after the calculation is completed, bandwidth is allocated to the newly connected wireless AP in the buffer pool; Propose a user satisfaction evaluation index based on reinforcement learning; Define the user satisfaction evaluation indicators based on reinforcement learning respectively, and calculate the weight coefficient of each indicator after definition; A linear weighted combination method is introduced to perform a linear weighted combination of each evaluation indicator and the corresponding weight coefficient, and a large-connection user satisfaction evaluation function is established to achieve a comprehensive evaluation of the user satisfaction based on industrial 50GPON large connections.

2. A method for evaluating user satisfaction based on industrial 50GPON large connections as claimed in claim 1, characterized in that: Build and configure the data collection environment to record the number of access request services of wireless APs, including: Use OPNET to build a large-connection industrial simulation system model, extract log information containing characteristic value parameters of each wireless AP service type, wavelength, time slot and subcarrier; standardize the extracted characteristic value parameter log information to obtain the corrected request service; The wireless AP request services that have been corrected by the standardization process are classified and put into corresponding buffer pools according to different categories of request services.

3. A method for evaluating user satisfaction based on industrial 50GPON large connections as claimed in claim 1, characterized in that: The characteristic value parameters include each wireless AP service type, wavelength, time slot and subcarrier.

4. A method for evaluating user satisfaction based on industrial 50GPON large connections as claimed in claim 1, characterized in that: Extract the log information of the eigenvalue parameters and establish a multi-granularity bandwidth demand matrix, with the log information of the eigenvalue parameters as the elements of the matrix; It is to extract the log information of wireless AP business, establish the access conditions of multi-granularity demand matrix, and use mathematical models to establish AP i Connection requirements ij : d ij =[d i1 d i2 d i3 d i4 ] where d ij (j=1, 2, 3, 4) are the bandwidth requirements of 50GPONi for service type, wavelength, time slot and subcarrier respectively; In order to avoid frequent bandwidth re-allocation, two thresholds, namely the low threshold η1 and the high threshold η2, and a connection demand change matrix are defined, which is shown as follows: a ij =[a i1 a i2 a i3 a i4 ] where a ij Indicates d ij If the change of d ij If the ratio of its previous round is lower than η1, then a ij will be set to -1, which means that d ij There is a large bandwidth reduction; if d ij If the ratio of a to its previous round is higher than η2, then a ij will be set to 1, which means that d ij There is a large bandwidth increase; otherwise, it is considered that d ij There is no large change in bandwidth demand, i.e. a ij will be set to 0.

5. A method for evaluating user satisfaction based on industrial 50GPON large connections as claimed in claim 1, characterized in that: The multi-granularity bandwidth requirement matrix is ​​further standardized to obtain the corrected wireless AP bandwidth requirement. Specifically, the "maximum-minimum" standard deviation method is used to further standardize the evaluation indicators with different variation ranges: Among them, r ij Represents the bandwidth demand utility description function value of the i-th indicator of the j-th wireless AP.

6. A method for evaluating user satisfaction based on industrial 50GPON large connections as claimed in claim 4, characterized in that: The corrected wireless AP bandwidth requirements are classified and placed into corresponding buffer pools according to different categories. After completing the standardization of the demand matrix, the characteristic values ​​such as the number of requested services and traffic data of the wireless AP are classified by calculating the sum of the sub-matrices, and the classified wireless APs are placed into corresponding buffer pools. Specifically, the connection demand change matrix a ij The value of d ij It can be divided into 3 parts: EF, AF and BF, demand matrix d ij can be interpreted as the sum of the following three sub-matrices: Each sub-matrix corresponds to EF, AF and BF respectively.

7. A method for evaluating user satisfaction based on industrial 50GPON large connections as claimed in claim 4, characterized in that: The multi-granularity bandwidth allocation algorithm is used to calculate the remaining bandwidth allocated in the previous round, and after the calculation is completed, bandwidth is allocated to the newly connected wireless AP in the buffer pool, including: Two thresholds, namely, the low threshold η1 and the high threshold η2, are defined to avoid frequent bandwidth reconnection allocation; According to the bandwidth information to be allocated by the newly connected wireless AP in the buffer pool, a connection demand change matrix a is defined. i ; According to the defined wireless AP connection demand change matrix a i , define a variable v i To indicate wireless AP i Whether bandwidth reallocation is required; According to the variable v i , define a matrix v that represents the bandwidth allocation of all wireless APs.

8. The method for evaluating the user satisfaction index based on industrial 50GPON large connection according to claim 1, characterized in that: The user satisfaction evaluation index based on reinforcement learning is defined respectively, and the weight coefficient of each index is calculated after the definition; specifically, a user satisfaction evaluation index based on reinforcement learning is proposed by using a mathematical model, and the user satisfaction evaluation index based on reinforcement learning includes respectively defining calculation formulas for instantaneous throughput, average weighted switching time and fairness constraints, and calculating the weight coefficient of each index after the definition.

9. A method for evaluating user satisfaction based on industrial 50GPON large connections as claimed in claim 1, characterized in that: A linear weighted combination method is introduced to perform a linear weighted combination of each evaluation index and the corresponding weight coefficient, and a large-connection user satisfaction evaluation function is established to achieve a comprehensive evaluation of the user satisfaction based on industrial 50GPON large connections. The expression of the user satisfaction evaluation function is: C=e1·max(US mt )+e2·min(US switch_t )+e3·exp(Z) Among them, US mt is the instantaneous throughput satisfaction, US switc_t is the switching time satisfaction, Z is the network fairness satisfaction, e1, e2, e2 are the utility weights of each indicator respectively, and the reward strategy is to maximize the instantaneous throughput, minimize the user switching time, and satisfy the user fairness.

Images