Communication device, parameter generation device, communication system, method, and computer program

Abstract

To provide a communication device capable of efficiently allocating computational resources even when communicating with a plurality of communication nodes at different frequencies.SOLUTION: A communication device 24 includes a frequency measurement unit 35 that measures the frequency of communication with each of a plurality of communication nodes 111 to 11N and generates observation data consisting of a binary time series representing the measured frequency for each communication node, a clustering unit 36 that classifies the observation data into one of a plurality of clusters based on a mixed probability distribution having a maximum likelihood parameter estimated by machine learning, and a computational resource allocation unit 37 that allocates an allocation amount of computational resources associated with the cluster to each communication node.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present disclosure relates to a communication technology using machine learning in a communication system. [Background technology]

[0002] In recent years, communication technologies using machine learning have been proposed to improve communication quality in communication systems such as fixed communication systems and mobile communication systems (for example, fifth generation mobile communication systems). This type of communication technology is disclosed, for example, in Patent Document 1 (JP Patent Publication No. 2023-123991). Patent Document 1 discloses a technology that uses a machine learning model such as an artificial neural network to allocate resources such as frequency bands and time for wireless communication in a mobile communication system. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2023-123991 A Summary of the Invention [Problem to be solved by the invention]

[0004] In a communication system that dynamically allocates computational resources according to the communication status with a large number of communication nodes, it is preferable to efficiently allocate computational resources to each communication node. In particular, when there is a possibility that communication frequency with an unspecified number of communication nodes (e.g., mobile terminals using a fifth generation mobile communication system) is high, efficient allocation of computational resources to each communication node becomes important.

[0005] However, when a communication system performs parallel processing using shared computational resources, simultaneously allocating the same amount of computational resources to a communication node that communicates infrequently and a communication node that communicates frequently is inefficient and results in a lot of waste from the perspective of effective use of computational resources.

[0006] In view of the above, an object of the present disclosure is to provide a communication device, a parameter generation device, a communication system, a method, and a computer program that enable efficient allocation of computational resources even when communicating with multiple communication nodes at different frequencies. [Means for solving the problem]

[0007] A communication device according to a first aspect of the present disclosure is characterized in that it comprises a frequency measurement unit that measures a frequency of communication between a plurality of communication nodes and each communication node and generates observation data consisting of a binary time series representing the measured frequency for each of the communication nodes, a clustering unit that classifies the observation data into one of a plurality of clusters based on a mixed probability distribution having a maximum likelihood parameter estimated by machine learning, and a resource allocation unit that allocates to each communication node an allocation amount of computational resources associated with the cluster into which the observable data is classified.

[0008] A parameter generation device according to a second aspect of the present disclosure includes a training data storage unit in which training data consisting of a binary time series representing a frequency of communication with each of a predetermined number of communication nodes is stored, a machine learning unit that uses a training data set read from the training data storage unit to perform machine learning based on a mixed probability distribution to estimate a maximum likelihood parameter of the mixed probability distribution, and a parameter storage unit that stores data on the estimated maximum likelihood parameter, and is characterized in that the training data for each of the communication nodes is previously associated with an allocated amount of computational resources.

[0009] A communication system according to a third aspect of the present disclosure includes the communication device according to the first aspect and the parameter generating device according to the second aspect.

[0010] A method according to a fourth aspect of the present disclosure is a method for allocating computational resources in a communication device, comprising the steps of measuring a frequency of communication between a plurality of communication nodes and each of the communication nodes, generating observational data consisting of a binary time series representing the measured frequency for each of the communication nodes, classifying the observational data into one of a plurality of clusters based on a mixed probability distribution having a maximum likelihood parameter estimated by machine learning, and allocating to each of the communication nodes an amount of computational resources corresponding to the cluster into which the observational data has been classified.

[0011] A computer program according to a fifth aspect of the present disclosure is a computer program for a communications device having one or more processors, configured to, when executed by the one or more processors, cause the communications device to perform a method according to the fourth aspect.

[0012] A method according to a sixth aspect of the present disclosure includes the steps of: reading a training dataset from a training data storage unit in which training data consisting of a binary time series representing the frequency of communication between a plurality of communication nodes and each of the communication nodes is stored; using the read training dataset to perform machine learning based on a mixed probability distribution to estimate a maximum likelihood parameter of the mixed probability distribution; and storing data of the estimated maximum likelihood parameter in a parameter storage unit, wherein the training data for each of the communication nodes is previously associated with an allocated amount of computational resources.

[0013] A computer program according to a seventh aspect of the present disclosure is a computer program for a parameter generating device having one or more processors, configured to, when executed by the one or more processors, cause the parameter generating device to perform the method according to the sixth aspect. Effect of the Invention

[0014] According to the first to seventh aspects of the present disclosure, even when communication is performed with a plurality of communication nodes at different frequencies, it becomes possible to efficiently allocate computational resources. [Brief description of the drawings]

[0015] [Figure 1] 1 is a block diagram illustrating a schematic configuration example of a communication network system according to an embodiment of the present disclosure. [Diagram 2] FIG. 2 is a schematic diagram illustrating an example of a hardware configuration for implementing components of the communication system. [Diagram 3] 13 is a flowchart illustrating an example of a procedure for a computational resource allocation process. [Figure 4] 4A and 4B are schematic diagrams for explaining the frequency of location registration requests. [Diagram 5] 11 is a diagram illustrating an example of observed data indicating a measured frequency of location registration requests. [Figure 6] 13 is a flowchart illustrating another example of a procedure for a computational resource allocation process. [Figure 7] 7A and 7B are schematic diagrams for explaining the frequency of user communications. [Figure 8] FIG. 10 is a diagram illustrating an example of observed data indicating the measured frequency of packet communication signals. [Figure 9] 13 is a flowchart illustrating an example of a procedure for a parameter generation process. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0016] Various embodiments according to the present disclosure will be described in detail below with reference to the drawings. In addition, components with the same reference numerals throughout the drawings have the same configurations and functions.

[0017] 1 is a block diagram illustrating a schematic configuration example of a communication network system 1 according to an embodiment of the present disclosure. As illustrated in FIG. 1, the communication system 1 includes user equipment (UE) 111 to 113 as communication nodes.N and base stations 201 to 20 capable of wireless communication with N These base stations 201 to 20 N and a communication system 21 connected to the UEs 111 to 111. N is an integer equal to or greater than 2. N The subscribers 101 to 10 of the service of the communication network system 1 N UE11 is a mobile terminal used by n ~11 N Examples of such devices include, but are not limited to, smartphones, mobile phones, tablet devices, wearable devices, and laptop computers.

[0018] The communication network system 1 in Fig. 1 is configured as a mobile communication system conforming to the fifth generation (5G) communication standard, and the communication system 21 includes a core network of the fifth generation (5G) mobile communication system. As shown in Fig. 1, the communication system 21 includes an access and mobility management function / security anchor function (AMF / SEAF) 22, a unified data management unit (UDM) 23, a unified data repository unit (UDR) 24, a user plane function (UPF) 25, and a parameter generation device 26. The communication system 21 also includes a plurality of other function units such as a session management function (SMF), but these function units are omitted in Fig. 1.

[0019] The AMF / SEAF 22 is a communication device that performs, for example, mobility management, connection management, and primary authentication using a SUCI (Subscription Concealed Identifier). The UDM 23 is a communication device that manages information such as subscriber information (subscription data) using information stored in the UDR 24. The UDR 24 is a communication device that stores subscriber information and context information such as the state of a user terminal.

[0020] The UDR 24 stores location registration information of the subscriber's user terminal. n (n is an arbitrary number) is, for example, a user terminal 11 in order to use the service of the communication network system 1. n When the power is turned on, when the location registration area is changed, or periodically, the base station 20 n The location registration request is sent to the base station 20. n The location registration request is received by the UDR 24 via the AMF / SEAF 22 and the UDM 23. The UDR 24 executes a location registration process in response to the location registration request, and if the location registration process is successful, n The UDR 24 has a function of performing parallel processing using common computing resources (e.g., multiple processor cores) to execute the location registration process. However, the user terminal 11 that issues location registration requests with low frequency may not be able to perform the location registration process. n1 and a user terminal 11 that transmits location registration requests with high frequency. n2 If there exists a user terminal 11 n1 and user terminal 11 n2 Therefore, the UDR 24 of this embodiment has a computational resource allocation function 30 that efficiently allocates computational resources even in such a case.

[0021] The computational resource allocation function 30 of the UDR 24 includes a data storage unit 31, a frequency measurement unit 35, a clustering unit 36, and a resource allocation unit 37. The data storage unit 31 stores parameter data 32 indicating a parameter group of a mixture probability distribution having a maximum likelihood parameter estimated by machine learning, and a lookup table 33. The configuration of the computational resource allocation function 30 will be described in detail later.

[0022] The UPF25 has a function to process user plane (U-plane) communications compliant with the 5G communications standard, and N The UPF 25 is a communication device that serves as an anchor point for user communications between the user terminal 11 and a data network (DN) 15 such as the Internet. The UPF 25 has a function of performing parallel processing using common computing resources (e.g., multiple processor cores) to process user plane communications. However, the UPF 25 is a communication device that serves as an anchor point for user communications between the user terminal 11 and a data network (DN) 15 such as the Internet. n1 and a user terminal 11 that frequently performs user communication. n2 If there exists a user terminal 11 n1 and user terminal 11 n2 It is inefficient to simultaneously allocate the same amount of computational resources to both of them. Therefore, the UPF 25 of this embodiment has a computational resource allocation function 40 that efficiently allocates computational resources even in such a case.

[0023] The computational resource allocation function 40 of the UPF 25 includes a data storage unit 41, a frequency measurement unit 45, a clustering unit 46, and a resource allocation unit 47. The data storage unit 41 stores parameter data 42 indicating a parameter group of a mixture probability distribution having a maximum likelihood parameter estimated by machine learning, and a lookup table 43. The configuration of the computational resource allocation function 40 will be described in detail later.

[0024] The parameter generating device 26 includes a training data storage unit 51 in which training data sets 52A and 52B are stored, a machine learning unit 50 that performs machine learning based on a mixed probability distribution using the training data sets 52A and 52B to estimate a maximum likelihood parameter of the mixed probability distribution, and a parameter storage unit 53 that stores parameter data 54A and 54B indicating a parameter group of the mixed probability distribution having the estimated maximum likelihood parameter. The training data set 52A is a data set used to generate parameter data 54A for the computational resource allocation function 30 of the UDR 24, and the training data set 52B is a data set used to generate parameter data 54B for the computational resource allocation function 40 of the UPF 25. The parameter generating device 26 supplies the parameter data 54A to the computational resource allocation function 30 of the UDR 24 at an appropriate timing. The computational resource allocation function 30 can store the parameter data 54A in the data storage unit 31 as parameter data 32 and use it. Furthermore, the parameter generating device 26 supplies the parameter data 54B to the computational resource allocation function 40 of the UPF 25 at an appropriate timing. The computational resource allocation function 40 can store and use the parameter data 54B in the data storage unit 41 as parameter data 42. The configuration of the parameter generating device 26 will be described in detail later.

[0025] All or part of the components of the communication system 21 described above (AMF / SEAF 22, UDM 23, UDR 24, UPF 25, and parameter generating device 26) may be realized by one computer including one or more processors, or by multiple computers interconnected via communication paths. All or part of the components of the communication system 21 can be realized by one or more processors including one or more processing units that execute processing according to computer program codes (instructions) read from a non-volatile memory (computer-readable recording medium). For example, a CPU (Central Processing Unit), a GPU (Graphics Processing Unit), or an NPU (Neural Network Processing Unit) can be used as the processing unit.

[0026] 2 is a schematic diagram of an information processing device (computer) 100, which is an example of a hardware configuration for implementing the components of the communication system 21. The information processing device 100 includes a processor 101 including a plurality of processor cores μC, ..., μC, a random access memory (RAM) 102, a non-volatile memory 103, a large capacity memory 104, an input / output interface 105, and a signal path 106. The signal path 106 is a bus for mutually connecting the processor 101, the RAM 102, the non-volatile memory 103, the large capacity memory 104, and the input / output interface 105. The RAM 102 is a data storage area used when the processor 101 executes digital signal processing. The non-volatile memory 103 has a data storage area in which the code (group of instructions) of a computer program executed by the processor 101 is stored.

[0027] Next, the computational resource allocation function 30 of the UDR 24 will be described in detail with reference to Figures 3 to 5. Figure 3 is a flowchart showing an example of a computational resource allocation process performed by the computational resource allocation function 30.

[0028] The frequency measurement unit 35 in FIG. n and collects location registration requests received from the UE 11 and calculates the frequency of the location registration requests in a predetermined period (for each UE 11). n The frequency measurement unit 35 measures the frequency of communication with the target device (step S11 in FIG. 3). n can be generated (step S12 in FIG. 3).

[0029] 4A and 4B are schematic diagrams for explaining the frequency of location registration requests. N The location registration requests RS1 to RS N The base stations 201 to 20 N , and is observed by the UDR 24 via the AMF / SEAF 22 and the UDM 23. FIG. 4B shows the location registration request RS observed at a predetermined time interval. i ,…,RS j ,…,RS k ,…,RS p ,RS q ,…,RS r ,…,RS s Here, the subscripts i, j, k, r, and s are integers in the range of 1 to N that are different from one another. In FIG. 4B, circles ("○") representing observed location registration requests are arranged along the time axis t, with the upward direction indicating location registration requests observed more frequently and the downward direction indicating location registration requests observed less frequently. For example, when UE11 i A location registration request RS i was observed 15 times during the specified period, and UE11 s A location registration request RS s It can be seen that was observed twice within the specified period.

[0030] FIG. 5 shows the location registration request RS n Observational data x representing the measured frequency of n In the example of FIG. 5, a location registration request RS nWhen the location registration request RS n When is not observed, the observed data value is "0". n is represented as a binary vector whose elements are binary variables that take the values ​​"0" or "1."

[0031] As described above, the parameter generating device 26 supplies the parameter data 54A indicating a parameter set of a mixed probability distribution having a maximum likelihood parameter estimated by machine learning to the computational resource allocation function 30 of the UDR 24, and the computational resource allocation function 30 of the UDR 24 can use the parameter data 54A as the parameter data 32. The clustering unit 36 ​​shown in FIG. 1 can acquire the parameter data 32 from the data storage unit 31 and execute clustering (cluster classification) using the parameter set of the mixed probability distribution indicated by the parameter data 32. That is, the clustering unit 36 ​​classifies the observed data x n can be classified into one of a plurality of clusters (step S13 in FIG. 3).

[0032] In this embodiment, the observation data x n Since each of the binary variables that are elements of can be considered to follow a Bernoulli distribution, a mixed Bernoulli distribution can be used as the mixed probability distribution represented by the parameter data 32. As machine learning, a likelihood estimation method using the EM (Expectation-Maximization) algorithm can be used.

[0033] According to the maximum likelihood estimation method using the EM algorithm of the mixed Bernoulli distribution, the observed data x n The proportion of people who belong to the kth cluster is the burden rate γ nk It can be calculated as follows: nk is expressed by the following equation (1).

[0034]

number

[0035] Here, π k ,μ k is the maximum likelihood parameter estimated by the maximum likelihood estimation method using the EM algorithm. nk More details will be provided later.

[0036] The clustering unit 36 ​​clusters the observed data x n The burden rate γ n1 ,…,γ nK The users can be classified into a cluster corresponding to the highest burden rate among the above (step S13 in FIG. 3).

[0037] The lookup table 33 in the data storage unit 31 has a predetermined correspondence between clusters and the amount of computational resources allocated to them. The resource allocation unit 37 refers to the lookup table 33 and calculates the amount of computational resources allocated to the clusters. n The resource allocation unit 37 then determines the amount of computational resources that is associated with the cluster into which the UE 11 is classified (step S14 in FIG. 3). The amount of computational resources that can be allocated may be, for example, the number of processor cores that are allocated. Next, the resource allocation unit 37 allocates the determined amount of computational resources to the UE 11. n (Step S15 in FIG. 3).

[0038] As described above, the computational resource allocation function 30 of the UDR 24 allocates the computational resources to each UE 11. n Observation data x consisting of a binary time series representing the frequency of measurements for each communication node n Using the machine learning, we perform clustering (cluster classification) based on a mixture probability distribution with maximum likelihood parameters estimated by machine learning, and estimate the observed data x n Since the amount of computational resources assigned to the cluster into which the UEs 111 to 11 are classified is determined, N Even if communication (reception of location registration requests) is performed with each of the UDR 24 at different frequencies, the computational resources of the UDR 24 can be efficiently allocated.

[0039] Next, the computational resource allocation function 40 of the UPF 25 will be described in detail with reference to Fig. 6 to Fig. 8. Fig. 6 is a flowchart showing an example of a procedure of computational resource allocation processing by the computational resource allocation function 40.

[0040] The frequency measurement unit 45 in FIG. n The frequency measurement unit 45 monitors user communications (e.g., packet communications) with the user and measures the frequency of the user communications in a predetermined period (step S21 in FIG. 6). n can be generated (step S22 in FIG. 6).

[0041] 7A and 7B are schematic diagrams for explaining the frequency of user communications. As illustrated in FIG. N and DN (data network) 15. N is observed by the UPF 25. FIG. 7B shows the packet communication signal PS i ,…,PS j ,…,PS k ,…,PS p ,PS q ,…,PS r ,…,PS s 7B is a time chart showing an example of the above. Here, the subscripts i, j, k, r, and s are integers in the range of 1 to N that are different from one another. In FIG. 7B, circles ("○") representing observed packet communication signals are arranged along the time axis t, with packet communication signals observed more frequently going up and packet communication signals observed less frequently going down. For example, UE11 i Packet communication signal between PS and DN15 i was observed 17 times during the specified period, and UE11 r Packet communication signal between PS and DN15 s It can be seen that was observed four times within the specified period.

[0042] FIG. 8 shows the packet communication signal PS nObservational data x representing the measured frequency of n In the example of FIG. 8, the packet communication signal PS n When the packet communication signal PS n When is not observed, the observed data value is "0". n is represented as a binary vector whose elements are binary variables that take the values ​​"0" or "1."

[0043] As described above, the parameter generating device 26 supplies the parameter data 54B indicating a parameter set of a mixture probability distribution having a maximum likelihood parameter estimated by machine learning to the computational resource allocation function 40 of the UPF 25, and the computational resource allocation function 40 of the UPF 25 can use the parameter data 54B as parameter data 42. The clustering unit 46 shown in FIG. 1 can acquire the parameter data 42 from the data storage unit 41 and execute clustering (cluster classification) using the parameter set of the mixture probability distribution indicated by the parameter data 42. That is, the clustering unit 36 ​​classifies the observation data x n can be classified into one of a plurality of clusters (step S23 in FIG. 6).

[0044] In this embodiment, the observation data x n Since each of the binary variables that are elements of can be considered to follow a Bernoulli distribution, a mixed Bernoulli distribution can be used as the mixed probability distribution represented by the parameter data 42. As machine learning, a likelihood estimation method using the EM (Expectation-Maximization) algorithm can be used.

[0045] According to the maximum likelihood estimation method using the EM algorithm of the mixed Bernoulli distribution, the observed data x n The proportion of people who belong to the kth cluster is the burden rate γ nk It can be calculated as follows: nkSince is expressed by the above formula (1), the clustering unit 46 calculates the observed data x n The burden rate γ n1 ,…,γ nK The users can be classified into a cluster corresponding to the highest burden rate among the above (step S23 in FIG. 6).

[0046] The lookup table 43 in the data storage unit 41 has a correspondence relationship between clusters and the amount of computational resources allocated to them. The resource allocation unit 47 refers to the lookup table 43 and calculates the amount of computational resources allocated to the clusters. n The resource allocation unit 37 then determines the amount of computational resources that is associated with the cluster into which the UE 11 is classified (step S24 in FIG. 6). The amount of computational resources that can be allocated may be, for example, the number of processor cores that are allocated. Next, the resource allocation unit 37 allocates the determined amount of computational resources to the UE 11. n (Step S25 in FIG. 6).

[0047] As described above, the computational resource allocation function 40 of the UPF 25 allocates the computational resources to each UE 11. n Observation data x consisting of a binary time series representing the frequency of measurements for each communication node n Using the machine learning, we perform clustering (cluster classification) based on a mixture probability distribution with maximum likelihood parameters estimated by machine learning, and the observed data x n Since the amount of computational resources assigned to the cluster into which the UEs 111 to 11 are classified is determined, N Even if communication (transmission and reception of communication signals) is performed with each of the devices at different frequencies, the computational resources of the UPF 25 can be efficiently allocated.

[0048] Next, the parameter generating device 26 will be described in detail below with reference to Fig. 9. Fig. 9 is a flowchart showing an example of a procedure for parameter generation processing by the parameter generating device 26.

[0049] The training data set 52A is the training data x corresponding to the observation data of FIG. nThe training data set 52B includes training data x corresponding to the observation data in FIG. n The training data x n Each of the parameters 52A and 52B is associated with an allocation amount of computational resources in advance. A method for generating parameter data 54A based on training data set 52A will be described below. A method for generating parameter data 54B based on training data set 52B is basically the same as a method for generating parameter data 54A based on training data set 52A, so a detailed description thereof will be omitted.

[0050] The machine learning unit 50 reads out the training data set 52A from the training data storage unit 51 (step S31 in FIG. 9), and executes machine learning based on a mixture probability distribution using the training data set 52A to estimate a maximum likelihood parameter of the mixture probability distribution (step S32 in FIG. 9). After that, the machine learning unit 50 stores a parameter group of the mixture probability distribution having the estimated maximum likelihood parameter as parameter data 54A in the parameter storage unit 53 (step S33 in FIG. 9).

[0051] 9, a maximum likelihood estimation method using the EM algorithm of a mixed Bernoulli distribution can be performed. The procedure of the maximum likelihood estimation method using the EM algorithm will be described below.

[0052] N binary vectors x1, x2, …, x N A set of binary data sets X is denoted by K parameter vectors μ1, μ2, …, μ K The set of binary data sets X and parameter data sets M are expressed as follows (T is the transpose symbol):

[0053]

number

[0054] Here, the n-th binary vector x nand the kth parameter vector μ k Each of these is represented as a vector with D elements as follows:

[0055]

number

[0056] Binary vector x n Element x of n Each of (i) (i=1,2,…,D) is a binary variable that can take the value “0” or “1”. Each binary vector x n are independently generated by the following mixed Bernoulli distribution:

[0057]

number

[0058] Here, π is a parameter data set consisting of K mixture ratios, which can be expressed as follows:

[0059]

number

[0060] The likelihood function P(X|M,π) is expressed by the following equation.

[0061]

number

[0062] The log-likelihood function lnP(X|M,π) is expressed by the following equation.

[0063]

number

[0064] The EM algorithm is a method for finding the maximum likelihood solution of the parameter data set M,π of a mixed Bernoulli distribution so that the likelihood (the likelihood function P(X|M,π) or the logarithmic likelihood function lnP(X|M,π)) is maximized. Below, we will explain the procedure for maximum likelihood estimation using the EM algorithm.

[0065] First, execute the initial step. That is, the parameter data M={μ1, μ2, ..., μ K},π={π1,π2,…,π K}, and initialize the likelihood.

[0066] Next, the E-step is performed. That is, a value called the Responsibility γ is calculated using the current parameter set M,π according to the following formula: nk Calculate the burden rate γ nk is a binary vector x n represents the proportion of clusters belonging to the kth cluster.

[0067]

number

[0068] Next, execute the M step. That is, the current burden rate γ nk Update the parameter set,M,π,using,.

[0069]

number

[0070] Next, it is determined whether a predetermined convergence condition is satisfied. Specifically, if the likelihood converges within a predetermined numerical range, if the parameter set M,π converges, or if both the likelihood and the parameter set M,π converge, it is determined that the convergence condition is satisfied. If it is determined that the convergence condition is not satisfied, the process returns to step E and repeats the calculation. If the predetermined convergence condition is not satisfied even after repeating the calculation a predetermined number of times or for a certain period of time, it is possible to change any or all of the parameter set M,π and the number of clusters K, and execute the processing from the initial step.

[0071] Various embodiments according to the present disclosure have been described above with reference to the drawings, but these embodiments are merely examples, and various forms other than these embodiments are possible. For example, in the above embodiment, the communication network system 1 is a mobile communication system conforming to the fifth generation (5G) communication standard, but is not limited thereto. The computational resource allocation functions 30, 40 and the parameter generation device 26 can also be applied to components of a mobile communication system conforming to the fourth generation (4G) communication standard or a communication standard after the fifth generation, or components of a fixed communication system other than a mobile communication system or a non-terrestrial network (NTN) system.

[0072] It should be understood that the above-described embodiment can be appropriately modified, added, and improved without departing from the spirit and scope of the present invention. The scope of the present invention should be interpreted based on the description of the claims, and should be understood to include equivalents thereof. [Explanation of symbols]

[0073] 1: Communication network systems, 101-10 N :Subscriber, 111~11 N : User equipment (UE), 15: Data network (DN), 201-20 N: base station, 21: communication system, 22: access mobility management function / security anchor function (AMF / SEAF), 23: integrated data management unit (UDM), 24: integrated data repository unit (UDR), 25: user plane function (UPF), 26: parameter generation device, 30: computational resource allocation function, 31: data storage unit, 32: parameter data, 33: lookup table, 35: frequency measurement unit, 36: clustering unit, 37: resource allocation unit, 40: computational resource allocation Functions, 41: data storage unit, 42: parameter data, 43: lookup table, 45: frequency measurement unit, 46: clustering unit, 47: resource allocation unit, 50: machine learning unit, 51: training data storage unit, 52A, 52B: training data set, 53: parameter storage unit, 54A, 54B: parameter data, 100: information processing device, 101: processor, 102: random access memory (RAM), 103: non-volatile memory, 104: large capacity memory, 105: input / output interface.

Claims

1. a frequency measurement unit that measures a frequency of communication between a plurality of communication nodes and each of the communication nodes and generates observation data consisting of a binary time series that represents the measured frequency for each of the communication nodes; a clustering unit that classifies the observation data into one of a plurality of clusters based on a mixture probability distribution having a maximum likelihood parameter estimated by machine learning; a resource allocation unit that allocates, to each communication node, a computational resource of an amount corresponding to the cluster into which the observable data is classified; A communication device comprising:

2. The communication device according to claim 1 , wherein the mixed probability distribution is a mixed Bernoulli distribution.

3. The communication device according to claim 1 , wherein the machine learning includes maximum likelihood estimation using an Expectation-Maximization (EM) algorithm.

4. 2. The communication device according to claim 1, A lookup table is further provided to define a correspondence relationship between the cluster and the amount of the computing resource allocated to the cluster; The resource allocation unit refers to the lookup table to determine an allocation amount to be allocated to each communication node.

5. 2. The communication device according to claim 1, each of the plurality of communication nodes is a mobile terminal that transmits a location registration request signal to a mobile communication system; The frequency measurement unit measures a frequency of transmission of the location registration request signal as a frequency of the communication.

6. The communications device of claim 1 , wherein the computational resources include a plurality of processor cores.

7. a training data storage unit storing training data consisting of a binary time series representing the frequency of communication between a predetermined number of communication nodes and each of the communication nodes; a machine learning unit that performs machine learning based on a mixture probability distribution using the training data set read from the training data storage unit to estimate maximum likelihood parameters of the mixture probability distribution; a parameter storage unit for storing data of the estimated maximum likelihood parameters; Equipped with The training data for each communication node is associated in advance with an allocated amount of computational resources. A parameter generating device comprising:

8. 8. The parameter generating device according to claim 7, wherein the mixed probability distribution is a mixed Bernoulli distribution.

9. 8. The parameter generating device according to claim 7, wherein the machine learning includes maximum likelihood estimation using an Expectation-Maximization (EM) algorithm.

10. 8. A parameter generating device according to claim 7, wherein each of the plurality of communication nodes is a mobile terminal that transmits a location registration request signal to a mobile communication system, and the frequency of communication is a frequency of transmission of the location registration request signal.

11. A communication device according to any one of claims 1 to 6, A parameter generating device according to any one of claims 7 to 10, A communication system comprising:

12. 1. A method for allocating computational resources in a communication device, comprising: measuring a frequency of communication between a plurality of communication nodes and each of the communication nodes; generating observation data consisting of a binary time series representing the measured frequency for each of the communication nodes; classifying the observed data into one of a plurality of clusters based on a mixture probability distribution having a maximum likelihood parameter estimated by machine learning; allocating, to each of the communication nodes, a computational resource of an amount corresponding to the cluster into which the observable data was classified; 23. A method comprising:

13. The method of claim 12 , wherein the mixture probability distribution is a mixture of Bernoulli distributions.

14. The method of claim 12 , wherein the machine learning includes maximum likelihood estimation using an Expectation-Maximization (EM) algorithm.

15. The method according to claim 12 , further comprising the step of determining an allocation amount to be allocated to each of the communication nodes by referring to a lookup table that defines a correspondence relationship between clusters and allocation amounts of computing resources.

16. A computer program for a communication device having one or more processors, the computer program being configured, when executed by said processor or processors, to cause said communication device to carry out a method according to any one of claims 12 to 15.

17. reading a training data set from a training data storage unit in which training data consisting of a binary time series representing a frequency of communication between a plurality of communication nodes and each of the communication nodes is stored; Using the retrieved training data set, perform machine learning based on a mixture probability distribution to estimate maximum likelihood parameters of the mixture probability distribution; storing the estimated maximum likelihood parameter data in a parameter storage unit; Equipped with The training data for each communication node is associated in advance with an allocated amount of computational resources. A method comprising:

18. 20. The method of claim 17, wherein the mixture probability distribution is a mixed Bernoulli distribution.

19. The method of claim 17 , wherein the machine learning includes maximum likelihood estimation using an Expectation-Maximization (EM) algorithm.

20. 18. The method according to claim 17, wherein each of the plurality of communication nodes is a mobile terminal that transmits a location registration request signal to a mobile communication network system, and the frequency of transmitting the location registration request signal is the frequency of the communication.

21. A computer program for a parameter generating device having one or more processors, the computer program being configured, when executed by said processor or processors, to cause said parameter generating device to perform a method according to any one of claims 17 to 20.

Images