Communication device, information processing device, communication method, and information processing method

Abstract

To provide a communication device, an information processing device, a communication method, and an information processing method that can achieve highly convenient communication path switching.SOLUTION: A communication device includes a prediction unit that, when quality of wireless communication using a first wireless communication method satisfies a predetermined condition, predicts whether a user wants to continue the wireless communication using the first wireless communication method, using a prediction model of the behavior of the user regarding the wireless communication. The prediction model is a learning model generated based on at least one of a behavioral parameter of the user regarding the wireless communication and an environmental parameter regarding the wireless communication related to the behavioral parameter. The prediction unit predicts whether the user wants to continue using the wireless communication using the first wireless communication method, every predetermined time interval after the quality of the wireless communication using the first wireless communication method satisfies the predetermined condition.SELECTED DRAWING: Figure 10

Description

Technical Field

[0001] The present disclosure relates to a communication device, an information processing device, a communication method, and an information processing method.

Background Art

[0002] Communication devices such as smartphones can often be connected to multiple communication paths (for example, Wi-Fi or a cellular network). The communication device appropriately switches the currently connected communication path (for example, a wireless LAN (Local Area Network) network such as Wi-Fi) to another communication path (for example, a cellular network) according to the state of the communication path. For example, a conventional communication device estimates the quality of a communication path from changes in the wireless environment and communication conditions, and switches the communication path based on the result.

Prior Art Documents

Non-Patent Documents

[0003]

Non-Patent Document 1

Summary of the Invention

Problems to be Solved by the Invention

[0004] [[ID=3,4]] However, even if a communication device determines that a switch in the communication path is necessary, some users may want to continue using the communication path they are currently connected to. For example, even if the communication quality of a wireless LAN deteriorates and the communication device determines that it is necessary to switch to a cellular network, some users may want to continue using the wireless LAN due to cost or other reasons. In this case, the user will find the automatic switching of the communication path by the communication device inconvenient.

[0005] Therefore, this disclosure proposes a communication device, an information processing device, a communication method, and an information processing method that enable convenient switching of communication channels.

[0006] It should be noted that the above-mentioned problems or objectives are merely one of several problems or objectives that can be solved or achieved by the multiple embodiments disclosed herein. [Means for solving the problem]

[0007] To solve the above problems, one embodiment of a communication device according to the present disclosure includes a prediction unit that, when the quality of wireless communication using a first wireless communication method satisfies predetermined conditions, predicts whether the user wants to continue using the wireless communication using the first wireless communication method, using a prediction model of the user's behavior regarding the wireless communication, wherein the prediction model is a learning model generated based on at least one of the user's behavioral parameters regarding the wireless communication and environmental parameters related to the wireless communication relating to the behavioral parameters, and the prediction unit makes a prediction whether the user wants to continue using the wireless communication using the first wireless communication method at predetermined time intervals after the quality of the wireless communication using the first wireless communication method satisfies the predetermined conditions. [Brief explanation of the drawing]

[0008] [Figure 1] This is a diagram illustrating the overview of this embodiment. [Figure 2] This is a diagram to explain the predictive model. [Figure 3] This figure shows an example configuration of a communication system according to the embodiment of this disclosure. [Figure 4] This figure shows an example of a server configuration according to the embodiments of this disclosure. [Figure 5] This figure shows an example configuration of a terminal device according to the embodiments of this disclosure. [Figure 6] This is a diagram to explain associative learning. [Figure 7] This figure shows an example of applying associative learning to prediction in this embodiment. [Figure 8] This diagram shows the functional configuration of communication system 1. [Figure 9] This is a flowchart of the learning process. [Figure 10] This is a flowchart showing the bearer switching process. [Figure 11] This is a diagram illustrating an example of applying a predictive model. [Figure 12] This is a diagram illustrating another example of applying a predictive model. [Figure 13] This is a diagram illustrating another example of applying a predictive model. [Modes for carrying out the invention]

[0009] Embodiments of this disclosure will be described in detail below with reference to the drawings. In each of the following embodiments, the same parts will be denoted by the same reference numerals to avoid redundant descriptions.

[0010] Furthermore, in this specification and drawings, multiple components having substantially the same functional configuration may be distinguished by adding different numbers after the same reference numeral. For example, multiple components having substantially the same functional configuration may be distinguished as terminal devices 201 and 202 as needed. However, if there is no need to particularly distinguish each of multiple components having substantially the same functional configuration, only the same reference numeral will be used. For example, if there is no need to particularly distinguish terminal devices 201 and 202, they will simply be referred to as terminal device 20.

[0011] One or more embodiments (including examples and variations) described below can each be implemented independently. On the other hand, at least some of the multiple embodiments described below may be implemented in appropriate combination with at least some of other embodiments. These multiple embodiments may include different novel features. Therefore, these multiple embodiments can contribute to solving different objectives or problems and can exhibit different effects.

[0012] <<1. Overview>> Terminal devices such as smartphones can often be connected to multiple communication paths. Conventional terminal devices appropriately switch communication paths according to the state of the communication paths. For example, conventional terminal devices estimate the quality of communication paths from changes in the wireless environment and communication status, and switch communication paths based on the results.

[0013] However, even if the terminal device determines that it is necessary to switch the communication path, some users may want to continue using the currently connected communication path as it is. For example, even if the communication quality of a wireless LAN (hereinafter also referred to as Wi-Fi) deteriorates and the terminal device determines that it is necessary to switch to a cellular network, some users may want to continue using Wi-Fi, for example, due to cost issues. In this case, the user will feel inconvenienced by the automatic switching of the communication path by the terminal device. There is also an issue that it is difficult for the user to understand why the communication path has been switched.

[0014] Therefore, in this embodiment, the above problems are solved as follows.

[0015] FIG. 1 is a diagram showing an overview of this embodiment. Terminal devices such as smartphones can often be connected to multiple communication paths (hereinafter also referred to as bearers). In the example of FIG. 1, the terminal device can be connected to two bearers, a wireless LAN network such as Wi-Fi (registered trademark) and a cellular network.

[0016] In this embodiment, the terminal device first predicts the communication quality of the currently used bearer. For example, if the currently used bearer is Wi-Fi, the terminal device predicts the wireless quality of Wi-Fi. Then, when the communication quality of the currently used bearer deteriorates (for example, when the wireless quality of Wi-Fi meets a predetermined standard), the terminal device predicts the user's connection intention (whether the user wants to continue using the currently used bearer).

[0017] As a method for predicting communication quality and connection intention, a method using machine learning can be considered. In this embodiment, the terminal device performs these predictions using a prediction model distributed from the server. FIG. 2 is a diagram for explaining the prediction model. In this embodiment, as the prediction model, a quality prediction model for predicting communication quality and an intention prediction model for predicting the user's connection intention are used.

[0018] The quality prediction model is a learning model that learns at least one of the user's behavior parameters related to wireless communication and the environmental parameters related to the wireless communication related to the behavior parameters as input data, and information related to communication quality as the correct label (teacher data). When the terminal device inputs parameters into the quality prediction model, the quality prediction model outputs the future communication quality of the corresponding bearer (for example, 0 (good) to 1 (bad)).

[0019] The intention prediction model is a learning model that learns at least one of the user's behavior parameters related to wireless communication and the environmental parameters related to the wireless communication related to the behavior parameters as input data, and a value indicating whether the user wants to continue using the current bearer as the correct label (teacher data). Note that the terminal device may obtain the value to be used as the correct label by directly asking the user via a dialog, or may obtain it based on the user's behavior towards the bearer (for example, an action such as manually turning off Wi-Fi). When the terminal device inputs parameters into the intention prediction model, the intention prediction model outputs the degree of desire to use the corresponding bearer of the user (for example, 0 (want to use) to 1 (do not want to use)).

[0020] The terminal device predicts the user's connection intention at regular intervals, after the quality of the currently used bearer (e.g., Wi-Fi) meets predetermined conditions. If the terminal device determines that the user does not want to continue using the current bearer, it switches the currently used bearer (e.g., Wi-Fi) to another bearer (e.g., a cellular network).

[0021] Thus, in this embodiment, when the communication quality of the current bearer deteriorates, the terminal device does not simply switch the current bearer to another bearer, but rather predicts the user's connection intention (e.g., their preference for using the current bearer) before switching. The terminal device then switches the current bearer to another bearer if it predicts that the user does not want to continue using the current bearer (i.e., is willing to switch to another bearer). This reduces unintended switching by the user, thereby improving the user experience (UX).

[0022] Having outlined the basics of this embodiment, the communication system 1 according to this embodiment will now be described in detail.

[0023] <<2. Communication System Configuration>> First, let's explain the configuration of communication system 1.

[0024] Figure 3 shows an example configuration of a communication system 1 according to an embodiment of the present disclosure. The communication system 1 comprises a server 10 and terminal devices 20 configured to connect to a plurality of communication paths (hereinafter also referred to as bearers). The communication system 1 may comprise a plurality of servers 10 and terminal devices 20. In the example of Figure 3, the communication system 1 comprises servers 101, 102, etc. as servers 10, and terminal devices 201, 202, 203, etc. as terminal devices 20. The terminal devices 20 connect to the servers 10 via an upstream network.

[0025] Here, the upstream network refers to the network above the terminal device 20. In the example in Figure 3, the upstream networks are networks N1 and N2. Networks N1 and N2 are communication networks such as LAN (Local Area Network), WAN (Wide Area Network), cellular network, fixed telephone network, regional IP (Internet Protocol) network, and the Internet. Networks N1 and N2 may include wired networks or wireless networks. Networks N1 and N2 may also include core networks. Core networks are, for example, EPC (Evolved Packet Core) or 5GC (5G Core network). Of course, network N may also be a data network connected to the core network. The data network may be a service network of a telecommunications carrier, for example, an IMS (IP Multimedia Subsystem) network. Alternatively, the data network may be a private network such as an internal corporate network.

[0026] Note that while only two upstream networks are shown in the example in Figure 3, the number of upstream networks is not limited to two. For example, if the upstream network is a cellular network consisting of a wireless access network and a core network, then multiple core networks may exist as upstream networks. In this case, each core network may have a different data network.

[0027] The terminal device 20 of this embodiment can connect to the upstream network using multiple communication paths. In this case, at least one of the multiple communication paths may be a wireless communication path. For example, the communication path may be a wireless communication path (wireless access network) between the terminal device 20 and a base station. Alternatively, the communication path may be a wireless communication path between the terminal device 20 and an access point. Of course, the multiple communication paths may include wired communication paths (e.g., wired LAN). Furthermore, the communication path may be the upstream network itself.

[0028] When multiple communication paths include wireless communication paths, the terminal device 20 may be configured to connect to the upstream network using radio access technologies (RATs) such as LTE (Long Term Evolution), NR (New Radio), Wi-Fi, and Bluetooth (registered trademark). In this case, the terminal device 20 may be configured to use different radio access technologies (wireless communication methods). For example, the terminal device 20 may be configured to use NR and Wi-Fi. Also, the terminal device 20 may be configured to use different cellular communication technologies (e.g., LTE and NR). LTE and NR are types of cellular communication technologies that enable mobile communication of the terminal device by arranging multiple cell-like areas covered by base stations.

[0029] In the following explanation, "LTE" includes LTE-A (LTE-Advanced), LTE-A Pro (LTE-Advanced Pro), and EUTRA (Evolved Universal Terrestrial Radio Access). Similarly, "NR" includes NRAT (New Radio Access Technology) and FEUTRA (Further EUTRA). A single base station may manage multiple cells. In the following explanation, cells supporting LTE are referred to as LTE cells, and cells supporting NR are referred to as NR cells.

[0030] NR is the next generation (fifth generation) of wireless access technology following LTE (fourth generation communication including LTE-Advanced and LTE-Advanced Pro). NR is a wireless access technology that can support a variety of use cases, including eMBB (Enhanced Mobile Broadband), mMTC (Massive Machine Type Communications), and URLLC (Ultra-Reliable and Low Latency Communications). NR is being developed with the aim of creating a technical framework that addresses the usage scenarios, requirements, and deployment scenarios in these use cases.

[0031] Furthermore, the terminal device 20 may be able to connect to the upstream network using wireless access technologies (wireless communication methods) other than LTE, NR, Wi-Fi, and Bluetooth. For example, the terminal device 20 may be able to connect to the upstream network using LPWA (Low Power Wide Area) communication. Also, the terminal device 20 may be able to connect to the upstream network using a proprietary wireless communication standard.

[0032] Here, LPWA communication refers to wireless communication that enables low-power, wide-area communication. For example, LPWA wireless refers to IoT (Internet of Things) wireless communication using specific low-power radio (e.g., the 920MHz band) or ISM (Industry-Science-Medical) bands. The LPWA communication used by the terminal device 20 may also conform to the LPWA standard. Examples of LPWA standards include ELTRES, ZETA, SIGFOX, LoRaWAN, and NB-IoT. Of course, the LPWA standard is not limited to these, and other LPWA standards may also be used.

[0033] Furthermore, multiple communication paths may include virtual networks. For example, multiple communication paths to which terminal device 20 can connect may include virtual networks such as VLANs (Virtual Local Area Networks) and physical networks such as IP communication paths. In this case, terminal device 20 may perform routing based on routing protocols such as OSPF (Open Shortest Path First) and BGP (Border Gateway Protocol).

[0034] In addition, multiple communication channels may include one or more overlay networks, or one or more network slicing.

[0035] It should be noted that the devices in the diagram can be considered as devices in a logical sense. In other words, some or all of the devices in the diagram may be implemented as virtual machines (VMs), containers, Docker, etc., and these may be implemented on the same physical hardware.

[0036] Furthermore, in this embodiment, a communication device is a device that has a communication function, and in the example of Figure 3, at least the terminal device 20 is a communication device. The server 10 may also be considered a communication device.

[0037] The configuration of each device constituting communication system 1 will be described in detail below. Note that the configurations of each device shown below are merely examples. The configuration of each device may differ from those shown below.

[0038] <2-1. Server Configuration> First, let's describe the configuration of server 10.

[0039] Server 10 is an information processing device (computer) that provides various services to terminal devices 20 via the upstream network (e.g., networks N1 and N2). For example, Server 10 is an application server or a web server. Server 10 may be a PC server, a midrange server, or a mainframe server. Server 10 may also be an information processing device that performs data processing (edge ​​processing) near users or terminals. For example, Server 10 may be an information processing device (computer) installed in or attached to a base station. Of course, Server 10 may also be an information processing device that performs cloud computing.

[0040] Figure 4 shows an example configuration of a server 10 according to the present disclosure. The server 10 comprises a storage unit 11, a communication unit 12, and a control unit 13. Note that the configuration shown in Figure 4 is a functional configuration, and the hardware configuration may differ. Furthermore, the functions of the server 10 may be distributed and implemented across multiple physically separated configurations. For example, the server 10 may be composed of multiple information processing devices.

[0041] The memory unit 11 is a data read / write storage device such as DRAM (Dynamic Random Access Memory), SRAM (Static Random Access Memory), flash memory, or hard disk. The memory unit 11 functions as a storage means for the server 10. The memory unit 11 stores prediction models (learning models) to be distributed to the terminal device 20, such as quality prediction models and intention prediction models. This information will be described later.

[0042] The communication unit 12 is a communication interface for communicating with other devices. For example, the communication unit 12 is a network interface. For example, the communication unit 12 is a LAN (Local Area Network) interface such as a NIC (Network Interface Card). The communication unit 12 may be a wired interface or a wireless interface. The communication unit 12 functions as a communication means for the server 10. The communication unit 12 communicates with terminal devices 20 and 30 according to the control of the control unit 13.

[0043] The control unit 13 is a controller that controls various parts of the server 10. The control unit 13 is implemented by a processor such as a CPU (Central Processing Unit) or an MPU (Micro Processing Unit). For example, the control unit 13 is implemented by the processor executing various programs stored in the internal storage device of the server 10 using RAM (Random Access Memory) or the like as a working area. The control unit 13 may also be implemented by an integrated circuit such as an ASIC (Application Specific Integrated Circuit) or an FPGA (Field Programmable Gate Array). CPUs, MPUs, ASICs, and FPGAs can all be considered controllers.

[0044] The control unit 13 comprises an acquisition unit 131 and a generation unit 132. Each block constituting the control unit 13 (acquisition unit 131 to generation unit 132) is a functional block that represents the function of the control unit 13. These functional blocks may be software blocks or hardware blocks. For example, each of the above-mentioned functional blocks may be a single software module implemented in software (including a microprogram), or a single circuit block on a semiconductor chip (die). Of course, each functional block may also be a single processor or a single integrated circuit. The control unit 13 may be composed of functional units different from the above-mentioned functional blocks. The configuration of the functional blocks is arbitrary.

[0045] <2-2. Terminal Device Configuration> Next, the configuration of the terminal device 20 will be described.

[0046] Terminal device 20 is a communication device that communicates with other communication devices such as base stations and access points. Terminal device 20 is configured to be connectable to multiple communication paths (bearers). For example, terminal device 20 can be connected to two bearers: Wi-Fi (registered trademark) and a cellular network.

[0047] The terminal device 20 can be any form of computer. For example, the terminal device 20 may be a mobile device such as a mobile phone, smart device (smartphone or tablet), PDA (Personal Digital Assistant), or notebook PC. The terminal device 20 may also be a wearable device such as a smartwatch. The terminal device 20 may also be an xR device such as an AR (Augmented Reality) device, VR (Virtual Reality) device, or MR (Mixed Reality) device. The terminal device 20 may also be an imaging device equipped with communication functions (e.g., a camcorder), or a motorcycle or mobile relay vehicle equipped with communication equipment such as an FPU (Field Pickup Unit). The terminal device 20 may also be an M2M (Machine to Machine) device or an IoT (Internet of Things) device. The terminal device 20 may also be a router with multiple communication channels.

[0048] Furthermore, the terminal device 20 may be capable of LPWA communication with other communication devices (for example, base stations, access points, and other terminal devices 20). Also, the wireless communication used by the terminal device 20 may be wireless communication using millimeter waves. In addition, the wireless communication used by the terminal device 20 may be wireless communication using radio waves, or wireless communication using infrared or visible light (optical wireless).

[0049] Furthermore, the terminal device 20 may also be a mobile device. The mobile device is a portable wireless communication device. In this case, the terminal device 20 may be a wireless communication device installed on the mobile device, or it may be the mobile device itself. For example, the terminal device 20 may be a vehicle that moves on roads, such as an automobile, bus, truck, or motorcycle, or a wireless communication device mounted on such a vehicle. The mobile device may be a mobile terminal, or a mobile device that moves on land, underground, on water, or underwater. The mobile device may also be a mobile device that moves within the atmosphere, such as a drone or helicopter, or a mobile device that moves outside the atmosphere, such as an artificial satellite.

[0050] The terminal device 20 may simultaneously connect to and communicate with multiple base stations or multiple cells. For example, if one base station supports a communication area via multiple cells (e.g., pCell, sCell), it is possible to combine those multiple cells and communicate between the base station and the terminal device 20 using carrier aggregation (CA), dual connectivity (DC), or multi-connectivity (MC) technologies. Alternatively, the terminal device 20 can communicate with those multiple base stations via cells from different base stations using coordinated multi-point transmission and reception (CoMP) technology.

[0051] Figure 5 shows an example of the configuration of a terminal device 20 according to the present disclosure. The terminal device 20 includes a storage unit 21, a communication unit 22, a control unit 23, a sensor unit 24, and a plurality of communication units 25 (communication unit 25 1~ twenty five n The system comprises (where n is any integer) and . Note that the configuration shown in Figure 5 is a functional configuration, and the hardware configuration may differ from this. Furthermore, the functions of the terminal device 20 may be implemented in a distributed manner across multiple physically separated configurations.

[0052] The memory unit 21 is a data read / write storage device such as DRAM, SRAM, flash memory, or hard disk. The memory unit 21 functions as a storage means for the terminal device 20. The memory unit 11 stores, for example, prediction models (learning models) such as quality prediction models and intention prediction models. This information will be described later.

[0053] The communication unit 22 is a communication interface for communicating with other devices on the network (e.g., server 10). For example, the communication unit 22 is a network interface. For example, the communication unit 22 is a LAN interface such as a NIC. The communication unit 22 may be a wired interface or a wireless interface. The communication unit 22 functions as a communication means for the terminal device 20. The communication unit 22 communicates with other communication devices according to the control of the control unit 23. The communication unit 22 may have the same configuration as the communication unit 25.

[0054] The control unit 23 is a controller that controls various parts of the terminal device 20. The control unit 23 is implemented by a processor such as a CPU or MPU. For example, the control unit 23 is implemented by the processor executing various programs stored in the memory device inside the terminal device 20 using RAM or the like as a working area. The control unit 23 may also be implemented by an integrated circuit such as an ASIC or FPGA. CPUs, MPUs, ASICs, and FPGAs can all be considered controllers. In addition, the control unit 23 may be implemented by a GPU in addition to, or instead of, the CPU.

[0055] The control unit 23 comprises an acquisition unit 231, a learning unit 232, a prediction unit 233, and a communication control unit 234. Each block constituting the control unit 23 (acquisition unit 231 to communication control unit 234) is a functional block that represents the function of the control unit 23. These functional blocks may be software blocks or hardware blocks. For example, each of the above-mentioned functional blocks may be a single software module implemented in software (including microprograms), or a single circuit block on a semiconductor chip (die). Of course, each functional block may also be a single processor or a single integrated circuit. The control unit 23 may be composed of functional units different from the above-mentioned functional blocks. The configuration of the functional blocks is arbitrary.

[0056] The sensor unit 24 is a sensor that acquires various information for predicting the quality of the communication path (the communication path formed by the communication unit 25) for connecting to the upstream network. For example, if the communication path is a wireless communication path, the sensor unit 24 is a sensor that detects the received signal-to-noise ratio (S / N) of radio waves received from a base station or access point. Of course, the information acquired by the sensor unit 24 is not limited to the received S / N, as long as it can be used to predict the quality of the communication path.

[0057] Each of the multiple communication units 25 is a communication interface for connecting to the upstream network. Each of the multiple communication units 25 forms a different communication path to the server 10. Each of the multiple communication units 25 may support a different wireless access technology (wireless communication method). For example, communication unit 251 may support LTE or NR, and communication unit 252 may support Wi-Fi. In addition, communication units 25 may support other wireless access technologies such as Bluetooth and LPWA. The communication unit 25 may also have the same configuration as communication unit 22.

[0058] <2-3. Predictive Models (Learning Models)> Next, the prediction model (learning model) used in this embodiment will be described.

[0059] <2-3-1. About the Learning Model> As described above, the storage unit 11 of the server 10 and the storage unit 21 of the terminal device 20 store prediction models such as a quality prediction model and an intention prediction model. The quality prediction model is a learning model for predicting the future quality of the bearer currently being used by the terminal device 20. The intention prediction model is a learning model for predicting the user's connection intention regarding the bearer.

[0060] A learning model is a machine learning model, such as a neural network model. A neural network model consists of layers called an input layer, hidden layers (or hidden layers), and an output layer, each containing multiple nodes, with each node connected via an edge. Each layer has a function called an activation function, and each edge is weighted. A learning model has one or more hidden layers (or hidden layers). When the learning model is a neural network model, learning the learning model means setting, for example, the number of hidden layers (or hidden layers), the number of nodes in each layer, or the weights of each edge.

[0061] Here, the neural network model may be a deep learning model. In this case, the neural network model may be a model of the form known as DNN (Deep Neural Network). Alternatively, the neural network model may be a model of the form known as CNN (Convolutional Neural Network), RNN (Recurrent Neural Network), or LSTM (Long Short-Term Memory). Of course, the neural network model is not limited to these forms.

[0062] Furthermore, the learning model is not limited to a neural network model. For example, the learning model may be a reinforcement learning model. In reinforcement learning, actions (settings) that maximize value are learned through trial and error. In addition, the learning model may be a logistic regression model.

[0063] Furthermore, the learning model may consist of multiple models. For example, the learning model may consist of multiple neural network models. More specifically, the learning model may consist of multiple neural network models selected from, for example, CNN, RNN, and LSTM. When the learning model consists of multiple neural network models, these multiple neural network models may be in a dependent relationship or a parallel relationship.

[0064] As described above, the storage unit 11 of the server 10 and the storage unit 21 of the terminal device 20 store learning models (predictive models) such as quality prediction models and intention prediction models. The learning models are used in the bearer switching process described later. The learning models will be explained in detail below.

[0065] The quality prediction model is a trained model (trained model) that has learned using, for example, at least one of the user's behavioral parameters related to wireless communication and environmental parameters related to wireless communication related to said behavioral parameters as input data, and information about communication quality as correct labels (training data). When the terminal device 20 inputs parameters to the quality prediction model, the quality prediction model outputs, for example, the future communication quality of the bearer in question (e.g., 0 (good) to 1 (poor)).

[0066] The intent prediction model is a trained model (trained model) that has learned using, for example, at least one of the user's behavioral parameters related to wireless communication and environmental parameters related to wireless communication related to said behavioral parameters as input data, and a value indicating whether the user wants to continue using the current bearer as the ground truth label (training data). When the terminal device 20 inputs parameters to the intent prediction model, the intent prediction model outputs, for example, the user's desire to use the bearer in question (e.g., 0 (want to use) to 1 (do not want to use)). (In the following explanation, the information indicating whether the user wants to continue using the current bearer may be referred to as connection intent information.)

[0067] In Figures 4 and 5, textual information such as "quality prediction model" and "intent prediction model" is shown as the information stored in memory units 11 and 21. However, in reality, memory units 11 and 21 store strings of characters and numerical values ​​that indicate the structure of the model and connection coefficients.

[0068] A quality prediction model may be a model that has been trained to output information about the future communication quality of a bearer when parameters are input, using a set of data consisting of parameters (behavioral parameters, and / or behavioral parameters) and information about communication quality as training data. In this case, the first training model may include an input layer for inputting parameters, an output layer for outputting information about communication quality, a first element belonging to any layer between the input layer and the output layer other than the output layer, and a second element whose value is calculated based on the first element and the weight of the first element. The model may also be a model that causes a computer to function so that, based on the information input to the input layer, it outputs information about communication quality from the output layer according to the parameters input to the input layer by performing calculations based on the first element and the weight of the first element (i.e., connection coefficient), with each element belonging to each layer other than the output layer as the first element.

[0069] Furthermore, the intention prediction model may be a model that has been trained to output the corresponding connection intention information when parameters are input, using data in the form of a set of parameters (behavioral parameters, and / or behavioral parameters) and connection intention information as training data. In this case, the first learning model may include an input layer for inputting parameters, an output layer for outputting connection intention information, a first element belonging to any layer between the input and output layers other than the output layer, and a second element whose value is calculated based on the first element and the weight of the first element. The model may also be a model that causes the computer to function so that, based on the information input to the input layer, it outputs connection intention information from the output layer according to the parameters input to the input layer by performing calculations based on the first element and the weight of the first element (i.e., connection coefficient), with each element belonging to each layer other than the output layer as the first element.

[0070] Here, we assume that the learning model is implemented as a neural network having one or more hidden layers, such as a DNN. In this case, the first element included in the learning model corresponds to one of the nodes in the input layer or hidden layer. The second element corresponds to the next node, which is the node to which values ​​are transmitted from the node corresponding to the first element. Furthermore, the weights of the first element correspond to connection coefficients, which are weights considered for the values ​​transmitted from the node corresponding to the first element to the node corresponding to the second element.

[0071] Furthermore, let's assume that the learning model is implemented as a regression model represented by "y = a1*x1 + a2*x2 + ... + ai*xi". In this case, the first element included in the learning model corresponds to the input data (xi) such as x1 and x2. Also, the weights of the first element correspond to the coefficient ai that corresponds to xi. Here, the regression model can be considered as a simple perceptron having an input layer and an output layer. When each model is considered as a simple perceptron, the first element can be considered as one of the nodes in the input layer, and the second element can be considered as a node in the output layer.

[0072] The terminal device 20 calculates the information to be output using a model with an arbitrary structure, such as a neural network or a regression model. Specifically, the first learning model is configured to output communication quality information or connection intention information when at least one of the behavioral parameters and environmental parameters is input. For example, the server 10 sets the coefficients based on the similarity between the measured communication quality data and the value obtained by inputting at least one of the behavioral parameters and environmental parameters into the learning model. The server 10 uses such a learning model to generate communication quality information or connection intention information from at least one of the behavioral parameters and environmental parameters.

[0073] In the example above, a model was shown that outputs communication quality information or connection intent information when parameters are input, as an example of a learning model. However, the learning model according to the embodiment may also be a model generated based on the results obtained by repeatedly inputting and outputting data to the learning model.

[0074] Furthermore, if server 10 performs learning or output information generation using GAN (Generative Adversarial Network), the learning model may be a model that constitutes a part of the GAN.

[0075] The learning device that trains the learning model (for example, the learning model) may be the server 10, the terminal device 20, or another information processing device. For example, suppose the server 10 trains the learning model. In this case, the server 10 trains the learning model and stores the trained learning model in the storage unit 11. More specifically, the server 10 sets the connection coefficient of the learning model so that when at least one of the behavioral parameters and environmental parameters is input to the learning model, the learning model outputs communication quality information or connection intention information.

[0076] For example, the server 10 or terminal device 20 inputs parameters to the input layer nodes of the learning model and propagates the data through each intermediate layer to the output layer of the learning model, thereby outputting communication quality information or connection intent information. The server 10 or terminal device 20 then modifies the connection coefficient of the learning model based on the difference between the communication quality information or connection intent information actually output by the learning model and the communication quality information or connection intent information used as the correct label (training data). At this time, the server 10 or terminal device 20 may use a method such as backpropagation to modify the connection coefficient. At this time, the server 10 or terminal device 20 may modify the connection coefficient based on the cosine similarity between the vector representing the first measured data and the vector representing the value actually output by the learning model.

[0077] Any learning algorithm may be used for learning. For example, the server 10 or terminal device 20 may use learning algorithms such as neural networks, support vector machines, clustering, reinforcement learning, random forests, or decision trees to train the learning model.

[0078] Furthermore, the learning algorithm used in this embodiment may be one in which the server 10 and the terminal device 20 learn independently, or it may be one in which the server 10 and the terminal device 20 learn in cooperation. Here, as an example of a learning algorithm in which the server 10 and the terminal device 20 learn in cooperation, federated learning can be mentioned. Federated learning will be described below.

[0079] <2-3-2. About Associative Learning> Federated learning is a type of algorithm used to optimize machine learning models. Federated learning allows for the training of a model without exposing the private data of individual devices to external sources.

[0080] Figure 6 is a diagram illustrating federative learning. In federative learning, a group of devices holding data and a server that manages them work together to advance the learning process. First, each device downloads a model from a server in the cloud, trains on that model using its own data, and updates it. Then, these devices upload the updated model to the server. The server aggregates the large number of updated models to generate a single new model. The server then distributes the new model to the devices. Each device trains using the new model. Each device and the server repeat these processes. In federative learning, since each device updates its model and uploads the updated model to the cloud server, it becomes possible to perform training without exposing the private data on each device.

[0081] In this embodiment, federative learning is used to predict the future communication quality of the currently connected bearer and to predict the user's connection intention. Using user activity logs and access point-identifying information such as BSSIDs is effective for these predictions. However, this information can reveal user privacy information. For example, user activity logs can reveal the user's hobbies and preferences, and BSSIDs can reveal the user's current location. Therefore, in this embodiment, by applying federative learning to these predictions, it is possible to achieve highly accurate predictions while protecting user privacy.

[0082] Figure 7 shows an example of applying federated learning to predictions (prediction of communication quality and prediction of connection intent) in this embodiment.

[0083] First, let's explain the generation of the quality prediction model. Each of the multiple terminal devices 20 stores at least one of the following parameters as features: user behavior parameters related to wireless communication and environmental parameters related to wireless communication related to those behavior parameters, and information about communication quality as correct labels (training data) in the storage unit 21 (Step S1). Then, each of the multiple terminal devices 20 performs model training using the data in the terminal and uploads the updated model to the server 10 (Step S2). The server 10 aggregates the uploaded models and generates a new model. Then, the server 10 distributes the new model to each of the terminal devices 20 (Step S3).

[0084] Next, the generation of the intention prediction model will be explained. Each of the multiple terminal devices 20 stores at least one of the following parameters in the memory unit 21: the user's behavioral parameters related to wireless communication and the environmental parameters related to wireless communication related to those behavioral parameters, as features, and a value indicating whether the user wants to continue using the current bearer as the correct label (training data) (Step S1). Subsequently, each of the multiple terminal devices 20 performs model training using the data in the terminal and uploads the updated model to the server 10 (Step S2). The server 10 aggregates the uploaded models and generates a new model. Then, the server 10 distributes the new model to each of the terminal devices 20 (Step S3).

[0085] The server 10 and terminal device 20 can generate a predictive model without collecting data by repeating these processes (steps S1 to S3).

[0086] <2-4. Functional Configuration of the Communication System> Next, the functional configuration of communication system 1 will be described.

[0087] Figure 8 shows the functional configuration of communication system 1. More specifically, Figure 8 shows the functional configuration required when server 10 and terminal device 20 generate a prediction model through federated learning and perform bearer switching processing based on the generated prediction model.

[0088] First, the acquisition unit 231 of the terminal device 20 acquires the user's behavioral parameters related to wireless communication and environmental parameters related to wireless communication related to those behavioral parameters. The behavioral parameters may be sensor information acquired by the sensor unit 24. The sensor information may be information provided from outside the terminal device 20. The environmental parameters may be information about the bearer currently in use, acquired by the communication control unit 234. The behavioral parameters and environmental parameters will be described later.

[0089] The acquisition unit 231 of the terminal device 20 formats the acquired parameters into an appropriate form and stores the formatted parameters as training data in the storage unit 21. At this time, the acquisition unit 231 assigns correct labels regarding the user's connection intention to the training data.

[0090] Next, the learning unit 232 of the terminal device 20 performs training of the prediction models (quality prediction model and intention prediction model) based on the training data stored in the memory unit 21. At this time, the learning unit 232 performs training at appropriate times. For example, the learning unit 232 performs training when the user is not using the terminal device 20, when the terminal device 20 is charging, and / or when the battery has sufficient charge.

[0091] Next, the communication unit 22 of the terminal device 20 uploads the prediction model generated by learning to the server 10. At this time, the communication unit 22 uploads the prediction model to the server 10 at an appropriate time. For example, the communication unit 22 uploads the prediction model to the server 10 when it connects to free Wi-Fi, and / or at a predetermined time once a day, or at regular times such as the end of the month.

[0092] Server 10 generates a new prediction model by aggregating prediction models uploaded from multiple terminal devices 20. Server 10 then distributes the new prediction model to the terminal devices 20. The communication unit 22 of the terminal device 20 stores the new prediction model in the storage unit 21.

[0093] The prediction unit 233 of the terminal device 20 acquires prediction models (quality prediction model and intention prediction model) from the storage unit 21. The prediction unit 233 then acquires prediction results by inputting the parameters acquired by the acquisition unit 231 into the prediction models. More specifically, the prediction unit 233 acquires prediction results for the future communication quality of the bearer currently being used by the user by inputting behavioral parameters and / or environmental parameters into the quality prediction model. The prediction unit 233 also acquires prediction results for the user's connection intention regarding the bearer currently being used by inputting behavioral parameters and / or environmental parameters into the intention prediction model. The prediction unit 233 predicts the user's connection intention at predetermined time intervals after the communication quality of the bearer currently being used meets predetermined conditions (for example, after the predicted value of communication quality falls below a predetermined threshold).

[0094] The communication control unit 234 of the terminal device 20 performs bearer switching processing based on the prediction results. For example, if the predicted value of the communication quality of the currently used bearer falls below a predetermined threshold, and the prediction result of the connection intention predicts that the user wishes to switch bearers, the currently used bearer (e.g., Wi-Fi) is switched to another bearer (e.g., cellular network).

[0095] <<3. Operation of the Communication System>> The configuration of communication system 1 has been described above. Next, the operation of communication system 1 having this configuration will be described. The operation of communication system 1 can be divided into the learning process of the prediction model and the bearer switching process using the prediction model.

[0096] As described above, predictive models (quality predictive models and intention predictive models) are learning models generated based on at least one of the behavioral parameters and environmental parameters. Predictive models (quality predictive models and intention predictive models) may also be learning models generated based on both the behavioral parameters and environmental parameters.

[0097] In the following explanation, the prediction model is assumed to be an intent prediction model, but the prediction model may also be a quality prediction model. In this case, the information used as the correct label (training data) should be replaced with information about communication quality as appropriate.

[0098] <3-1. Learning Process> First, let's explain the training process for the prediction model. Figure 9 is a flowchart of the training process. The following processes are executed, for example, by the control unit 23 of the terminal device 20. The training process will be explained below with reference to the flowchart in Figure 9.

[0099] First, the terminal device 20 starts connecting to a predetermined bearer (step S101). Then, the terminal device 20 starts acquiring logs related to wireless communication (step S102). For example, the terminal device 20 starts acquiring user behavior parameters related to wireless communication and environmental parameters related to wireless communication related to those behavior parameters.

[0100] Here, the following (A1) to (A3) can be considered as examples of behavioral parameters.

[0101] (A1) Walking condition (A2) Acceleration applied to terminal device 20 (A3) A parameter indicating whether or not reconnection will be performed within a fixed time for the immediate neighborhood.

[0102] Here, (A1) is a parameter that indicates, for example, the user's walking state, for example, whether the user is currently walking or stationary. Also, (A2) is the acceleration value detected by the accelerometer installed in the terminal device 20. (A3) is a parameter that indicates whether the user reconnected to the same bearer within n seconds after disconnecting from the predetermined bearer.

[0103] Furthermore, the following (B1) to (B3) can be considered as examples of environmental parameters.

[0104] (B1) Predicted value of communication quality (B2) Radio field strength (B3) Number of packets remaining (B4) Communication speed (B5) Packet Counter (B6) Connection frequency band (B7) Communication status

[0105] Here, (B1) is a value indicating the predicted deterioration of the communication quality of the currently connected bearer. (B2) is a parameter indicating how strong the signal strength the terminal device 20 can receive is, for example, RSSI (Received Signal Strength Indicator). (B3) is a parameter indicating the number of packets currently lingering in the terminal device 20. (B4) is the theoretical value of the communication speed. (B5) is the value obtained by counting the received and / or transmitted packets. (B6) is a parameter indicating the frequency band to which the connection is currently made; if the currently connected bearer is Wi-Fi, this value indicates whether the currently connected frequency band is 2.4GHz / 5GHz. (B7) is a parameter indicating the communication status of the currently connected bearer.

[0106] The parameters acquired by the terminal device 20 are not limited to the examples above. For example, behavioral parameters may include parameters indicating the user's monthly cumulative data usage, as shown in (C1) to (C2) below, and parameters related to the application the user is currently using. Environmental parameters may include parameters related to the terminal device 20's battery consumption, the importance of the current traffic, and information about the terminal device 20's current location, as shown in (C3) to (C5) below. Examples of parameters acquired by the terminal device 20 are as follows.

[0107] (C1) A parameter indicating the user's monthly cumulative data usage. (C2) The application currently being used by the user, and the average throughput of the communication used by that application. (C3) Parameters related to the battery consumption of terminal device 20 (C4) Current traffic importance (C5) Information regarding the current location of terminal device 20 (C6) Context, Activity Recognition results (C7) The amount of remaining data available to the user, and the pricing plan. (C8) Disconnection status of the surrounding area (C9) User preferences (for example, whether the user wants to use Wi-Fi) (C10) Communication status of a bearer other than the bearer currently in use

[0108] Next, the terminal device 20 determines whether the connection to the bearer connected in step S101 has been disconnected (step S103). If it has not been disconnected (step S103: No), it repeats step S103 until the connection is disconnected. On the other hand, if it has been disconnected (step S203: Yes), it determines, based on information about the user's behavior, whether the user wishes to continue communication using the current bearer (the bearer connected in step S101) (step S104).

[0109] The following (D1) to (D8) are expected behaviors if the user does not wish to continue communication using the current bearer (behavior that the user would have preferred to switch to).

[0110] (D1) Action in which the user turns off the connection to the current bearer. (D2) User's response to a notification displayed on terminal device 20 (for example, the user's response to a notification asking whether they want to switch bearers or not, indicating that they want to switch). (D3) User's communication usage status for the current month (for example, whether data usage is insufficient at the end of the month) (D4) Actions in which a user connects to a different access point (for example, an action in which a user connects to a different SSID (Service Set Identifier)) (D5) The user turns off the screen of terminal device 20. (D6) The user throws the terminal device 20. (D7) The action of closing the application the user is currently using. (D8) The user restarts the terminal device 20.

[0111] Furthermore, the following behaviors (E1) to (E4) are expected if the user wishes to continue using the current bearer for communication (behavior they did not want to occur).

[0112] (E1) An action in which the user reconnects the terminal device 20 to the same access point (for example, an action in which the user reconnects to the same SSID) (E2) User reconnection action to the current bearer (e.g., Wi-Fi) (e.g., turning Wi-Fi ON -> OFF -> ON) (E3) The user puts terminal device 20 into airplane mode and then immediately switches it back. (E4) The user wanders around looking for the direction with the strongest signal.

[0113] If it is determined that the user wishes to continue communication using the current bearer (Step S104: Yes), the terminal device 20 labels some or all of the acquired logs as a sample of when the user wanted to continue the connection (Step S105). On the other hand, if it is determined that the user does not wish to continue communication using the current bearer (Step S104: No), the terminal device 20 labels some or all of the acquired logs as a sample of when the user did not want to continue the connection (Step S106).

[0114] Here, we will explain labeling with a specific example. For example, suppose terminal device 20 determines that "the user wants to continue using Wi-Fi" if "the user has connected to the same Wi-Fi (same access point) and used it for several tens of seconds or more after disconnecting from the Wi-Fi (designated access point)." In this case, terminal device 20 performs labeling as follows.

[0115] (Pre-processing) First, terminal device 20 performs the following preprocessing on the acquired logs. First, terminal device 20 saves parameters for each Wi-Fi session within terminal device 20. Then, terminal device 20 determines whether or not the connections are to the same access point based on the similarity of the Wi-Fi parameters. Then, terminal device 20 groups the connections to the same access point. Finally, for connections to the same access point, terminal device 20 calculates the time between the previous Wi-Fi disconnection and the next Wi-Fi connection.

[0116] (Labeling) Terminal device 20 searches the pre-processed logs for samples that satisfy the following rule: "The user disconnected from Wi-Fi, reconnected within n seconds, and continued to use Wi-Fi for m seconds or more after reconnecting." Terminal device 20 labels samples that satisfy this rule as samples where the user wanted to continue connecting to Wi-Fi.

[0117] (Other examples of labeling) Note that labeling is not limited to this example. For example, terminal device 20 may label a sample where, after disconnecting communication using a predetermined bearer (e.g., Wi-Fi or cellular), the user reconnects to communication using the same bearer (regardless of whether it is the same access point), and continues to use the communication for a predetermined time or longer after reconnection, as a sample of when the user wanted to continue communication using the predetermined bearer.

[0118] Returning to the flow in Figure 9, the terminal device 20 trains a predictive model based on the acquired parameters (for example, at least one of behavioral parameters and environmental parameters) (step S107). Then, the terminal device 20 uploads the trained predictive model to the server 10 (step S108). Once the upload is complete, the terminal device 20 terminates the training process.

[0119] Server 10 retrieves prediction models from multiple terminal devices 20. It then aggregates the uploaded prediction models to generate a new prediction model. Server 10 then distributes the new prediction model to each terminal device 20. The terminal devices 20 save the new prediction model to the storage unit 21.

[0120] <3-2. Bearer Switching Process> Next, we will explain the bearer switching process. Figure 10 is a flowchart of the bearer switching process. The following process is executed, for example, by the control unit 23 of the terminal device 20. The bearer switching process will be explained below with reference to the flowchart in Figure 10.

[0121] First, the terminal device 20 starts connecting to a predetermined bearer (step S201). Then, the terminal device 20 starts acquiring behavioral parameters and environmental parameters (step S202). The behavioral parameters and environmental parameters are the same as those described in <3-1. Learning Process> above.

[0122] Next, the terminal device 20 obtains information on the prediction models (quality prediction model, intention prediction model) from the storage unit 21. Then, the terminal device 20 inputs the parameters (behavioral parameters and / or environmental parameters) obtained in step S202 into the prediction models to predict the future communication quality of the currently connected bearer and the user's connection intention regarding the bearer (step S203). More specifically, the terminal device 20 obtains the prediction result for the future communication quality of the currently used bearer by inputting the behavioral parameters and / or environmental parameters into the quality prediction model. The terminal device 20 also obtains the prediction result for the user's connection intention regarding the currently used bearer by inputting the behavioral parameters and / or environmental parameters into the intention prediction model.

[0123] Next, the terminal device 20 determines whether the communication quality of the bearer currently in use is deteriorating based on the result of step S203 (step S204). For example, the terminal device 20 determines whether the output value of the quality prediction model is below a predetermined threshold. Alternatively, the terminal device 20 may determine whether the communication quality of the bearer currently in use is deteriorating based on the current output value of the sensor unit 24 or the communication unit 25, rather than the prediction result of the quality prediction model.

[0124] If the communication quality has not deteriorated (step S204: No), the terminal device 20 returns to step S202. If the communication quality has deteriorated (step S204: Yes), the terminal device 20 determines whether the user wishes to continue connecting to the bearer currently in use (step S205).

[0125] If the user wishes to continue the connection (step S205: Yes), the terminal device 20 returns to step S202. In this case, the terminal device 20 continues the connection to the current bearer. The terminal device 20 may wait for a predetermined amount of time before returning to step S202.

[0126] On the other hand, if the user does not wish to continue the connection (step S205: No), the terminal device 20 switches the communication using the current bearer to communication using another bearer (step S206). Once the switch is complete, the terminal device 20 terminates the bearer switching process.

[0127] <<4. Variation>> The above-described embodiment is merely an example, and various modifications and applications are possible.

[0128] <4-1. Variations of the application of predictive models> In the embodiment described above, the terminal device 20 used a predictive model generated by the server 10 through federated learning to make predictions about the user's connection intentions, etc. However, the application of predictive models is not limited to this example. The following examples come to mind regarding the application of predictive models.

[0129] (First application example) Installing a trained predictive model on a device. Figure 11 is a diagram illustrating an example of the application of a predictive model. In the first application example, the predictive model is installed on the terminal device 20. In this case, the predictive model installed on the terminal device 20 may be generated by learning within the terminal device 20. In this case, the terminal device 20 learns the user's preferences regarding the bearer to be used and switches bearers based on the output value (predicted value) of the predictive model. Of course, the predictive model installed on the terminal device 20 may also be generated by the server 10 through federated learning.

[0130] (Second application example) Installing the trained model on another device. Figure 12 illustrates another example of the application of the predictive model. In the second application example, the predictive model generated by the terminal device 20 is adapted for use in other communication devices used by the user. These other communication devices are, for example, mobile devices or wearable devices. In this case, the terminal device 20 learns the user's preferences regarding the bearer to be used. The terminal device 20 then transmits the predictive model generated through learning to the other communication device. The other communication device switches bearers based on the output values ​​(predicted values) of the predictive model. This makes it possible to switch bearers based on predicting the user's intent, even in terminals where it is difficult for the other terminal device to grasp the user's intent.

[0131] (Third application example) A system where parameters are reported from the terminal and inference is performed in another location. Figure 13 illustrates another example of the application of the predictive model. In this third application example, terminal device 20 reports parameters to other communication devices on the network. The other communication devices generate a predictive model based on the parameters from terminal device 20. The other communication devices then determine whether terminal device 20 should perform a bearer switch based on the output values ​​(predicted values) of the predictive model and the surrounding environment. Based on the determination result, the other communication devices instruct terminal device 20 to switch bearers. Terminal device 20 performs the bearer switch based on the instruction from the other communication devices. This enables advanced network control that cannot be completed by a single terminal device 20 alone.

[0132] <4-2. Variations regarding labeling> In the above-described embodiment (step S106 of the learning process), the labeling of samples was explained with specific examples, but the examples of labeling are not limited to these examples.

[0133] For example, terminal device 20 may label a sample from when the user turns off wireless communication using the current bearer as a sample from when the user did not want to continue communication using the current bearer.

[0134] Furthermore, the terminal device 20 may label a sample of when the user switches from communication to a predetermined access point using the current bearer to communication to another access point as a sample of when the user did not want to continue communication using the current bearer.

[0135] <4-3. Other variations> The control device for controlling the server 10 and terminal device 20 in this embodiment may be implemented by a dedicated computer system or by a general-purpose computer system.

[0136] For example, a communication program for performing the above-described operations is stored in a computer-readable recording medium such as an optical disc, semiconductor memory, magnetic tape, or flexible disk and distributed. Then, for example, the control device is configured by installing the program on a computer and executing the above-described process. In this case, the control device may be an external device to the server 10 and terminal device 20 (for example, a personal computer). Alternatively, the control device may be an internal device to the server 10 and terminal device 20 (for example, a control unit 13 or control unit 23).

[0137] Alternatively, the above communication program may be stored on a disk device provided by a server on a network such as the Internet, and made available for download to a computer. Furthermore, the above functions may be implemented through the cooperation of an OS (Operating System) and application software. In this case, the parts other than the OS may be stored on a medium and distributed, or the parts other than the OS may be stored on a server and made available for download to a computer.

[0138] Furthermore, among the processes described in the above embodiments, all or part of the processes described as being performed automatically can be performed manually, or all or part of the processes described as being performed manually can be performed automatically by known methods. In addition, the processing procedures, specific names, and information including various data and parameters shown in the above document and drawings can be arbitrarily changed unless otherwise specified. For example, the various information shown in each figure is not limited to the information shown.

[0139] Furthermore, the components of each illustrated device are functionally conceptual and do not necessarily need to be physically configured as shown. In other words, the specific forms of distribution and integration of each device are not limited to those shown, and all or part of them can be functionally or physically distributed and integrated in any unit according to various loads and usage conditions. This distribution and integration configuration may also be performed dynamically.

[0140] Furthermore, the above-described embodiments can be combined as appropriate in areas where the processing content is not contradictory. Also, the order of each step shown in the flowchart of the above-described embodiments can be changed as appropriate.

[0141] Furthermore, for example, this embodiment can also be implemented as any configuration that makes up a device or system, such as a processor as a system LSI (Large Scale Integration), a module using multiple processors, a unit using multiple modules, or a set with additional functions added to a unit (i.e., a configuration of a part of a device).

[0142] In this embodiment, a system refers to a collection of multiple components (devices, modules (parts), etc.), regardless of whether all components are located in the same enclosure. Therefore, multiple devices housed in separate enclosures and connected via a network, and a single device containing multiple modules within a single enclosure, are both considered systems.

[0143] Furthermore, for example, this embodiment can adopt a cloud computing configuration in which a single function is shared and processed collaboratively by multiple devices via a network.

[0144] <<5. Conclusion>> As described above, according to one embodiment of the present disclosure, when the quality of wireless communication using a predetermined bearer (first wireless communication method) satisfies predetermined conditions, the terminal device 20 predicts whether the user wants to continue using the predetermined bearer, using a predictive model of the user's wireless communication behavior. At this time, the predictive model is a learning model generated based on at least one of the user's wireless communication behavior parameters and the wireless communication environment parameters related to said behavior parameters. The terminal device 20 makes a prediction as to whether the user wants to continue using wireless communication using the predetermined bearer at predetermined time intervals after the quality of wireless communication using the predetermined bearer satisfies predetermined conditions. Based on the prediction result, the terminal device 20 switches from wireless communication using the predetermined bearer (first wireless communication method) to wireless communication using another bearer (second wireless communication method).

[0145] This allows the terminal device 20 to reduce unintended switching by the user. As a result, a highly convenient terminal device 20 can be realized.

[0146] Although the embodiments of this disclosure have been described above, the technical scope of this disclosure is not limited to the embodiments described above, and various modifications are possible without departing from the gist of this disclosure. Furthermore, components from different embodiments and modifications may be combined as appropriate.

[0147] Furthermore, the effects described in each embodiment of this specification are merely illustrative and not limiting, and other effects may also occur.

[0148] Furthermore, this technology can also be configured as follows. (1) The system includes a prediction unit that, when the quality of wireless communication using a first wireless communication method satisfies predetermined conditions, uses a predictive model of the user's behavior regarding the wireless communication to predict whether the user wants to continue the wireless communication using the first wireless communication method, The predictive model is a learning model generated based on at least one of the user's behavioral parameters related to wireless communication and environmental parameters related to wireless communication relating to those behavioral parameters. The prediction unit makes a prediction at predetermined time intervals whether the user wants to continue using the wireless communication using the first wireless communication method, after the quality of the wireless communication using the first wireless communication method satisfies the predetermined conditions. Communication device. (2) The system further includes a communication control unit that switches the wireless communication to another wireless communication method based on the results of the prediction. The communication device described in (1) above. (3) The predictive model is a learning model generated based on both the user's behavioral parameters related to wireless communication and the environmental parameters related to wireless communication relating to those behavioral parameters. The communication device described in (1) or (2) above. (4) The predictive model is a learning model generated by aggregating multiple models, each generated based on at least one of the user's behavioral parameters related to wireless communication and the environmental parameters related to wireless communication relating to those behavioral parameters, through federated learning. The communication device described in (1) or (2) above. (5) The aforementioned environmental parameters include at least one of the following: RSSI (Received Signal Strength Indicator), a parameter indicating the number of packets remaining in the communication device, a theoretical value of the communication speed, and a parameter indicating the frequency band being used. A communication device as described in any of (1) to (4) above. (6) The aforementioned environmental parameters include parameters relating to the battery consumption of the communication device. A communication device as described in any of (1) to (5) above. (7) The aforementioned environmental parameters include the importance of the current traffic. A communication device as described in any of (1) to (6) above. (8) The aforementioned environmental parameters include information regarding the current location of the communication device. A communication device as described in any of (1) to (7) above. (9) The behavioral parameters include at least one of the following: a parameter indicating the user's walking state, the acceleration applied to the communication device, and a parameter indicating whether or not reconnection was performed within a short time interval. A communication device as described in any of (1) to (8) above. (10) The aforementioned behavioral parameters include a parameter indicating the user's monthly cumulative data usage. A communication device as described in any of (1) to (9) above. (11) The aforementioned behavioral parameters include parameters related to the application being used by the user. A communication device as described in any of (1) to (10) above. (12) An acquisition unit that acquires at least one of the following parameters: behavioral parameters related to wireless communication using a user's first wireless communication method, and environmental parameters related to said wireless communication relating to said behavioral parameters. The system includes a learning unit that learns a predictive model that predicts whether the user wants to continue the wireless communication using the first wireless communication method when the quality of the wireless communication using the first wireless communication method meets predetermined conditions, The learning unit learns the predictive model based on at least one of the user's behavioral parameters related to wireless communication and the environmental parameters related to wireless communication relating to those behavioral parameters. Information processing device. (13) The learning unit learns the predictive model based on both the user's behavioral parameters related to wireless communication and the environmental parameters related to wireless communication that relate to those behavioral parameters. The information processing device described in (12) above. (14) The system includes a determination unit that determines whether the parameter is a parameter used when the user wanted to continue communication using the first wireless communication method, or a parameter used when the user did not want to continue communication using the first wireless communication method. The learning unit learns the prediction model based on the discrimination result of the discrimination unit and the parameters. The information processing apparatus described in (12) or (13) above. (15) The discrimination unit determines, after disconnection of communication using the first wireless communication method, the parameters when the user reconnects to communication using the first wireless communication method and continues to use the communication for a predetermined time or longer after reconnection, as the parameters when the user wanted to continue using communication using the first wireless communication method. The information processing device described in (14) above. (16) The discrimination unit determines, after disconnecting communication to a predetermined access point using the first wireless communication method, the parameters when the user reconnects to the predetermined access point and continues to maintain the connection to the predetermined access point for a predetermined time or longer after reconnection, as the parameters when the user wanted to continue communication using the first wireless communication method. The information processing device described in (14) or (15) above. (17) The discrimination unit determines that the parameters when the user turns off wireless communication using the first wireless communication method are the parameters when the user does not want to continue communication using the first wireless communication method. An information processing device as described in any of (14) to (16) above. (18) The discrimination unit determines the parameters when the user switches from communication to a predetermined access point using the first wireless communication method to communication to another access point as parameters when the user does not want to continue communication using the first wireless communication method. An information processing device as described in any of (14) to (17) above. (19) The system includes a prediction step in which, when the quality of wireless communication using a first wireless communication method satisfies predetermined conditions, a prediction model of the user's behavior regarding the wireless communication is used to predict whether the user wants to continue the wireless communication using the first wireless communication method, The predictive model is a learning model generated based on at least one of the user's behavioral parameters related to wireless communication and environmental parameters related to wireless communication relating to those behavioral parameters. In the prediction step, a prediction is made at predetermined time intervals after the quality of the wireless communication using the first wireless communication method satisfies the predetermined conditions, to determine whether the user wishes to continue using the wireless communication using the first wireless communication method. Communication method. (20) An acquisition step to acquire behavioral parameters related to wireless communication using a user's first wireless communication method, and environmental parameters related to said wireless communication relating to said behavioral parameters, The system includes a learning step of training a predictive model that predicts whether the user wants to continue the wireless communication using the first wireless communication method when the quality of the wireless communication using the first wireless communication method meets predetermined conditions, In the learning step, the predictive model is trained based on at least one of the user's behavioral parameters related to wireless communication and the environmental parameters related to wireless communication relating to those behavioral parameters. Information processing methods. [Explanation of Symbols]

[0149] 1. Communication System 10 servers 20 Terminal devices 11, 21 Storage section 12, 22, 25 Communications Department 13, 23 Control Unit 24. Recovery Unit 131, 231 Acquisition Department 132 Generation part 232 Learning Department 233 Prediction Section 234 Communication Control Unit N1, N2 Network

Claims

1. The system includes a prediction unit that, when the quality of wireless communication using a first wireless communication method satisfies predetermined conditions, uses a predictive model of the user's behavior regarding the wireless communication to predict whether the user wants to continue the wireless communication using the first wireless communication method, The predictive model is a learning model generated based on at least one of the user's behavioral parameters related to wireless communication and environmental parameters related to wireless communication relating to those behavioral parameters. The prediction unit makes a prediction at predetermined time intervals whether the user wants to continue using the wireless communication using the first wireless communication method, after the quality of the wireless communication using the first wireless communication method satisfies the predetermined conditions. Communication device.

2. The system further includes a communication control unit that switches the wireless communication to another wireless communication method based on the results of the prediction. The communication device according to claim 1.

3. The predictive model is a learning model generated based on both the user's behavioral parameters related to wireless communication and the environmental parameters related to wireless communication relating to those behavioral parameters. The communication device according to claim 1.

4. The predictive model is a learning model generated by aggregating multiple models, each generated based on at least one of the user's behavioral parameters related to wireless communication and the environmental parameters related to wireless communication relating to those behavioral parameters, through federated learning. The communication device according to claim 1.

5. The aforementioned environmental parameters include at least one of the following: RSSI (Received Signal Strength Indicator), a parameter indicating the number of packets remaining in the communication device, a theoretical value of the communication speed, and a parameter indicating the frequency band being used. The communication device according to claim 1.

6. The aforementioned environmental parameters include parameters relating to the battery consumption of the communication device. The communication device according to claim 1.

7. The aforementioned environmental parameters include the importance of the current traffic. The communication device according to claim 1.

8. The aforementioned environmental parameters include information regarding the current location of the communication device. The communication device according to claim 1.

9. The behavioral parameters include at least one of the following: a parameter indicating the user's walking state, the acceleration applied to the communication device, and a parameter indicating whether or not reconnection was performed within a short time interval. The communication device according to claim 1.

10. The aforementioned behavioral parameters include a parameter indicating the user's monthly cumulative data usage. The communication device according to claim 1.

11. The aforementioned behavioral parameters include parameters related to the application being used by the user. The communication device according to claim 1.

12. An acquisition unit that acquires at least one of the following parameters: behavioral parameters related to wireless communication using a user's first wireless communication method, and environmental parameters related to said wireless communication relating to said behavioral parameters. The system includes a learning unit that learns a predictive model that predicts whether the user wants to continue the wireless communication using the first wireless communication method when the quality of the wireless communication using the first wireless communication method meets predetermined conditions, The learning unit learns the predictive model based on at least one of the user's behavioral parameters related to wireless communication and the environmental parameters related to wireless communication relating to said behavioral parameters. Information processing device.

13. The learning unit learns the predictive model based on both the user's behavioral parameters related to wireless communication and the environmental parameters related to wireless communication that relate to those behavioral parameters. The information processing apparatus according to claim 12.

14. The system includes a determination unit that determines whether the parameter is a parameter used when the user wanted to continue communication using the first wireless communication method, or a parameter used when the user did not want to continue communication using the first wireless communication method. The learning unit learns the prediction model based on the discrimination result of the discrimination unit and the parameters. The information processing apparatus according to claim 12.

15. The discrimination unit determines, after disconnection of communication using the first wireless communication method, the parameters when the user reconnects to communication using the first wireless communication method and continues to use the communication for a predetermined time or longer after reconnection, as the parameters when the user wanted to continue using communication using the first wireless communication method. The information processing apparatus according to claim 14.

16. The discrimination unit determines, after disconnecting communication to a predetermined access point using the first wireless communication method, the parameters when the user reconnects to the predetermined access point and continues to maintain the connection to the predetermined access point for a predetermined time or longer after reconnection, as the parameters when the user wanted to continue communication using the first wireless communication method. The information processing apparatus according to claim 14.

17. The discrimination unit determines the parameters when the user turns off wireless communication using the first wireless communication method as the parameters when the user does not want to continue communication using the first wireless communication method. The information processing apparatus according to claim 14.

18. The discrimination unit determines the parameters when the user switches from communication to a predetermined access point using the first wireless communication method to communication to another access point as parameters when the user does not want to continue communication using the first wireless communication method. The information processing apparatus according to claim 14.

19. The system includes a prediction step in which, when the quality of wireless communication using a first wireless communication method satisfies predetermined conditions, a prediction model of the user's behavior regarding the wireless communication is used to predict whether the user wants to continue the wireless communication using the first wireless communication method, The predictive model is a learning model generated based on at least one of the user's behavioral parameters related to wireless communication and environmental parameters related to wireless communication relating to those behavioral parameters. In the prediction step, a prediction is made at predetermined time intervals after the quality of the wireless communication using the first wireless communication method satisfies the predetermined conditions, to determine whether the user wishes to continue using the wireless communication using the first wireless communication method. Communication method.

20. An acquisition step to acquire behavioral parameters related to wireless communication using a user's first wireless communication method, and environmental parameters related to said wireless communication relating to said behavioral parameters, The system includes a learning step of training a predictive model that predicts whether the user wants to continue the wireless communication using the first wireless communication method when the quality of the wireless communication using the first wireless communication method meets predetermined conditions, In the learning step, the predictive model is trained based on at least one of the user's behavioral parameters related to wireless communication and the environmental parameters related to wireless communication relating to those behavioral parameters. Information processing methods.

Images