Bandwidth estimation device, bandwidth estimation method, and program
The bandwidth estimation device improves accuracy by classifying traffic by service and using machine learning to estimate required bandwidth, addressing inaccuracies in existing methods and reducing equipment costs.
Patent Information
- Application Number
- JP2023522108
- Authority / Receiving Office
- JP · JP
- Patent Type
- Patents
- Current Assignee / Owner
- Filing Date
- 2021-05-20
- Publication Date
- 2024-11-20
- Estimated Expiration
- 2041-05-20
AI Technical Summary
Existing bandwidth estimation methods in communication networks fail to accurately distinguish between various services with different traffic patterns, leading to inaccuracies in estimating required bandwidth, which can result in increased equipment costs.
A bandwidth estimation device that classifies traffic information by service using tags and DPI technology, estimates required bandwidth for each service based on contract bandwidth information, and utilizes machine learning to improve accuracy by considering user bandwidth utilization rates and traffic periodicity.
Enhances the accuracy of bandwidth estimation by service, allowing for precise determination of equipment needs and reducing costs by minimizing data collection and processing time.
Smart Images

Figure 0007587190000001 
Figure 0007587190000002 
Figure 0007587190000003
Abstract
Description
[Technical field]
[0001] The present disclosure relates to a bandwidth estimation device, a bandwidth estimation method, and a program. [Background technology]
[0002] When designing a communication network in which a link between communication devices accommodates multiple lines, information on the bandwidth required by the link (hereinafter referred to as "required bandwidth") is required. The communication bandwidth of each line has an upper limit of the bandwidth according to the contract for each line (hereinafter referred to as "contracted bandwidth"), but the number of lines accommodated in the link and the contracted bandwidth change over time as lines are newly added, contracts are changed, or lines are deleted. Therefore, in designing a communication network, it is important to calculate the required bandwidth taking into account contracted bandwidth information, which is information on the contracted bandwidth of each line. A network designer estimates the required bandwidth and judges whether a new line can be accommodated (accommodation judgment) depending on whether the required bandwidth exceeds the bandwidth of the existing equipment. If the required bandwidth exceeds the bandwidth of the existing equipment, an additional facility will be required. Therefore, estimating the required bandwidth with high accuracy leads to suppression of equipment costs.
[0003] As a method for estimating the required bandwidth, a method has been reported in which the bandwidth is estimated based on information on the traffic of each line flowing through a link, obtained from a communication device. For example, Patent Document 1 describes a method for calculating the future required bandwidth from the fluctuation of the bandwidth of existing lines. [Prior art documents] [Patent documents]
[0004] [Patent Document 1] JP 2009-118274 A Summary of the Invention [Problem to be solved by the invention]
[0005] In communication networks to which many users' devices are connected, traffic caused by various services such as mobile (video images, SNS (Social Networking Service)), IoT (Internet of Things), and mobile devices flows, and in recent years, the traffic has become increasingly diversified. In the method described in Patent Document 1, traffic information is acquired and analyzed without distinguishing between various services. However, collectively analyzing services that differ in various factors such as communication bandwidth usage, trends in increases and decreases in the number of users, and daily and weekly fluctuations may lead to a decrease in the accuracy of estimating the required bandwidth. Thus, there is room for improvement in the accuracy of estimating the required bandwidth.
[0006] In view of the above problems, an object of the present disclosure is to provide a bandwidth estimation device, a bandwidth estimation method, and a program that can improve the accuracy of estimating a required bandwidth. [Means for solving the problem]
[0007] In order to solve the above problems, the bandwidth estimation device disclosed herein is a bandwidth estimation device that estimates the bandwidth required for a link between communication devices, which accommodates lines of multiple users and through which traffic resulting from multiple services flows, and includes: a classification unit that acquires traffic information regarding the traffic from the communication devices and classifies the acquired traffic information by service; and an estimation unit that estimates the required bandwidth for each service based on the traffic information for each service and contracted bandwidth information regarding the contracted bandwidth of each of the multiple users.
[0008] In addition, in order to solve the above-mentioned problems, the bandwidth estimation method disclosed herein is a bandwidth estimation method that estimates the bandwidth required for a link between communication devices that accommodates lines of multiple users and through which traffic resulting from multiple services flows, and obtains traffic information regarding the traffic from the communication devices, classifies the obtained traffic information by service, and estimates the required bandwidth for each service based on the traffic information for each service and contracted bandwidth information regarding the contracted bandwidth of each of the multiple users.
[0009] In order to solve the above problem, a program according to the present disclosure causes a computer to function as the above-mentioned bandwidth estimation device. Effect of the Invention
[0010] According to the bandwidth estimation device, the bandwidth estimation method, and the program of the present disclosure, it is possible to improve the accuracy of estimating the required bandwidth. [Brief description of the drawings]
[0011] [Figure 1] 1 is a diagram illustrating an example of the configuration of a bandwidth estimation device according to a first embodiment of the present disclosure. [Figure 2A] 2 is a diagram for explaining estimation of a required bandwidth for each service by the estimation unit shown in FIG. 1. FIG. [Figure 2B] 2 is a diagram for explaining estimation of a required bandwidth for each service by the estimation unit shown in FIG. 1. FIG. [Figure 2C] 2 is a diagram for explaining estimation of a required bandwidth for each service by the estimation unit shown in FIG. 1. FIG. [Diagram 3] 2 is a flowchart showing an example of the operation of the bandwidth estimation device shown in FIG. [Figure 4] FIG. 11 is a diagram illustrating an example of the configuration of a bandwidth estimation device according to a second embodiment of the present disclosure. [Diagram 5] 5 is a diagram showing an example of learning data input to a learning unit shown in FIG. 4; [Figure 6] 5 is a diagram showing an example of an estimation result of the estimation unit shown in FIG. 4. [Figure 7] FIG. 13 is a diagram illustrating an example of the configuration of a bandwidth estimation device according to a third embodiment of the present disclosure. [Figure 8] 8 is a diagram for explaining extraction of traffic periodicity by a periodicity extraction unit shown in FIG. 7. FIG. [Figure 9] 8 is a diagram showing an example of learning data input to a learning unit shown in FIG. 7; FIG. [Figure 10] FIG. 11 is a diagram illustrating an example of the configuration of a bandwidth estimation device according to a fourth embodiment of the present disclosure. [Figure 11] 11 is a diagram for explaining a determination of the presence or absence of a periodic feature by a feature determining unit shown in FIG. 10. FIG. [Figure 12] FIG. 11 is a diagram illustrating an example of the configuration of a bandwidth estimation device according to a fourth embodiment of the present disclosure. [Figure 13] 2 illustrates an example of a hardware configuration of the bandwidth estimation device illustrated in FIG. 1. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
[0012] Hereinafter, embodiments of the present disclosure will be described with reference to the drawings.
[0013] (First embodiment) 1 is a diagram illustrating a configuration example of a bandwidth estimation device 10 according to a first embodiment of the present disclosure. The bandwidth estimation device 10 according to the present disclosure estimates a required bandwidth of a link 21 between communication devices 20, which accommodates lines of multiple users and through which traffic resulting from multiple services flows.
[0014] As shown in FIG. 1, a bandwidth estimation device 10 according to this embodiment includes a service classification unit 11, a traffic collection unit 12, a user information DB (Database) 13, and an estimation unit 14.
[0015] The service classification unit 11 acquires traffic information, which is information about traffic (bandwidth) flowing through the link 21, from the communication device 20. The service classification unit 11 classifies the acquired traffic information by service. That is, the service classification unit 11 classifies the traffic information of the traffic caused by each service by each service. For example, when designing a network, services are classified in advance by tags, V-LAN (Virtual Local Area Network), priority, SID (Segment ID), etc., so that the service classification unit 11 can classify the traffic information by service based on the classification information. In addition, the service classification unit 11 can classify the traffic information by service by reading the contents of packets and detecting the service contents by, for example, DPI (Deep Packet Inspection) technology. The service classification unit 11 outputs the traffic information classified by service to the traffic collection unit 12.
[0016] The traffic collection unit 12 collects traffic information for each service output from the service classification unit 11, and outputs the information to the estimation unit .
[0017] The user information DB 13 is a database that stores information on the contract bandwidth, which is the upper limit of the bandwidth according to the contract of each user's line (hereinafter referred to as "contract bandwidth information"). The user information DB 13 outputs the stored contract bandwidth information to the estimation unit 14.
[0018] The estimation unit 14 estimates the required bandwidth for each service based on the traffic information for each service output from the traffic collection unit 12 and the contracted bandwidth information output from the user information DB 13 .
[0019] The estimation of the required bandwidth for each service by the estimation unit 14 will be described with reference to Figures 2A to 2C. The estimation unit 14 performs the process described with reference to Figures 2A to 2C for each service.
[0020] The estimation unit 14 extracts, for each contracted bandwidth, traffic information of traffic caused by a service (hereinafter referred to as a "target service") for which a required bandwidth is to be estimated. That is, the estimation unit 14 extracts traffic information of traffic caused by a target service for each contracted bandwidth (contracted bandwidth A, contracted bandwidth B, contracted bandwidth C...). Next, as shown in FIG. 2A, the estimation unit 14 evaluates the distribution of the bandwidth (past bandwidth) indicated in the traffic information extracted for each contracted bandwidth, and obtains the median and variance of the past bandwidth. As a bandwidth distribution function, for example, a normal distribution can be used, but is not limited to this.
[0021] Next, the estimation unit 14 approximates the relationship between the median value of the past bandwidth and the contracted bandwidth with an arbitrary function, as shown in Fig. 2B. As an approximation of the relationship between the median value of the bandwidth and the contracted bandwidth, for example, a linear approximation can be used, but is not limited to this.
[0022] Next, as shown in FIG. 2C, the estimation unit 14 calculates the median of the future bandwidth for a certain contracted bandwidth X based on the relationship between the above-mentioned bandwidth median and the contracted bandwidth and the future contracted bandwidth information prepared in advance. Finally, the estimation unit 14 obtains the distribution of traffic (bandwidth) caused by the service to be estimated based on the variance of the past bandwidth and the median of the future bandwidth, and derives the required bandwidth. For example, when the SLA (Service Agreement Level) of the network is 99%, the estimation unit 14 may determine the required bandwidth so as to cover 99% of the obtained distribution. In addition, the estimation unit 14 may derive the required bandwidth by adding a buffer that takes into account detouring traffic when a failure occurs in the communication network. By estimating the required bandwidth for each service, the required bandwidth of the link 21 can also be estimated.
[0023] Next, the operation of the bandwidth estimation device 10 according to the present embodiment will be described.
[0024] FIG. 3 is a flowchart showing an example of the operation of the bandwidth estimation device 10 according to the present embodiment, and is a diagram for explaining a bandwidth estimation method performed by the bandwidth estimation device 10. In FIG.
[0025] The service classification unit 11 acquires traffic information on the traffic flowing through the link 21 from the communication device 20, and classifies the acquired traffic information by service (step S11).
[0026] The estimation unit 14 estimates a required bandwidth for each service based on the traffic information for each service and the contracted bandwidth information stored in the user information DB 13 (step S12).
[0027] As described above, the bandwidth estimation device 10 according to this embodiment includes a service classification unit 11 and an estimation unit 14. The service classification unit 11 acquires traffic information on traffic flowing through the link 21 from the communication device 20, and classifies the acquired traffic information by service. The estimation unit 14 estimates a required bandwidth for each service based on the traffic information for each service and contracted bandwidth information on the contracted bandwidth of each of a plurality of users.
[0028] By classifying traffic information by service and estimating the required bandwidth for each service, it is possible to estimate the required bandwidth after classifying various services with different trends in bandwidth increase and decrease. Therefore, it is possible to improve the accuracy of estimating the required bandwidth in a communication network in which multiple lines are accommodated in link 21 and traffic of multiple services flows. In addition, by estimating the required bandwidth for each service, it is possible to determine the accommodation for each service, so that the need for additional equipment can be determined for each service, and the equipment cost can be reduced. In addition, since high estimation accuracy can be obtained even with data (traffic information) collected for a short period of time, the amount of data that needs to be recorded in traffic collection unit 12 can be suppressed, and the equipment cost can be reduced.
[0029] Second embodiment Fig. 4 is a diagram showing an example of the configuration of a bandwidth estimation device 10A according to the second embodiment of the present disclosure. In Fig. 4, the same components as those in Fig. 1 are denoted by the same reference numerals, and the description thereof will be omitted.
[0030] 4, the bandwidth estimation device 10A according to the present embodiment includes a service classification unit 11, a traffic collection unit 12, a user information DB 13, and an estimation unit 14A. The bandwidth estimation device 10A according to the present embodiment is different from the bandwidth estimation device 10 according to the first embodiment in that the estimation unit 14 is changed to an estimation unit 14A.
[0031] The estimation unit 14A includes a learning unit 141.
[0032] The learning unit 141 creates a learning model 142 by machine learning using the past contract bandwidth information of each of the multiple users stored in the user information DB 13 and the past traffic information of each of the multiple services output from the traffic collection unit 12 as learning data. Specifically, as shown in FIG. 5, the learning unit 141 creates a learning model 142 by machine learning using the past contract bandwidth information of each of the multiple users stored in the user information DB 13 and the past traffic information of each of the multiple services output from the traffic collection unit 12 as learning data. n-1 The contracted bandwidth information of each user and the traffic information of each service are input as learning data. The learning unit 141 optimizes the parameters of the learning model 142 based on the input learning data. That is, the learning unit 141 extracts the correlation between the contracted bandwidth information of each user and the traffic information of each service input as learning data.
[0033] The estimation unit 14A estimates the time t n ,t n+1 ,..., the estimation unit 14A inputs contracted bandwidth information (current and future contracted bandwidth information) of each of the multiple users in each area and estimates the required bandwidth for each service. n ,t n+1 ,... estimates the required bandwidth for each of the multiple services in each of the above. The estimation unit 14A may compare the estimated required bandwidth with an upper limit bandwidth allowed by the communication device 20, and perform an accommodation determination to determine whether the communication device 20 can accommodate the line.
[0034] As described above, in this embodiment, the bandwidth estimation device 10A includes an estimation unit 14A. The estimation unit 14A creates a learning model 142 by machine learning using past contract bandwidth information of each of the multiple users and past traffic information of each of the multiple services as learning data. Then, the estimation unit 14A inputs the contract bandwidth information of each of the multiple users to the learning model 142, and estimates the required bandwidth required for each service.
[0035] In the bandwidth estimation device 10 according to the first embodiment, the required bandwidth is estimated from the relationship between the contracted bandwidth of the user and the traffic, so that the accuracy of the estimation decreases when the bandwidth utilization rate differs for each user. To solve this problem, it is necessary to consider the bandwidth utilization rate of each user as a parameter, but this requires the analysis of a huge amount of data. By using machine learning as in this embodiment, it becomes possible to perform analysis in a short time, and the accuracy of the required bandwidth estimation can be improved compared to the first embodiment.
[0036] In the present embodiment, the contract bandwidth information of each of a plurality of users and the traffic information of each of a plurality of services are learned as learning data, but the learning data is not limited to this.Time information, destination information, source information, V-LAN, tag, packet length for each priority, number of packets, etc. may be further included in the learning data.
[0037] Also, in FIG. 6, an example is shown in which the estimation unit 14A outputs a required bandwidth value for each service as the estimation result, but this is not limited to this, and the estimation unit 14A may output a set of the required bandwidth value and the probability of taking that value as the estimation result.
[0038] (Third embodiment) Fig. 7 is a diagram showing an example of the configuration of a bandwidth estimation device 10B according to a third embodiment of the present disclosure. In Fig. 7, the same components as those in Fig. 4 are denoted by the same reference numerals, and the description thereof will be omitted.
[0039] 7, the bandwidth estimation device 10B according to the present embodiment includes a service classification unit 11, a traffic collection unit 12, a user information DB 13, an estimation unit 14B, and a periodicity extraction unit 15. The bandwidth estimation device 10B according to the present embodiment differs from the bandwidth estimation device 10A according to the second embodiment in that the estimation unit 14A is changed to an estimation unit 14B, and that a periodicity extraction unit 15 is added.
[0040] The periodicity extraction unit 15 extracts the periodicity of the traffic caused by each of the multiple services based on the traffic information for each service output from the traffic collection unit 12.
[0041] FIG. 8 is a diagram for explaining extraction of traffic periodicity by the periodicity extraction unit 15. As shown in FIG. 8, a time-series change in traffic volume is obtained from the traffic information. The periodicity extraction unit 15 extracts information on the traffic periodicity (periodicity information) by performing frequency analysis such as Fourier transform, discrete Fourier transform, and wavelet transform on the time-series change in traffic volume. FIG. 8 shows an example in which the time-series change in traffic volume is converted into amplitude (corresponding to the magnitude of periodicity) at each frequency (corresponding to the reciprocal of the period) by Fourier transform. The periodicity extraction unit 15 outputs the extracted traffic periodicity information for each service to the estimation unit 14B.
[0042] The estimation unit 14B differs from the estimation unit 14A in that the learning unit 141 is changed to a learning unit 141B.
[0043] The learning unit 141B creates a learning model 142B by machine learning using the past contract bandwidth information of each of the multiple users stored in the user information DB 13, the past traffic information of each of the multiple services output from the traffic collection unit 12, and the traffic periodicity of each of the multiple services extracted by the periodicity extraction unit 15 as learning data. Specifically, as shown in FIG. 9, the learning unit 141B creates a learning model 142B by machine learning using the past contract bandwidth information of each of the multiple users stored in the user information DB 13, the past traffic information of each of the multiple services output from the traffic collection unit 12, and the traffic periodicity of each of the multiple services extracted by the periodicity extraction unit 15 as learning data. n-1The contracted bandwidth information of each user, the traffic information of each service, and the periodicity information of the traffic of each service (frequency and amplitude indicating the periodicity of the traffic) are input as learning data. The learning unit 141B optimizes the parameters of the learning model 142B based on the input learning data. That is, the learning unit 141B extracts the correlation between the contracted bandwidth information of each user, the traffic information of each service, and the periodicity of the traffic of each service, which are input as learning data.
[0044] The estimation unit 14B estimates the time t n ,t n+1 ,...The contracted bandwidth information (current and future contracted bandwidth information) of each of the multiple users in each area is input, and the required bandwidth for each service is estimated.
[0045] As described above, in this embodiment, the bandwidth estimation device 10B includes a periodicity extraction unit 15 and an estimation unit 14B. The periodicity extraction unit 15 extracts periodicity of traffic caused by each of the multiple services based on traffic information of each of the multiple services. The estimation unit 14B includes a learning unit 141B. The learning unit 141B creates a learning model 142B by machine learning using contract bandwidth information of each of the multiple users, traffic information of each of the multiple services, and the periodicity extracted by the periodicity extraction unit 15 as learning data.
[0046] By adding the traffic periodicity of each of multiple services to the learning data, it is possible to learn information about the time dependency of traffic that differs for each service, which further improves the accuracy of estimating the required bandwidth for multiple services. In addition, for new services, the traffic periodicity may change from testing to normal operation. In this case, extracting the traffic periodicity makes it possible to learn by taking into account information about the operation mode of the service.
[0047] (Fourth embodiment) Fig. 10 is a diagram showing an example of the configuration of a bandwidth estimation device 10C according to a fourth embodiment of the present disclosure. In Fig. 10, the same components as those in Fig. 7 are denoted by the same reference numerals, and the description thereof will be omitted.
[0048] 10, the bandwidth estimation device 10C according to the present embodiment includes a service classification unit 11, a traffic collection unit 12, a user information DB 13, an estimation unit 14C, a periodicity extraction unit 15, and a feature determination unit 16. The bandwidth estimation device 10C according to the present embodiment differs from the bandwidth estimation device 10B according to the third embodiment in that the estimation unit 14B is changed to an estimation unit 14C, and that a feature determination unit 16 is added.
[0049] The characteristic determination unit 16 determines the presence or absence of periodicity characteristics of the traffic for each service extracted by the periodicity extraction unit 15. For example, the characteristic determination unit 16 determines whether the traffic of the service changes with a characteristic period, the magnitude of the peak at a characteristic frequency, whether the traffic is stationary or random, etc.
[0050] FIG. 11 is a diagram showing an example of the determination of the periodicity characteristics by the characteristics determining section 16. In FIG.
[0051] As described above, the time series change of the traffic is converted into a spectrum of frequency and amplitude by Fourier transform as shown in Fig. 11. The feature determining unit 16 fits the spectrum of frequency and amplitude with an arbitrary function, and determines that there is periodicity if there is a peak that deviates from the fitting curve at an arbitrary rate, and determines that there is no periodicity if there is no peak. The feature determining unit 16 may also perform a similar analysis on the partial coefficients of the spectrum of frequency and amplitude, for example.
[0052] Referring again to FIG. 10, the feature determining unit 16 outputs the determination result to the estimation unit 14C.
[0053] The estimation unit 14C differs from the estimation unit 14B in that the learning unit 141B is changed to a learning unit 141C.
[0054] The learning unit 141C creates a learning model 142C by machine learning using the traffic information for each service, the contract bandwidth information, and the traffic periodicity information for each service as learning data. Here, the learning unit 141C creates a plurality of learning models 142C with different parameters used for learning, for example.
[0055] The estimation unit 14C determines the required bandwidth for each service by using a learning model 142C according to the periodicity characteristics determined by the characteristic determination unit 16, out of the multiple learning models 142C created by the learning unit 141C.
[0056] For example, when traffic has a time-series periodicity (high correlation between time and traffic), it is necessary to apply a model suitable for handling a non-stationary state. On the other hand, when traffic does not have a time-series periodicity (low correlation between time and traffic) (when data is stationary or random), it is necessary to apply a model suitable for handling a stationary state, which has a smaller calculation processing load than a model suitable for handling a non-stationary state. The estimation unit 14C selects the learning model 142C according to the characteristics of the traffic caused by the service based on the determination result of the characteristic determination unit 16. In this way, by selecting the learning model 142C according to the characteristics of the traffic caused by the service, it is possible to further improve the estimation accuracy and reduce the calculation processing load. In addition, when estimating the required bandwidth of a new service for which it is not known what learning model 142C should be used, it is possible to improve the accuracy of estimating the required bandwidth for the new service by using the learning model 142C used to estimate the required bandwidth of another service with similar traffic periodicity.
[0057] In the present embodiment, an example has been described in which the learning model 142C used to estimate the required bandwidth is selected from the multiple learning models 142C based on the result of the determination by the characteristic determination unit 16, but the present invention is not limited to this. For example, in the bandwidth estimation device 10 shown in Fig. 1, when multiple algorithms for estimating the required bandwidth are prepared, an algorithm to be used to estimate the required bandwidth may be selected from the multiple prepared algorithms according to the characteristics of the traffic caused by the service.
[0058] Fifth embodiment Fig. 12 is a diagram showing an example of the configuration of a bandwidth estimation device 10D according to a fifth embodiment of the present disclosure. In Fig. 12, the same components as those in Fig. 10 are denoted by the same reference numerals, and the description thereof will be omitted.
[0059] 12, the bandwidth estimation device 10D according to the present embodiment includes a service classification unit 11, a traffic collection unit 12, a user information DB 13, an estimation unit 14D, a periodicity extraction unit 15, a feature determination unit 16, and an evaluation unit 17. The bandwidth estimation device 10D according to the present embodiment is different from the bandwidth estimation device 10C shown in FIG. 10 in that the estimation unit 14C is changed to an estimation unit 14D and that an evaluation unit 17 is added.
[0060] The evaluation unit 17 compares the necessary bandwidth of the service estimated by the estimation unit 14D with the bandwidth actually required for the service, and outputs the comparison result to the estimation unit 14D. The evaluation unit 17 outputs, for example, the difference between the necessary bandwidth of the service estimated by the estimation unit 14D and the bandwidth actually required as the comparison result to the estimation unit 14D.
[0061] The estimation unit 14D, like the estimation unit 14C, selects a learning model C142C from the multiple learning models 142C according to the determination result of the periodicity characteristics of the traffic caused by the service, and estimates the required bandwidth of the service. Then, the estimation unit 14D changes the parameters of the learning model 142C used to estimate the required bandwidth according to the result of the comparison by the evaluation unit 17. For example, when an error of an arbitrary ratio or more occurs between the required bandwidth estimated using the first learning model 142C-1 and the actually required bandwidth, the estimation unit 14D estimates the required bandwidth again using the second learning model 142C-2. Then, the estimation unit 14D uses, for example, the learning model 142 with the smallest error from the actually required bandwidth between the required bandwidth estimated by the first learning model 142C-1 and the required bandwidth estimated by the second learning model 142C-2, as a model for estimating the required bandwidth.
[0062] In this manner, in the present embodiment, the estimation of the required bandwidth can be further improved by evaluating the estimation result of the required bandwidth by the estimation unit 14D using the evaluation unit 17. In particular, a remarkable effect is expected when estimating the required bandwidth of a new service for which it is not known what kind of learning model 142C should be used.
[0063] In the present embodiment, the parameters of the learning model 142C used to estimate the required bandwidth are changed according to the result of the comparison by the evaluation unit 17, but the present invention is not limited to this. For example, in the bandwidth estimation device 10 shown in FIG. 1, the parameters of an algorithm for estimating the required bandwidth may be changed according to the result of the comparison by the evaluation unit 17. Also, the parameters related to the determination by the feature determination unit 16 may be changed according to the result of the comparison by the evaluation unit 17.
[0064] Next, a hardware configuration of the bandwidth estimation device 10 according to the present disclosure will be described. Note that, although the bandwidth estimation device 10 will be described below as an example, the same applies to the bandwidth estimation devices 10A, 10B, 10C, and 10D.
[0065] Fig. 13 is a diagram illustrating an example of a hardware configuration of a bandwidth estimation device 10 according to an embodiment of the present disclosure. Fig. 13 illustrates an example of a hardware configuration of the bandwidth estimation device 10 when the bandwidth estimation device 10 is configured by a computer capable of executing program instructions. Here, the computer may be a general-purpose computer, a dedicated computer, a workstation, a PC (Personal computer), an electronic notepad, or the like. The program instructions may be program code, code segments, or the like for performing necessary tasks.
[0066] 13, the bandwidth estimation device 10 includes a processor 210, a ROM (Read Only Memory) 220, a RAM (Random Access Memory) 230, a storage 240, an input unit 250, a display unit 260, and a communication interface (I / F) 270. Each component is communicably connected to each other via a bus 290. The processor 210 is specifically a CPU (Central Processing Unit), an MPU (Micro Processing Unit), a GPU (Graphics Processing Unit), a DSP (Digital Signal Processor), a SoC (System on a Chip), or the like, and may be configured by a plurality of processors of the same type or different types.
[0067] The processor 210 is a controller that controls each component and executes various arithmetic processing. That is, the processor 210 reads a program from the ROM 220 or the storage 240, and executes the program using the RAM 230 as a working area. The processor 210 controls each component and executes various arithmetic processing according to the program stored in the ROM 220 or the storage 240. In this embodiment, the ROM 220 or the storage 240 stores a program for causing a computer to function as the bandwidth estimation device 10 according to the present disclosure. The program is read and executed by the processor 210, thereby realizing each component of the bandwidth estimation device 10, that is, the service classification unit 11, the traffic collection unit 12, and the estimation unit 14.
[0068] The program may be provided in a form stored in a non-transitory storage medium such as a CD-ROM (Compact Disk Read Only Memory), a DVD-ROM (Digital Versatile Disk Read Only Memory), a USB (Universal Serial Bus) memory, etc. The program may also be provided in a form downloaded from an external device via a network.
[0069] The ROM 220 stores various programs and various data (e.g., contracted bandwidth information). The RAM 230 temporarily stores programs or data as a working area. The storage 240 is configured with an HDD (Hard Disk Drive) or an SSD (Solid State Drive) and stores various programs including an operating system and various data.
[0070] The input unit 250 includes a pointing device such as a mouse and a keyboard, and is used to perform various input operations.
[0071] The display unit 260 is, for example, a liquid crystal display, and displays various information. The display unit 260 may function as the input unit 250 by adopting a touch panel system.
[0072] The communication interface 270 is an interface for communicating with an external device such as the communication device 20, and uses standards such as Ethernet (registered trademark), FDDI, and Wi-Fi (registered trademark), for example.
[0073] A computer can be suitably used to function as each unit of the above-mentioned bandwidth estimation device 10. Such a computer can be realized by storing a program describing the processing contents for realizing the functions of each unit of the bandwidth estimation device 10 in a storage unit of the computer, and reading and executing the program by a processor of the computer. In other words, the program can cause the computer to function as the above-mentioned bandwidth estimation device 10. The program can also be recorded on a non-transitory recording medium. The program can also be provided via a network.
[0074] The following supplementary notes are further disclosed regarding the above embodiment.
[0075] (Additional note 1) A bandwidth estimation device that estimates a bandwidth required for a link between communication devices, the link accommodating multiple user lines and carrying traffic resulting from multiple services, comprising: a classification unit that acquires traffic information related to the traffic from the communication device and classifies the acquired traffic information for each service; a bandwidth estimation unit that estimates a required bandwidth for each of the services based on traffic information for each of the services and contract bandwidth information regarding a contract bandwidth for each of the multiple users;
[0076] (Additional note 2) In the bandwidth estimation device according to claim 1, The estimation unit is creating a learning model by machine learning using contract bandwidth information of each of the plurality of users and traffic information of each of the plurality of services as learning data; A bandwidth estimation device that inputs contracted bandwidth information of each of the multiple users into the learning model and estimates the required bandwidth for each of the services.
[0077] (Additional note 3) In the bandwidth estimation device according to claim 2, a periodicity extraction unit configured to extract a periodicity of traffic caused by each of the plurality of services based on traffic information of each of the plurality of services; The estimation unit creates the learning model through machine learning using contracted bandwidth information of each of the multiple users, traffic information of each of the multiple services, and the periodicity extracted by the periodicity extraction unit as learning data.
[0078] (Additional note 4) In the bandwidth estimation device according to claim 3, a feature determining unit that determines a feature of the periodicity extracted by the periodicity extracting unit, The estimation unit creates a plurality of learning models having different learning-related parameters, and estimates the required bandwidth using a learning model among the plurality of learning models that corresponds to the periodicity characteristics determined by the characteristic determination unit.
[0079] (Additional note 5) In the bandwidth estimation device according to claim 1, a periodicity extraction unit that extracts a periodicity of traffic caused by each of the plurality of services based on past traffic information of each of the plurality of services; a feature determination unit that determines a feature of the periodicity extracted by the periodicity extraction unit, The estimation unit estimates the required bandwidth using an algorithm corresponding to the periodicity characteristic determined by the characteristic determination unit, out of a plurality of algorithms for estimating the required bandwidth.
[0080] (Additional note 6) In the bandwidth estimation device according to claim 4, An evaluation unit that compares the required bandwidth of the service estimated by the estimation unit with a bandwidth actually required for the service, The estimation unit changes parameters of an algorithm or model used to estimate the required bandwidth according to a result of the comparison by the evaluation unit.
[0081] (Additional note 7) 1. A bandwidth estimation method for estimating a bandwidth required for a link between communication devices, the link accommodating multiple user lines and carrying traffic resulting from multiple services, comprising the steps of: acquiring traffic information relating to the traffic from the communication device, and classifying the acquired traffic information by the service; A bandwidth estimation method for estimating a required bandwidth for each of the services based on traffic information for each of the services and contract bandwidth information regarding a contract bandwidth for each of the multiple users.
[0082] (Additional note 8) A non-transitory storage medium storing a program executable by a computer, the non-transitory storage medium storing the program causing the computer to function as the control device described in appended claim 1.
[0083] Although the above-mentioned embodiment has been described as a representative example, it is obvious to those skilled in the art that many modifications and substitutions can be made within the spirit and scope of the present disclosure. Therefore, the present invention should not be interpreted as being limited by the above-mentioned embodiment, and various modifications or changes can be made without departing from the scope of the claims. For example, it is possible to combine multiple building blocks shown in the configuration diagram of the embodiment into one, or to divide one building block. [Explanation of symbols]
[0084] 10, 10A, 10B, 10C Bandwidth estimation device 11 Service Classification Section 12 Traffic Collection Department 13 User information DB 14, 14A, 14B, 14C, 14D Estimation section 141, 141B, 141C Learning Department 142, 142B, 142C Learning Model 15 Periodicity extraction part 16 Feature determination section 17 Evaluation Section 20. Communications Equipment 21 Link
Claims
1. A bandwidth estimation device that estimates a bandwidth required for a link between communication devices, the link accommodating multiple user lines and carrying traffic resulting from multiple services, comprising: a classification unit that acquires traffic information related to the traffic from the communication device and classifies the acquired traffic information for each service; and an estimation unit that estimates a required bandwidth for each of the services based on traffic information for each of the services and contract bandwidth information regarding a contract bandwidth for each of the plurality of users, the estimation unit comprising: creating a learning model by machine learning using contract bandwidth information of each of the plurality of users and traffic information of each of the plurality of services as learning data; A bandwidth estimation device that inputs contracted bandwidth information of each of the multiple users into the learning model and estimates the required bandwidth for each of the services.
2. 2. The bandwidth estimation device according to claim 1, a periodicity extraction unit configured to extract a periodicity of traffic caused by each of the plurality of services based on traffic information of each of the plurality of services; The estimation unit creates the learning model through machine learning using contracted bandwidth information of each of the multiple users, traffic information of each of the multiple services, and the periodicity extracted by the periodicity extraction unit as learning data.
3. 3. The bandwidth estimation device according to claim 2, a feature determining unit that determines a feature of the periodicity extracted by the periodicity extracting unit, The estimation unit creates a plurality of learning models having different learning-related parameters, and estimates the required bandwidth using a learning model among the plurality of learning models that corresponds to the periodicity characteristics determined by the characteristic determination unit.
4. A bandwidth estimation device that estimates a bandwidth required for a link between communication devices, the link accommodating multiple user lines and carrying traffic resulting from multiple services, comprising: a classification unit that acquires traffic information related to the traffic from the communication device and classifies the acquired traffic information for each service; an estimation unit that estimates a required bandwidth for each of the services based on traffic information for each of the services and contract bandwidth information regarding a contract bandwidth for each of the plurality of users; a periodicity extraction unit that extracts a periodicity of traffic caused by each of the plurality of services based on past traffic information of each of the plurality of services; a feature determination unit that determines a feature of the periodicity extracted by the periodicity extraction unit, The estimation unit estimates the required bandwidth using an algorithm corresponding to the periodicity characteristic determined by the characteristic determination unit, from among a plurality of algorithms for estimating the required bandwidth based on traffic information for each service and contracted bandwidth information for each of the plurality of users.
5. 5. The bandwidth estimation device according to claim 3, An evaluation unit that compares the required bandwidth of the service estimated by the estimation unit with a bandwidth actually required for the service, The estimation unit changes parameters of an algorithm or model used to estimate the required bandwidth according to a result of the comparison by the evaluation unit.
6. 1. A bandwidth estimation method for estimating a bandwidth required for a link between communication devices, the link accommodating multiple user lines and carrying traffic resulting from multiple services, comprising the steps of: acquiring traffic information relating to the traffic from the communication device, and classifying the acquired traffic information by the service; Estimating a required bandwidth for each of the services based on traffic information for each of the services and contract bandwidth information regarding a contract bandwidth for each of the plurality of users; creating a learning model by machine learning using contract bandwidth information of each of the plurality of users and traffic information of each of the plurality of services as learning data; A bandwidth estimation method for inputting contracted bandwidth information of each of the plurality of users into the learning model and estimating the required bandwidth for each of the services.
7. A program for causing a computer to function as the bandwidth estimation device according to any one of claims 1 to 5.
Citation Information
Patent Citations
Communication band calculation apparatus, method, and program
JP2009118274A
Prediction device, prediction method, and program
JP2020136894A
