Communication management device and communication management method

Abstract

To grasp an access tendency of a user for each base station for communication management of an entire communication network.SOLUTION: A communication management device 1 includes: an output symbol acquisition unit 10 that obtains an output symbol sequence including first identification information of a first web site that a communication terminal 2 accesses via a base station BS; an operation unit 12 that determines a probability that the communication terminal 2 accesses the first web site via each base station BS by using a learned hidden Markov model in which a parameter of the hidden Markov model is estimated beforehand, the hidden Markov model having each base station BS as a finite set of output symbols and having, as a finite set of observable output symbols, identification information of each web site that the communication terminal 2 accesses via each base station BS; and a determination unit 13 that determines, based on the determined access probability, the base station BS through which the communication terminal 2 most probably accesses the first web site.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a communication management device and a communication management method. [Background technology]

[0002] Conventionally, there has been known a technology that can identify a file on a cloud to which each communication terminal has made an access request by analyzing an access log of the cloud or the like. According to the technology disclosed in Patent Document 1, an analysis device that centrally manages a server unit such as a web server or multiple servers records an access log, and analyzes the access tendency of each of multiple communication terminals that accessed a specific website or the like. The results of such an analysis of access tendency are utilized by a provider of a service such as a website to provide a more personalized service to users.

[0003] However, the access log disclosed in Patent Document 1 is not a log that records access to any multiple websites that each communication terminal can access. Moreover, the access log disclosed in Patent Document 1 does not record which website was accessed from each base station that switches as the communication terminal moves on the radio access network (RA) or core network side. Therefore, it has been difficult to utilize the access log disclosed in Patent Document 1 and its analysis results for communication management of the entire communication network, for example, edge computing exemplified by MEC (Multi-access Edge Computing). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2018-156385 A Summary of the Invention [Problem to be solved by the invention]

[0005] As described above, according to the conventional technology, it is not possible to grasp the access trends of users on a base station basis because of communication management of the entire communication network.

[0006] The present invention has been made to solve the above-mentioned problems, and has an object to grasp the access trends of users on a base station basis for communication management of the entire communication network. [Means for solving the problem]

[0007] In order to solve the above-mentioned problems, the communication management device of the present invention comprises an output symbol acquisition unit configured to acquire an output symbol sequence including first identification information of a first website accessed via a base station in a communication area in which a communication terminal is located; a memory unit configured to store a learned hidden Markov model in which parameters of the hidden Markov model are estimated in advance, in which each base station that switches as the communication terminal moves is a finite set of hidden states and the identification information of each website accessed by the communication terminal via each base station is a finite set of output symbols in which the identification information of each website accessed by the communication terminal via each base station is observable; a calculation unit configured to use the learned hidden Markov model to calculate the probability that the communication terminal will access the first website having the first identification information acquired by the output symbol acquisition unit via each base station; a determination unit configured to determine the base station with the highest probability for the communication terminal to access the first website based on the probability of the communication terminal accessing the first website via each base station calculated by the calculation unit; and an output unit configured to output information of the determined base station with the highest probability as communication control information.

[0008] In addition, in the communication management device of the present invention, the output symbol acquisition unit may acquire, as learning data, an output symbol sequence indicating identification information of each of a plurality of websites including the first website accessed by the communication terminal via each of the base stations, and further include a learning unit configured to estimate the parameters of the hidden Markov model including a state transition probability distribution which is the probability that the connection of the communication terminal switches from a specified base station to another base station, and a symbol output probability distribution which is the probability that the communication terminal accesses a website of specified identification information via each of the base stations, maximizing the likelihood for the learning data, and the learned hidden Markov model may include the parameters estimated by the learning unit.

[0009] In order to solve the above-mentioned problems, the communication management device of the present invention may further include a first acquisition unit configured to acquire information about a cloud at a cloud site that is the shortest physical distance from a base station at which the communication terminal has the highest probability of accessing the first website, among the clouds at multiple cloud sites that provide the first website, based on the communication control information, and a communication control unit configured to instruct a core network of a specified communication standard to connect the communication terminal to the cloud at the cloud site, in accordance with the cloud information at the cloud site acquired by the first acquisition unit.

[0010] In addition, in the communication management device of the present invention, the output symbol acquisition unit may include a second acquisition unit configured to acquire a presence log indicating the base station of the communication area in which the communication terminal having the subscriber identification information was located and the time when the communication terminal was located from a core network of a specified communication standard, a third acquisition unit configured to acquire from the core network a terminal IP address assigned to the subscriber identification information included in the presence log acquired by the second acquisition unit, and a fourth acquisition unit configured to acquire from the core network an access log including the access time for each subscriber identification information to a website having a site IP address which is the identification information, based on the terminal IP address acquired by the third acquisition unit.

[0011] In addition, in the communication management device of the present invention, the second acquisition unit may acquire the presence log from an access and mobility management device provided in the core network, and the third acquisition unit may acquire the terminal IP address from a session management function provided in the core network.

[0012] In addition, in the communication management device of the present invention, the fourth acquisition unit may acquire the site IP address and access time for each communication terminal by referring to an access log for each website having an IP address that is the identification information for each terminal IP address, which is stored in a domain name system provided in the core network.

[0013] In order to solve the above-mentioned problems, a communication management method of the present invention includes an output symbol acquisition step of acquiring an output symbol sequence including first identification information of a first website accessed via a base station in a communication area in which a communication terminal is located; a storage step of storing in a storage unit a trained hidden Markov model in which parameters of the hidden Markov model are estimated in advance, in which each base station that switches as the communication terminal moves is a finite set of hidden states and the identification information of each website accessed by the communication terminal via each base station is a finite set of output symbols in which the identification information of each website accessed by the communication terminal via each base station is observable; a calculation step of using the trained hidden Markov model to calculate the probability that the communication terminal will access the first website having the first identification information acquired in the output symbol acquisition step via each base station; a determination step of determining a base station with the highest probability from which the communication terminal will access the first website based on the probability that the communication terminal will access the first website via each base station calculated in the calculation step; and an output step of outputting information of the base station with the highest probability determined as communication control information.

[0014] In addition, in the communication management method of the present invention, the output symbol acquisition step may include a learning step of acquiring, as learning data, an output symbol sequence indicating identification information of each of a plurality of websites including the first website accessed by the communication terminal via each of the base stations, and estimating the parameters of the hidden Markov model including a state transition probability distribution which is the probability that the connection of the communication terminal switches from a predetermined base station to another base station and a symbol output probability distribution which is the probability that the communication terminal accesses a website having predetermined identification information via each of the base stations, maximizing the likelihood for the learning data, and the learned hidden Markov model may include the parameters estimated in the learning step.

[0015] In addition, the communication management method of the present invention may further include a first acquisition step of acquiring, based on the communication control information, information on a cloud at a cloud site that is the shortest physical distance from a base station at which the communication terminal has the highest probability of accessing the first website, among the clouds at a plurality of cloud sites that provide the first website, and a communication control step of instructing a core network of a specified communication standard to connect the communication terminal to the cloud at the cloud site, in accordance with the cloud information at the cloud site acquired in the first acquisition step.

[0016] In addition, in the communication management method of the present invention, the output symbol acquisition step may include a second acquisition step of acquiring, from a core network of a specified communication standard, a presence log indicating a base station in a communication area in which the communication terminal having subscriber identification information was present and the time when the communication terminal was present, a third acquisition step of acquiring, from the core network, a terminal IP address assigned to the subscriber identification information included in the presence log acquired in the second acquisition step, and a fourth acquisition step of acquiring, from the core network, an access log including the access time to a website having a site IP address which is the identification information for each subscriber identification information, based on the terminal IP address acquired in the third acquisition step.

[0017] In addition, in the communication management method of the present invention, the second acquisition step may acquire the presence log from an access and mobility management device provided in the core network, and the third acquisition step may acquire the terminal IP address from a session management function provided in the core network.

[0018] In addition, in the communication management method of the present invention, the fourth acquisition step may acquire a site IP address and access time for each communication terminal by referring to an access log for each website having an IP address that is the identification information, for each terminal IP address, stored in a domain name system provided in the core network. Effect of the Invention

[0019] According to the present invention, a trained hidden Markov model is used to calculate the probability that a communication terminal will access a first website having first identification information via each base station, and a base station with the highest probability of accessing the first website by the communication terminal is determined based on the calculated probability. This makes it possible to grasp the access trends of users on a base station-by-base station basis for communication management of the entire communication network. [Brief description of the drawings]

[0020] [Figure 1] FIG. 1 is a block diagram showing a configuration of a communication management system including a communication management device according to an embodiment of the present invention. [Diagram 2] FIG. 2 is a diagram for explaining the data structure of the visit log that the communication management device according to this embodiment acquires from the AMF. [Diagram 3] FIG. 3 is a diagram for explaining the data structure of a table in which terminal IP addresses that the communication management device according to the present embodiment acquires from the SMF are stored. [Figure 4] FIG. 4 is a diagram for explaining the data structure of the access log that the communication management device according to the present embodiment acquires from the DNS. [Diagram 5] FIG. 5 is a diagram for explaining the learning unit and the calculation unit included in the communication management device according to the present embodiment. [Figure 6] FIG. 6 is a diagram for explaining the learning unit and the calculation unit included in the communication management device according to the present embodiment. [Figure 7] FIG. 7 is a diagram for explaining a data structure of the second storage unit stored in the communication management device according to the present embodiment. [Figure 8] FIG. 8 is a block diagram showing a hardware configuration of the communication management device according to the present embodiment. [Figure 9] FIG. 9 is a flowchart showing the learning process of the communication management device according to the present embodiment. [Figure 10]FIG. 10 is a flowchart showing the calculation process of the communication management device according to the present embodiment. [Figure 11] FIG. 11 is a sequence diagram showing a communication control process by the communication management system according to the present embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0021] DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Preferred embodiments of the present invention will now be described in detail with reference to FIGS.

[0022] [Communication management system configuration] 1 is a block diagram showing a configuration of a communication management system including a communication management device 1 according to an embodiment of the present invention. The communication management system according to this embodiment is provided in a 5G mobile communication network, and analyzes the access trends of websites accessed by a communication terminal 2 performing mobile communication for each base station BS0 to BSn. Furthermore, based on the analysis result, the communication management system performs communication control so that the communication terminal 2 communicates with a cloud 50 at a cloud site that is the shortest physical distance from a transit base station BS0 that has the highest probability of accessing a specific website.

[0023] The communication management system includes a communication management device 1 including an analysis device 1A and a communication control device 1B, a communication terminal 2 compatible with 5G mobile communication, a plurality of base stations BS0 to BSn, a core network 4, and a plurality of clouds 50.

[0024] The communication terminal 2 includes an SIM 20 and is realized as a mobile communication terminal such as a smartphone, a tablet computer, a laptop computer, or the like. The contract profile of the SIM 20 stores subscriber identification information of the user, and includes identifier information such as an International Mobile Subscriber Identity (IMSI) number assigned to a mobile phone line contract, a telephone number of the user who is a subscriber (Mobile Subscriber International Subscriber Directory Number (MSISDN)), and a SIM card number (Integrated Circuit Card Identifier (ICCID)). The communication terminal 2 is uniquely identified by the IMSI of the SIM 20.

[0025] A terminal IP address that uniquely identifies the terminal is further assigned to the communication terminal 2. A local IP address or a global IP address can be used as the terminal IP address. There are m communication terminals 2 (m is a positive integer).

[0026] The base stations BS0 to BSn (n is a positive integer) are composed of wireless base stations compatible with the 5G system, and relay communications between communication terminals 2 present in their communication areas and a core network 4. The base stations BS0 to BSn are connected to the core network 4 via a network L such as a backhaul link. Hereinafter, when there is no need to distinguish between the base stations BS1 to BSn, they may be collectively referred to as base station BS.

[0027] The core network 4 is connected to the communication management device 1 via a network NW such as a LAN or WAN. The core network 4 includes a plurality of UPFs (User Plane Functions) 40 in the U-plane, a plurality of AMFs (Access and Mobility Management Functions) 41 which are nodes in the C-plane, a SMF (Session Management Function) 42, and a DNS (Domain Name System) 43. Note that other functions included in the core network 4 are not shown in the figure.

[0028] The UPF 40 is a user plane function that processes data packets between a radio access network (RAN) and a data network such as the Internet. In this embodiment, a single C-plane can control a plurality of UPFs 40. The plurality of UPFs 40 are devices that are physically located at different positions in a 5G mobile communication network. Each UPF 40 is connected to each cloud 50 on the Internet, and the communication terminal 2 can access a website provided by each cloud 50 via each UPF 40.

[0029] The AMF 41 is an access and mobility management device, and manages the registration and wireless connection of the communication terminal 2 that has moved to each area. In this embodiment, a plurality of AMFs 41 are provided, each of which has a communication interface 41a for communicating with the communication management device 1. FIG. 2 shows a presence log table 410 managed by the AMF 41. The presence log is information in which the IMSI of the communication terminal 2, the base station BS of the communication area in which the communication terminal 2 has been present, and a timestamp which is the presence time x are associated with each other. For example, in the IMSI1 of the communication terminal 2, the base stations BS0 to BSn which have been switched in response to the movement of the communication terminal 2, and the time x at which the communication terminal 2 has been present at each of the base stations BS0 to BSn are recorded as timestamps.

[0030] The SMF 42 is a session management function, and performs the establishment, modification, release, etc. of a PDU (Packet Data Unit) session between the communication terminal 2 and a data network such as the Internet. The SMF 42 according to this embodiment includes a communication interface 42a for communicating with the communication management device 1. Furthermore, the SMF 42 stores information in which the IMSI of the communication terminal 2, a terminal IP address assigned to the IMSI, and the UPF 40 used by the communication terminal 2 are associated with each other, as shown in table 420 in Fig. 3 .

[0031] The DNS 43 is a domain name system that associates domain names with IP addresses and manages them on an IP network. The DNS 43 according to this embodiment includes a communication interface 43a and communicates with the communication management device 1 via the communication interface 43a. The DNS 43 according to this embodiment also stores an access log of the site IP address of a website requested by the communication terminal 2. As shown in FIG. 4, the DNS 43 stores a table 430 that associates the terminal IP address of the communication terminal 2, the site IP address, and a timestamp indicating the time t at which the site IP address was accessed by the terminal IP address.

[0032] The clouds 50 provide predetermined websites, web applications, etc., and each cloud 50 has a cloud base that is geographically separated from one another. A cloud base refers to a geographical location or area where physical devices constituting each cloud 50 are located. Each cloud 50 can be an edge server such as a server, a data center, or an MEC server. In the example of FIG. 1, cloud 1, cloud 2, and cloud Z (Z is a positive integer) are each distributed and located at geographically separated locations as cloud bases that provide a first website of a site IP address "site IP1."

[0033] For example, as shown in FIG. 1, a communication terminal 2 to which IMSI1 and terminal IP1 are assigned is present in the communication area of ​​a base station BS0. The communication terminal 2 connects to one of the clouds 50 via the base station BS0 and UPF 40, and accesses the first website "www.111" of the site IP1. In the example of FIG. 1, the distance between the base stations BS0, UPF 40, and cloud 50 is as follows: [distance R1 between base stations BS0, UPF_1, and cloud 1]>[distance R2 between base stations BS0, UPF_2, and cloud 2]>[distance RZ between base stations BS0, UPF_Z, and cloud Z]. Therefore, when the communication terminal 2 accesses the first website of the site IP1, it is possible to suppress the delay time using MEC by communicating with cloud Z via UPF_N, which takes the shortest route from the base station BS0.

[0034] On the other hand, depending on the access tendency of the user, the communication terminal 2 may access a specific website more frequently via a specific base station BS. For example, in recent remote work, a user linked to the communication terminal 2 accesses a specific website of the company where the user works while staying at home. In such a case, the user accesses the company's website more frequently via a base station BS close to the home, rather than a base station BS at the company's location. In this way, the communication management system according to the present embodiment learns the access tendency of the data network for each user, i.e., for each communication terminal 2, using a machine learning model, and determines the intermediate base station BS with the highest probability that the communication terminal 2 will access the specific website. Furthermore, the communication management system performs communication control so that the communication terminal 2 communicates with the cloud 50 of the cloud base located on the shortest route from the determined base station BS.

[0035] [Function block of communication management device] The communication management device 1 includes an analysis device 1A and a communication control device 1B. The analysis device 1A obtains the access tendency of each communication terminal 2 for each base station BS as a machine learning model using a hidden Markov model. The communication control device 1B performs communication control based on the analysis result obtained by the analysis device 1A so that the communication terminal 2 communicates with the cloud 50 at the cloud site with the shortest distance from the base station BS.

[0036] The analysis device 1A includes an output symbol acquisition unit 10, a learning unit 11, a calculation unit 12, a determination unit 13, a first storage unit 14, and an output unit 15. The output symbol acquisition unit 10 includes a second acquisition unit 10A, a third acquisition unit 10B, and a fourth acquisition unit 10C.

[0037] The output symbol acquiring unit 10 acquires an output symbol sequence including a site IP address (first identification information) of a first website accessed via a base station BS in a communication area where the communication terminal 2 is located. In addition, the output symbol acquiring unit 10 acquires, as learning data, an output symbol sequence indicating a site IP address which is identification information of each of a plurality of websites including the first website accessed by the communication terminal 2 via each of the base stations BS to BSn.

[0038] The second acquisition unit 10A acquires a visit log indicating base stations BS0 to BSn of a communication area in which a communication terminal 2 having an IMSI that is subscriber identification information is located and a visit time x from the core network 4. More specifically, the second acquisition unit 10A refers to a table 410 (FIG. 2) stored in the AMF 41 of the core network 4 and acquires a visit log for each communication terminal 2.

[0039] The third acquisition unit 10B acquires, from the core network 4, a terminal IP address assigned to the IMSI included in the presence log of the communication terminal 2 acquired by the second acquisition unit 10A. More specifically, the third acquisition unit 10B refers to a table 420 ( FIG. 3 ) stored in the SMF 42 of the core network 4, and acquires the terminal IP address assigned to the IMSI of the communication terminal 2.

[0040] The fourth acquisition unit 10C acquires an access log including an access time to each website having a site IP address for each IMSI from the core network 4, based on the terminal IP address of the communication terminal 2 acquired by the third acquisition unit 10B. More specifically, the fourth acquisition unit 10C refers to the table 430 (FIG. 4) stored in the DNS 43 included in the core network 4, and acquires the access log by using the terminal IP address of the communication terminal 2 as a key.

[0041] The information acquired by the output symbol acquisition unit 10, which is equipped with the second acquisition unit 10A, the third acquisition unit 10B, and the fourth acquisition unit 10C, makes it possible to determine which website, having which site IP address, a communication terminal 2, to which an IMSI and terminal IP address have been assigned, accessed at the time corresponding to the timestamp in which the communication terminal 2 is located in the communication area of ​​which base station BS.

[0042] The learning unit 11 estimates parameters of a hidden Markov model that maximizes the likelihood for the learning data acquired by the output symbol acquisition unit 10, including a state transition probability distribution A, which is the probability that the connection of the communication terminal 2 switches from a specified base station to another base station, and a symbol output probability distribution B, which is the probability that the communication terminal 2 accesses a website of a specified site IP address via each base station BS.

[0043] As described above, the communication management device 1 according to the present embodiment employs a hidden Markov model (HMM) as a machine learning model. The hidden Markov model is a model that probabilistically captures output symbols, which are observation values ​​that depend on state variables that change according to a Markov process. In other words, it is a model that can estimate which state a random variable that moves between multiple states is in at each point in time.

[0044] More specifically, the learning unit 11 uses a hidden Markov model in which each base station BS0 to BSn that switches as the communication terminal 2 moves is treated as a finite set of hidden states, and the site IP addresses of websites accessed by the communication terminal 2 via each base station BS0 to BSn are treated as a finite set of observable output symbols.

[0045] FIG. 5 is a diagram for explaining the hidden Markov model employed in this embodiment. As shown in FIG. 5, a Left-to-Right HMM is used in this embodiment. The Left-to-Right HMM is a model based on the assumption that states always transition from left to right, and that there is no ergodicity, that is, once a state transition occurs to the next state, the state cannot return to the previous state. Each circle shown in FIG. 5 indicates a base station BS0 to BSn in which the communication terminal 2 is present, that is, a hidden state that cannot be directly observed, and a set of states S={S 1 ,S 2 ,…,S n}.

[0046] In the example of Fig. 5, the communication terminal 2 moves from the base station BS0 in the initial state to the base station BSn in the final state, and the loop of each state shows a self-loop. In Fig. 5, the output probability of the site IP address of the website accessed by the communication terminal 2 via each base station BS0 to BSn in the hidden state, that is, the output symbol sequence that can be observed only, is b i (1), b i (2),…,b i This is shown in (k). i is the probability that the communication terminal 2 accesses each of k types of preset site IP addresses 1 to k via each of the base stations BS0 to BSn, which are respectively set as the base stations S. In this way, it can be considered that the site IP address represented by the symbol O is output from the base station BS represented by the state S.

[0047] The parameters of the hidden Markov model are the set of hidden states S mentioned above, as well as the type of output symbols K = {1, 2, ..., k} and the set of output symbols O = {O 1 ,O 2 ,…,O m}, the set of initial state probabilities π={π i}, a set of final states, a set of state transition probabilities A={a ij}, and the set of symbol output probabilities B={b i (O t )}.

[0048] Initial state probability π i The sum of these satisfies 1, as expressed by the following equation (1). This is because the Left-to-Right HMM starts with only the initial state i=0.

number

[0049] Regarding the final state, in a Left-to-Right HMM, there is only one final state.

[0050] State transition probability a ij is the state S i From state S j is the probability of transitioning to a certain state S i All possible states S to which we can transition j The sum of state transition probabilities to satisfies 1, as shown in the following equation (2).

number

[0051] Symbol output probability b i (O t ) is the state S i Output symbol O t The probability of outputting a state S i The sum of the symbol output probabilities of all symbols that can be output when transitioning from t is the output symbol observed at time t.

number

[0052] The hidden Markov model defined as above is represented as λ=(A,B,π). The learning unit 11 estimates parameters of the state transition probability distribution A and the symbol output probability distribution B of the hidden Markov model λ. In the hidden Markov model, since the hidden state sequence cannot be directly observed from the output symbol sequence, it is difficult to directly perform maximum likelihood estimation. Therefore, the learning unit 11 estimates these parameters by repeated calculations based on a maximum likelihood method called the EM (Expectation-maximization) algorithm. The learning unit 11 estimates parameters of the state transition probability distribution A and the symbol output probability distribution B by using the Baum-Welch algorithm, which is one of the EM algorithms well known as being particularly suitable for estimating parameters for the hidden Markov model.

[0053] The learning unit 11 executes the following first to fifth steps of the Baum-Welch algorithm learning steps.

[0054] [First step] First, the learning unit 11 sets initial values ​​of the state transition probability distribution A and the symbol output probability B. The learning unit 11 can use any value as the initial value.

[0055] [Second step] Next, the output symbol acquisition unit 10 obtains the output symbol sequence O t ={O 1 ,O 2 ,…,O m} is determined. The learning data is an output symbol sequence of the site IP addresses of websites accessed by the communication terminal 2 via each base station BS, which is acquired by the output symbol acquisition unit 10. As shown in FIG. 6, the output symbol acquisition unit 10 compares the timestamp of the presence log of the communication terminal 2 with the timestamp of the access log of the website, specifies which base station BS and which website were accessed, and determines the symbol output sequence O at time t. t Determine.

[0056] Specifically, as shown in FIG. 6, the output symbol acquisition unit 10 receives the time stamp x of the presence log of the communication terminal 2 stored in the table 410 of the AMF 41. 1 ~x n The site IP address designation of the website by the communication terminal 2 and the access time t 1 ,t 2 ,…,t n and the output symbol sequence O t In addition, the output symbol acquisition unit 10 determines the state S i In the example of FIG. 6, the state S is identified according to the base station BS number 0 to n. 1 ~Condition S n is identified. Note that in state S i Alternatively, the learning unit 11 may specify the above.

[0057] [Third Step] Next, the learning unit 11 sets the value of grid, which is a variable of the Forward algorithm, to α t (i) is defined by the following formula (4), and each α t The following calculation is performed.

number

[0058] [Fourth step] Next, the learning unit 11 changes the value of grid, which is a variable of the backward algorithm, to β t (t) is defined by the following formula (5), and each β t The following calculation is performed.

number

[0059] [5th ​​step] Next, the learning unit 11 calculates the state S i From state S j Transition probability tot Calculate (i,j) and determine the state transition probability a ij and the symbol output probability b j Recalculate (O).

number

[0060] After that, the learning unit 11 repeats the above third to fifth steps, and determines the point at which the parameters do not change or the likelihood does not change as the convergence point, and adopts the state transition probability distribution A and the symbol output probability distribution B at that time as estimates. In the Baum-Welch algorithm, the transition probability Γ t (i,j) is the state transition probability a ij and the symbol output probability b j The procedure of calculating from (O) corresponds to the expectation (E) step of the EM algorithm. Also, in the fifth step, the symbol output probability b j (O) is the transition probability Γ t The procedure of recalculating from (i,j) corresponds to the maximization (M) step of the EM algorithm.

[0061] Returning to Figure 1, the calculation unit 12 uses the trained hidden Markov model to determine the probability that the communication terminal 2 will access the first website having the site IP1 acquired by the output symbol acquisition unit 10 via each of the base stations BS0 to BSn.

[0062] More specifically, the calculation unit 12 calculates, for example, the symbol output probability b i (1), b i (2),…,b i Each state S for each of the site IP addresses 1 to k corresponding to the k types of output symbols of (k) i Symbol output probability b i (1), b i (2),…,b iFor example, the probability that communication terminal 2 accesses IP address 1, IP address 2, ..., IP address k via base station BS0 is b 0 (1)=0.7, b 0 (2)=0.2,...,b 0 It is calculated that (k)=0.1. In this manner, the calculation unit 12 uses the trained hidden Markov model to calculate the probability that the communication terminal 2 will access each of the site IP addresses 1 to k via each of the base stations BS0 to BSn.

[0063] The determination unit 13 determines the base station BS through which the communication terminal 2 has the highest probability of accessing the first website, based on the probability that the communication terminal 2 will access the first website of the site IP1 via each of the base stations BS0 to BSn, calculated by the calculation unit 12. The determination unit 13 compares, for example, the symbol output probability values ​​of the base stations BS0 to BSn shown in FIG. 5, and determines which base station BS0 to BSn through which the communication terminal 2 has the highest probability of accessing the site IP1. In the example of FIG. 5, the base station BS through which the communication terminal 2 has the highest number of accesses to the site IP1 (k=1) has a symbol output probability b 0 (1)=0.7 base station BS 0 It is determined that.

[0064] The first storage unit 14 stores the trained Markov model whose parameters have been estimated by the training unit 11.

[0065] The output unit 15 outputs, as communication control information, information on the base station BS with the highest probability that the communication terminal 2 will access the first website of the site IP1, which is determined by the determination unit 13. The output unit 15 can send the communication control information to the communication control device 1B.

[0066] Next, the communication control device 1B included in the communication management device 1 includes a first acquisition unit 16, a second storage unit 17, and a communication control unit 18. Based on information about a site IP1 of a first website actually accessed by the communication terminal 2, the communication control device 1B performs communication control so that communication is performed with the communication terminal 2 via cloud 50, which is a cloud base located on the shortest route from the intermediate base station BS where the communication terminal 2 has the highest probability of accessing the site IP1.

[0067] Based on the communication control information from the output unit 15, the first acquisition unit 16 acquires information on the cloud 50 at the cloud base that is the shortest physical distance from the base station BS and that has the highest probability of accessing the first website by the communication terminal 2, among the multiple clouds 50 that provide the first website. The first acquisition unit 16 acquires information on the cloud 50 at the cloud base on the shortest route, from the second storage unit 17.

[0068] The second storage unit 17 stores the table 170 of Fig. 7. The table 170 stores the physical distances between each of the base stations BS0 to BSn, each of the UPF_1 to UPF_N, and cloud 1 to cloud Z which are cloud bases providing each of the sites IP1 to IPk. The table 170 is information shared by a plurality of communication terminals 2. Here, as an example, consider the case where the transit base station BS with the highest probability that the communication terminal 2 will access the site IP1 is base station BS0.

[0069] In this case, the first acquisition unit 16 compares the distance of 30 km between the base stations BS0-UPF_1-cloud 1, the distance of 20 km between the base stations BS0-UPF_2-cloud 2, and the distance of 3 km between the base stations BS0-UPF_Z-cloud Z, which are shown in the dashed line area of ​​the table 170 shown in FIG. 7. Note that clouds 1 to Z are each cloud bases of the site IP1. As shown in FIG. 7, the communication terminal 2 takes the shortest route when communicating with cloud Z from base station BS0 via UPF_N. Therefore, the first acquisition unit 16 acquires information on UPF_N from the second storage unit 17.

[0070] The communication control unit 18 instructs the core network 4 to set the connection destination of the communication terminal 2 to the cloud Z according to the information on the cloud Z acquired by the first acquisition unit 16. More specifically, the communication control unit 18 sends an instruction to the SMF 42 of the core network 4 to cause the communication terminal 2 related to IMSI1 to communicate via UPF_N. Through the communication control by the communication control unit 18, the communication terminal 2 can access the site IP1 from the base station BS0 via the shortest route by communicating with the cloud Z, which is the cloud base of the site IP1, from the base station BS0 via the UPF_N.

[0071] [Hardware configuration of communication management device] Next, an example of a hardware configuration for realizing the communication management device 1 having the above-mentioned functions will be described with reference to FIG.

[0072] As shown in FIG. 8, the communication management device 1 can be realized, for example, by a computer having a processor 102, a main memory device 103, a communication interface 104, an auxiliary memory device 105, and an input / output (I / O) 106 connected via a bus 101, and a program that controls these hardware resources.

[0073] The main memory device 103 stores in advance programs for the processor 102 to perform various controls and calculations. The processor 102 and the main memory device 103 realize the functions of the communication management device 1, such as the output symbol acquisition unit 10, learning unit 11, calculation unit 12, and decision unit 13 of the analysis device 1A, and the first acquisition unit 16 and communication control unit 18 of the communication control device 1B, shown in FIG.

[0074] The communication interface 104 is an interface circuit for connecting the communication management device 1 to various external electronic devices via a network.

[0075] The auxiliary storage device 105 is composed of a readable / writable storage medium and a drive for reading and writing various information such as programs and data from and to the storage medium. The auxiliary storage device 105 can use a semiconductor memory such as a hard disk or a flash memory as the storage medium.

[0076] The auxiliary storage device 105 has a program storage area for storing a learning program executed by the communication management device 1. The auxiliary storage device 105 also has an area for storing a communication control program for controlling communication of the communication terminal 2. The first storage unit 14 and the second storage unit 17 described in Fig. 1 are realized by the auxiliary storage device 105. Furthermore, for example, the auxiliary storage device 105 may have a backup area for backing up the above-mentioned data and programs.

[0077] The input / output I / O 106 is an input / output device that inputs signals from an external device and outputs signals to an external device.

[0078] The analytical device 1A and the communication control device 1B may each have a separate hardware configuration. In that case, each of the analytical device 1A and the communication control device 1B has the hardware configuration shown in FIG.

[0079] [Operation of the communication management device] Next, the operation of the communication management device 1 having the above-mentioned configuration will be described with reference to the flowcharts of Fig. 9 and Fig. 10. Fig. 9 is a flowchart showing a hidden Markov model learning process by the analysis device 1A. Fig. 10 is a flowchart showing a calculation process by the analysis device 1A using a learned hidden Markov model whose parameters have been estimated.

[0080] First, the learning process by the analysis device 1A will be described with reference to Fig. 9. First, the second acquisition unit 10A of the output symbol acquisition unit 10 acquires the presence log of the communication terminal 2 from the AMF 41 (step S1). The second acquisition unit 10A refers to the table 410 of the AMF 41, and acquires the base stations BS0 to BSn of the communication area in which the IMSI1 assigned to the communication terminal 2 is present, and the timestamp indicating the presence time.

[0081] Next, third acquisition unit 10B of output symbol acquisition unit 10 acquires the terminal IP address of communication terminal 2 from SMF 42 (step S2). Specifically, third acquisition unit 10B refers to table 420 of SMF 42 and acquires terminal IP1 assigned to IMSI1 of communication terminal 2.

[0082] Next, the fourth acquisition unit 10C of the output symbol acquisition unit 10 acquires an access log of the website by the communication terminal 2 from the DNS 43 (step S3). More specifically, the fourth acquisition unit 10C refers to the table 430 stored in the DNS 43, and acquires the designation of the site IP address of the website by the terminal IP1 acquired in step S2 and a time stamp.

[0083] Next, the output symbol acquisition unit 10 acquires learning data for estimating parameters of the hidden Markov model based on the data acquired in steps S1 to S3 (step S4). The output symbol acquisition unit 10 acquires learning data for estimating parameters of the hidden Markov model based on the time stamp x 1 ~x n The time stamp t of the access log acquired in step S3 corresponds to 1 ~t n The communication terminal 2 acquires, as learning data, an output symbol sequence of the site IP address of the website designated by the IP address.

[0084] Next, the learning unit 11 estimates parameters of a hidden Markov model that maximizes the likelihood for the learning data acquired in step S4 (step S5). Specifically, the learning unit 11 executes the processes from the first step to the fifth step of the Baum-Welch algorithm described above, and estimates parameters including the state transition probability distribution A and the symbol output probability distribution B.

[0085] The parameters estimated in step S5 are stored in the first storage unit 14 as a trained hidden Markov model (step S6).

[0086] Next, with reference to FIG. 10, a calculation process using a trained hidden Markov model by the analysis device 1A included in the communication management device 1 will be described.

[0087] First, the calculation unit 12 loads a trained hidden Markov model from the first storage unit 14 (step S10). Next, the output symbol acquisition unit 10 acquires an output symbol sequence of a site IP address of a website that the communication terminal 2 is accessing via base stations BS0 to BSn (step S11).

[0088] Specifically, the output symbol acquisition unit 10 can acquire the site IP1 of the first website being accessed via the base station BS0 in the communication area in which the communication terminal 2 is present. The output symbol acquired in step S11 is information including the IMSI1 of the communication terminal 2, the site IP1 of the first website specified by the communication terminal 2 in the timestamp of the presence log, and the timestamp of the access log. From this information, it is understood that the communication terminal 2 is currently accessing the first website of the site IP1.

[0089] Next, the calculation unit 12 uses the trained hidden Markov model to obtain the probability that the communication terminal 2 will access the first website having the site IP1 acquired by the output symbol acquisition unit 10 via each of the base stations BS0 to BSn (step S12). For example, as shown in FIG. 5, the calculation unit 12 calculates the probability that the communication terminal 2 will access the first website having the site IP1 acquired by the output symbol acquisition unit 10 in each state Si The probability value of accessing the first website (k=1) of site IP1 via each of base stations BS0 to BSn corresponding to the site IP1 is calculated as “0.7” for base station BS0, “0.3” for base station BS1, ..., and “0.1” for base station BSn-1.

[0090] Next, the determination unit 13 determines the base station BS with the highest probability that the communication terminal 2 will access the first website of the site IP1 based on the probability value calculated in step S12 (step S13). For example, in the case of Fig. 5, the determination unit 13 selects the base station BS0 because the base station BS0 has the highest probability value "0.7" of accessing the first website of the site IP1.

[0091] Next, the output unit 15 outputs the information on the base station BS0 determined in step S13 as communication control information (step S14). The above is the process in the analysis device 1A of the communication management device 1.

[0092] Next, the communication control process in the communication control device 1B included in the communication management device 1 will be described with reference to the operation sequence of the communication management system in Fig. 11. First, the communication terminal 2 communicates with the cloud 1 via the base station BS0 in the communication area in which it is located and further via UPF_1, and accesses the first website of the site IP1 (step S100). Note that in step S100, the communication terminal 2 is connected to the cloud 1 via UPF_1 that was randomly assigned when the communication terminal 2 started communication.

[0093] Next, the SMF42 included in the core network 4 notifies the communication management device 1 of information that the communication terminal 2 having IMSI1 is accessing the site IP1 via UPF_1 (step S101). The SMF42 makes the notification by referring to the table 420. The output symbol acquisition unit 10 of the communication management device 1 acquires the notification from the SMF42.

[0094] Next, the communication control device 1B included in the communication management device 1 acquires the communication control information output from the output unit 15 of the analysis device 1A (step S102). For example, information on the base station BS0 with which the communication terminal 2 has the highest probability of accessing the first website of the site IP1 is acquired as the communication control information.

[0095] Next, the first acquisition unit 16 of the communication control device 1B acquires information on the UPF_N connected to cloud Z, a cloud base with the shortest physical distance from the base station BS0, among the clouds 50 that provide the first website of the site IP1, based on the information on the base station BS0 acquired in step S102 (step S103). Specifically, the first acquisition unit 16 refers to the table 170 stored in the second storage unit 17, and selects the UPF_N with the shortest physical distance between the base station BS0-UPF40-cloud 50, among the clouds 1 to Z that are the cloud bases of the first website of the site IP1.

[0096] Next, the communication control unit 18 of the communication control device 1B sends a communication control instruction to the SMF 42 indicating that the communication terminal 2 related to IMSI1 should communicate using the UPF_N (step S104). Next, in response to the communication control instruction, the SMF 42 sends an instruction to the UPF_N to establish a session with the communication terminal 2 of IMSI1 (step S105). Next, the UPF_N that has received the session establishment instruction establishes a session with the communication terminal 2 (step S106).

[0097] After that, UPF_N sends a notification indicating that a session has been established with the communication terminal 2 to the SMF 42 (step S107). Next, the SMF 42 instructs the communication terminal 2 associated with IMSI1 to communicate with the UPF_N (step S108). After that, in response to the communication instruction, the communication terminal 2 communicates with the UPF_N and cloud Z from the base station BS0, and accesses the first website of the site IP1 (step S109).

[0098] As described above, the communication management device 1 according to this embodiment estimates parameters of a hidden Markov model in which each base station BS0 to BSn that switches as the communication terminal 2 moves is treated as a finite set of hidden states, and the site IP addresses of each website accessed by the communication terminal 2 via each base station BS0 to BSn is treated as a finite set of observable output symbols. Therefore, for communication management of the entire communication network, it is possible to grasp the access trends of users for each base station BS0 to BSn.

[0099] Moreover, the communication management device 1 according to the present embodiment determines the base station BS0 with which the communication terminal 2 has the highest probability of accessing the first website of the site IP1. Therefore, by using MEC, communication management can be performed so that the communication terminal 2 communicates with the cloud 50 at the cloud base that is the shortest distance from the base station BS0. As a result, transmission delays in the communication of the communication terminal 2 can be suppressed.

[0100] In the above embodiment, a communication management system that complies with 5G has been exemplified, but the communication management system may also be one that complies with 6G or the like.

[0101] In addition, in the above-mentioned communication management device 1, the learning process related to the estimation of the parameters of the hidden Markov model is performed by the Baum-Welch algorithm. However, the estimation of the parameters of the hidden Markov model is not limited to the Baum-Welch algorithm. For example, structured variational inference may be used.

[0102] In addition, in the above-mentioned communication management device 1, the analysis device 1A and the communication control device 1B are provided in one device. However, the analysis device 1A and the communication control device 1B may be configured as independent devices distributed on a network. Also, each functional unit of the analysis device 1A and the communication control device 1B may be configured to be distributed on a network.

[0103] The above describes the embodiments of the communication management device and communication management method of the present invention, but the present invention is not limited to the described embodiments, and various modifications that a person skilled in the art can imagine may be made within the scope of the invention described in the claims. [Explanation of symbols]

[0104] 1...communication management device, 1A...analysis device, 1B...communication control device, 10...output symbol acquisition unit, 10A...second acquisition unit, 10B...third acquisition unit, 10C...fourth acquisition unit, 11...learning unit, 12...calculation unit, 13...determination unit, 14...first memory unit, 15...output unit, 16...first acquisition unit, 17...second memory unit, 18...communication control unit, 2...communication terminal, 20...SIM, 4...core network, 40...UPF, 41...AMF, 42...SMF, 43...DNS, 50...cloud, 101...bus, 102...processor, 103...main memory device, 41a, 42a, 43a, 104...communication interface, 105...auxiliary memory device, 106...input / output I / O, BS0 to BSn...base station, L, NW...network.

Claims

1. an output symbol acquisition unit configured to acquire an output symbol sequence including first identification information of a first website accessed via a base station in a communication area in which the communication terminal is located; a storage unit configured to store a trained hidden Markov model in which parameters of the hidden Markov model are estimated in advance, the hidden Markov model being a finite set of hidden states each of which is switched as the communication terminal moves, and a finite set of observable output symbols each of which is an identification information of each website accessed by the communication terminal via each of the base stations; a calculation unit configured to calculate a probability that the communication terminal will access the first website having the first identification information acquired by the output symbol acquisition unit, via each of the base stations, by using the trained hidden Markov model; and a determination unit configured to determine a base station through which the communication terminal has the highest probability of accessing the first website, based on the probability that the communication terminal will access the first website through each of the base stations calculated by the calculation unit; and an output unit configured to output information of the base station having the highest probability as communication control information; A communication management device comprising:

2. 2. The communication management device according to claim 1, the output symbol acquisition unit acquires, as learning data, an output symbol sequence indicating identification information of each of a plurality of websites including the first website accessed by the communication terminal via each of the base stations; a learning unit configured to estimate the parameters of the hidden Markov model including a state transition probability distribution, which is the probability that the connection of the communication terminal switches from a predetermined base station to another base station, and a symbol output probability distribution, which is the probability that the communication terminal accesses a website of predetermined identification information via each of the base stations, maximizing the likelihood for the learning data; The trained hidden Markov model includes the parameters estimated by the training unit. A communication management device comprising:

3. 2. The communication management device according to claim 1, a first acquisition unit configured to acquire, based on the communication control information, information on a cloud at a cloud site that is located at the shortest physical distance from a base station at which the communication terminal has the highest probability of accessing the first website, among clouds at a plurality of cloud sites that provide the first website; a communication control unit configured to instruct a core network of a predetermined communication standard to set the connection destination of the communication terminal to the cloud at the cloud site according to the cloud information of the cloud at the cloud site acquired by the first acquisition unit; A communication management device comprising:

4. 2. The communication management device according to claim 1, The output symbol acquisition unit a second acquisition unit configured to acquire, from a core network of a predetermined communication standard, a presence log indicating a base station of a communication area in which the communication terminal having the subscriber identification information is present and a time when the communication terminal is present; a third acquisition unit configured to acquire, from the core network, a terminal IP address assigned to the subscriber identification information included in the visit log acquired by the second acquisition unit; a fourth acquisition unit configured to acquire, from the core network, an access log including an access time to a website having a site IP address that is the identification information for each of the subscriber identification information, based on the terminal IP address acquired by the third acquisition unit; A communication management device comprising:

5. 5. The communication management device according to claim 4, The second acquisition unit acquires the visit log from an access and mobility management device included in the core network; The third acquisition unit acquires the terminal IP address from a session management function included in the core network. A communication management device comprising:

6. 6. The communication management device according to claim 4, The fourth acquisition unit acquires a site IP address and an access time for each of the communication terminals by referring to an access log for each of the terminal IP addresses to each website having an IP address that is the identification information, the access log being stored in a domain name system included in the core network. A communication management device comprising:

7. an output symbol acquisition step of acquiring an output symbol sequence including first identification information of a first website accessed via a base station in a communication area in which the communication terminal is located; a storage step of storing in a storage unit a trained hidden Markov model in which parameters of the hidden Markov model are estimated in advance, the hidden Markov model being a finite set of hidden states each of which is switched as the communication terminal moves, and a finite set of observable output symbols each of which is an identification information of each website accessed by the communication terminal via each of the base stations; a calculation step of calculating a probability that the communication terminal will access the first website having the first identification information acquired in the output symbol acquisition step, via each of the base stations, by using the trained hidden Markov model; a determining step of determining a base station through which the communication terminal has the highest probability of accessing the first website, based on the probability that the communication terminal will access the first website through each of the base stations obtained in the calculating step; an output step of outputting information of the base station having the highest probability as communication control information; A communication management method comprising:

8. 8. The communication management method according to claim 7, the output symbol acquiring step acquires, as learning data, an output symbol sequence indicating identification information of each of a plurality of websites including the first website accessed by the communication terminal via each of the base stations; Further, a learning step is provided for estimating the parameters of the hidden Markov model, the parameters including a state transition probability distribution, which is the probability that the connection of the communication terminal switches from a predetermined base station to another base station, and a symbol output probability distribution, which is the probability that the communication terminal accesses a website having predetermined identification information via each of the base stations, maximizing the likelihood for the learning data; The trained hidden Markov model includes the parameters estimated in the training step. A communication management method comprising:

9. 2. The communication management device according to claim 1, a first acquisition step of acquiring information on a cloud at a cloud site that is closest in physical distance to a base station at which the communication terminal has the highest probability of accessing the first website, from among clouds at a plurality of cloud sites that provide the first website, based on the communication control information; a communication control step of instructing a core network of a predetermined communication standard to set the connection destination of the communication terminal to the cloud at the cloud site according to the cloud information of the cloud at the cloud site acquired in the first acquisition step; A communication management method comprising:

10. 8. The communication management method according to claim 7, The output symbol acquisition step includes: a second acquisition step of acquiring, from a core network of a predetermined communication standard, a visit log indicating a base station of a communication area in which the communication terminal having the subscriber identification information was present and a time when the communication terminal was present; a third acquisition step of acquiring, from the core network, a terminal IP address assigned to the subscriber identification information included in the visit log acquired in the second acquisition step; a fourth acquisition step of acquiring from the core network an access log including an access time to a website having a site IP address, which is the identification information, for each of the subscriber identification information based on the terminal IP address acquired in the third acquisition step; A communication management method comprising:

11. The method of managing communications according to claim 10, The second acquisition step includes acquiring the visit log from an access and mobility management device included in the core network; The third acquisition step acquires the terminal IP address from a session management function included in the core network. A communication management method comprising:

12. 12. The communication management method according to claim 10, The fourth acquisition step refers to an access log for each of the terminal IP addresses to each website having an IP address that is the identification information, the access log being stored in a domain name system included in the core network, and acquires a site IP address and an access time for each of the communication terminals. A communication management method comprising:

Images