Connection destination switching control method, communication device, and program

Abstract

A connection destination switching control method executed by a communication device. The connection destination switching control method comprises: a data acquisition step for acquiring an observed value of reception power of a signal transmitted from a base station; a prediction processing step for predicting a future reception power by using, as input data for a prediction model, the observed value acquired by the data acquisition step; and a connection destination switching step for executing control for a handover on the basis of the future reception power predicted by the prediction processing step.

Description

[Technical Field]

[0001] The present invention relates to handover in which a user terminal switches a destination base station in a wireless communication system. [Background technology]

[0002] In a cellular wireless communication system, as a user terminal moves, the base station to which the user terminal is connected is switched, which is called handover (HO).

[0003] The user terminal detects an event that triggers handover processing based on the received power of a signal from a destination base station, which is the base station of the serving cell, and the received power of a signal from a neighboring base station, which is the base station of a neighboring cell (Non-Patent Document 2).

[0004] Incidentally, wireless communication systems beyond 6G will be required to meet more advanced requirements than 5G, such as ultra-high speed and large capacity, ultra-low latency and high reliability, and ultra-multiple connections. To meet these requirements, the use of high frequency bands will be expanded (i.e., cell sizes will be reduced), and complex cell configurations, known as New Network Topologies, in which the areas of multiple cells overlap, are being considered.

[0005] In such a wireless network configuration, it is expected that switching of connection destination base stations will occur more frequently than in the past. [Prior art documents] [Non-patent literature]

[0006] [Non-Patent Document 1] Changqing Luo, Jinlong Ji, Qianlong Wang, Xuhui Chen, and Pan Li, "Channel State Information Prediction for 5G Wireless Communications: A Deep Learning Approach", IEEE TRANSACTIONS ON NETWORK SCIENCE AND ENGINEERING, VOL. 7, NO. 1, JANUARY-MARCH 2020 [Non-patent document 2] Konishi et al., "Effects of Automatic Optimization by Self-Organizing Networks (SON) Technology for LTE / LTE-Advanced Systems" [Non-patent document 3] Docomo 6G White Paper. Summary of the Invention [Problem to be solved by the invention]

[0007] In the future, it is expected that wireless communication systems will use more high-frequency bands (reducing cell sizes) and that many cells will overlap. In this situation, the number of signals from base stations that a user terminal must check for handover conditions will increase, and the received power will fluctuate more rapidly than before.

[0008] This can cause handover processing to become overloaded, making it difficult to perform the handover process smoothly. As a result, it is thought that issues such as a significant drop in wireless quality before the handover or frequent handover processing may occur. In other words, there is a possibility that the user terminal will not be able to perform the handover properly.

[0009] The present invention has been made in view of the above points, and has an object to provide a technique that enables a user terminal to appropriately perform handover in a cellular wireless communication system. [Means for solving the problem]

[0010] According to the disclosed technology, there is provided a connection destination switching control method executed by a communication device, comprising: a data acquisition step of acquiring an observed value of the received power of a signal transmitted from a base station; The data acquired by the data acquisition step , a given section The observed value Statistics obtained by preprocessing a prediction processing step of predicting future received power by using the above as input data to a prediction model; a connection destination switching processing step of executing control for handover based on the future received power predicted by the prediction processing step, The statistical value is the median or average value of the observed values ​​in the predetermined interval. A connection destination switching control method is provided. [Effects of the Invention]

[0011] The disclosed technology provides a technology that enables a user terminal to appropriately perform handover in a cellular wireless communication system. [Brief explanation of the drawings]

[0012] [Figure 1] 1 is a system configuration diagram according to an embodiment of the present invention. [Figure 2] FIG. 10 is a diagram illustrating an example of an event that triggers a handover. [Figure 3] FIG. 10 is a diagram illustrating an example of a handover procedure. [Figure 4] FIG. 2 is a functional configuration diagram of a user terminal 100 according to the first embodiment. [Figure 5] FIG. 2 is a diagram illustrating an example of the hardware configuration of a user terminal 100. [Figure 6] FIG. 2 is a diagram illustrating an example of a prediction model used in the prediction processing unit 120. [Figure 7] FIG. 10 is a diagram illustrating an example of a prediction result. [Figure 8]FIG. 2 is a diagram illustrating an example of the operation of the user terminal 100. [Figure 9] FIG. 10 is a diagram for explaining a problem in receiving power prediction. [Figure 10] FIG. 10 is a functional configuration diagram of a user terminal 100 according to a second embodiment. [Figure 11] FIG. 10 is a functional configuration diagram of a user terminal 100 according to a second embodiment. [Figure 12] FIG. 1 is a diagram illustrating an example of movement of a user terminal 100. [Figure 13] FIG. 10 is a diagram for explaining an outline of a process in the second embodiment. [Figure 14] FIG. 10 is a diagram illustrating an example of input data and output data. [Figure 15] FIG. 10 is a diagram illustrating an example of input data and output data. [Figure 16] FIG. 10 is a diagram illustrating an example of input data and output data. [Figure 17] FIG. 10 is a diagram illustrating an example of input data and output data. [Figure 18] FIG. 10 is a diagram illustrating a processing procedure for acquiring a predicted value according to the second embodiment. [Figure 19] FIG. 10 is a diagram illustrating a processing procedure for acquiring a predicted value according to the second embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0013] Hereinafter, an embodiment of the present invention (the present embodiment) will be described with reference to the drawings. The embodiment described below is merely an example, and the embodiment to which the present invention is applied is not limited to the following embodiment. Below, the configuration and operation of the present embodiment will be described using Example 1 and Example 2.

[0014] Example 1 <System configuration> FIG. 1 illustrates a configuration example of a wireless communication system according to the first embodiment. The wireless communication system according to the first embodiment is a cellular wireless communication system. Although a large number of cells exist in the wireless communication system, only cell 1 and cell 2 are illustrated in FIG. 1. The wireless communication system according to the first embodiment is not limited to a specific system. For example, the wireless communication system according to the first embodiment may be any of 3G, LTE, 5G, and 6G.

[0015] 1, the base station 200-1 forms a cell 1, and the base station 200-2 forms a cell 2. The user terminal 100 connects to the base station of the serving cell and performs wireless communication.

[0016] 1, it is assumed that the user terminal 100 connected to the base station 200-1 is moving toward the cell 2. At this time, the user terminal 100 performs a handover to switch the connected base station 200 from the base station 200-1 to the base station 200-2.

[0017] <Example of handover processing> In the first embodiment, the handover method itself is not particularly limited, but as an example, an example of handover processing will be described with reference to FIGS. 2 and 3 based on the method used in LTE and LTE-Advanced.

[0018] The curves in Figure 2 show the changes in the received power from the serving cell and the received power from the neighboring cell at the user terminal 100. When the user terminal 100 detects an event (e.g., an A3 event) that triggers a handover (HO), it transmits a Measurement Report (MR) to the destination base station 200-1. This triggers the start of handover processing.

[0019] The event detection conditions are, for example, M n +HO offset,s,n >M s This means that the trigger continues for a certain period (TTT: Time to Trigger) or more. sis the received power of the serving cell s, and M n is the received power of neighbor cell n, and HO offset,s,n is an offset value that is uniquely set between cells sn.

[0020] As will be described later, in this embodiment, the user terminal 100 predicts future received power using observed values ​​of received power from the past to the present, and uses this predicted received power to detect an event that will trigger a handover.

[0021] Fig. 3 is a sequence diagram showing an example of handover processing. In the example of Fig. 3, it is assumed that the user terminal 100 switches the connection destination base station from the base station 200-1 to the base station 200-2 by handover.

[0022] At S1, the user terminal 100 detects an event that triggers a handover. At S2, the user terminal 100 transmits an MR to the base station 200-1. The MR includes the cell ID (which may also be called a base station ID) and the measurement result (received power) for each cell whose received power has been measured.

[0023] The base station 200-1, which has received the MR, decides to hand over the user terminal 100 to the adjacent base station 200-2, and in S3 transmits a handover command to the user terminal 100 instructing it to connect to the base station 200-2. The base station 200-1 also transmits information regarding communication with the user terminal 100 to the base station 200-2. Having received the handover command, the user terminal 100 connects to the base station 200-2 in S5, thereby completing the handover.

[0024] As mentioned above, it is expected that cell sizes will become smaller in the future, and that a large number of overlapping cells will occur, which will increase the number of signals from base stations that must be checked for event occurrence conditions, and that received power will fluctuate rapidly. This could result in an overload of handover-related processing, making it impossible to perform smooth handover processing.

[0025] In the first embodiment, in order to solve the above problem, the user terminal 100 uses a prediction technique such as deep learning to predict future (e.g., several seconds ahead) received power for each cell (base station) that is the target of received power measurement, using observation results of received power from the past to the present, thereby smoothly performing handover-related processing. The configuration and operation using a prediction technique such as deep learning will be described in more detail below.

[0026] <Configuration example of user terminal 100> Fig. 4 illustrates an example of a functional configuration of the user terminal 100 according to the first embodiment. As illustrated in Fig. 4, the user terminal 100 includes a data acquisition unit 110, a prediction processing unit 120, a connection destination switching processing unit 130, a prediction input data storage unit 140, and a prediction result data storage unit 150. The user terminal 100 may also be referred to as a communication device.

[0027] The data acquisition unit 110 receives signals from each base station and performs measurements to acquire cell IDs and received power (observed values). The data acquired by the data acquisition unit 110 is stored in the prediction input data holding unit 140 as input data for prediction. The prediction processing unit 120 uses the data read from the prediction input data holding unit 140 as input, predicts future received power from past received power (observed values), and stores the prediction result in the prediction result data holding unit 150.

[0028] The connection destination switching processing unit 130 performs connection destination switching processing (handover control) using the prediction result data read from the prediction result data storage unit 150. The connection destination switching processing includes specifying (narrowing down) the base station to be switched to, transmitting an MR, and processing for connecting to the destination base station after handover.

[0029] In the first embodiment (and the second embodiment), the user terminal 100 is responsible for the received power prediction process and handover control. However, the received power prediction process and handover control may be performed in a base station (which may also be called a communication device). In this case, the configuration of the base station is the same as that shown in FIG. 4 (FIG. 10 in the second embodiment). However, in this case, the base station receives the received power, which is an observed value, from the user terminal 100 and predicts future received power using this observed value. Furthermore, as handover control, the base station notifies the user terminal 100 of the destination base station to which the user terminal 100 will connect, identified based on the future received power.

[0030] <Hardware configuration example> Here, an example of the hardware configuration of the user terminal 100 of the first embodiment will be described. The hardware configuration of the user terminal 100 of the second embodiment is also as described below. Furthermore, the hardware configuration of the base station is also as described below.

[0031] The user terminal 100 (and the base station) can be realized by, for example, causing a computer to execute a program. Examples of such computers include a mobile phone and a smartphone.

[0032] That is, the user terminal 100 can be realized by using hardware resources such as a CPU and memory built into a computer to execute a program corresponding to the processing performed by the user terminal 100. The program can be recorded on a computer-readable recording medium (such as a portable memory) and stored or distributed. The program can also be provided via a network such as the Internet or email.

[0033] Fig. 5 is a diagram showing an example of the hardware configuration of the computer. The computer in Fig. 5 includes a drive device 1000, an auxiliary storage device 1002, a memory device 1003, a CPU 1004, an interface device 1005, a display device 1006, and an input device 1007, all of which are interconnected via a bus B.

[0034] A program for realizing processing on the computer is provided by a recording medium 1001 such as a CD-ROM or a memory card. When the recording medium 1001 storing the program is set in the drive device 1000, the program is installed from the recording medium 1001 to the auxiliary storage device 1002 via the drive device 1000. However, the program does not necessarily have to be installed from the recording medium 1001, but may be downloaded from another computer via a network. The auxiliary storage device 1002 stores the installed program as well as necessary files, data, etc.

[0035] The memory device 1003 reads and stores the program from the auxiliary storage device 1002 when an instruction to start the program is received. The CPU 1004 implements functions related to the user terminal 100 in accordance with the program stored in the memory device 1003. The interface device 1005 is used as an interface for connecting to a network, and functions as an input means and an output means via the network. The display device 1006 displays a GUI (Graphical User Interface) or the like according to the program. The input device 157 is composed of a keyboard, mouse, buttons, a touch panel, or the like, and is used to input various operation instructions.

[0036] <Example of prediction processing unit 120> A prediction model using a neural network is used as the prediction processing unit 120 in the first embodiment. Specifically, a DNN (deep neural network) that performs deep learning is used. The DNN uses an LSTM (long short-term memory), which is one of RNNs (recurrent neural networks) used for time series prediction. Note that the use of an LSTM is just one example. A GRU (gated recurrent unit) may be used instead of or in addition to the LSTM. Also, an RNN other than the LSTM and the GRU may be used.

[0037] FIG. 6 shows an example of a prediction model used in the prediction processing unit 120. The prediction model shown in FIG. 6 has an LSTM layer (50 nodes), three fully connected layers (each with 50 nodes), and one fully connected layer (1 node). Received power (observed values ​​in the first embodiment, and statistical values ​​in the second embodiment) is sequentially input to the prediction model. For example, when predicting received power one second after a certain time t, a predetermined number (e.g., 100) of received power values ​​from a predetermined period (e.g., 10 seconds) before time t to time t are input to the prediction model. The prediction model outputs a predicted value of received power at time t+1 based on the input received power.

[0038] The prediction target is not limited to received power; path loss can also be predicted. Figure 7 is a diagram comparing predictions made using a prediction model that uses LSTM, predictions made using a conventional method (Conv) that does not use LSTM, and actual measurements. The horizontal axis shows samples from observation points, and the vertical axis shows path loss. As shown in Figure 7, LSTM can make predictions that are closer to the actual measurements than the conventional method.

[0039] The learning of the prediction model shown in FIG. 6 may be performed using an existing method such as backpropagation by using the actual measured values ​​as correct data.

[0040] <Processing Procedure> An example of the operation of the user terminal 100 having the configuration of FIG. 4 will be described with reference to FIG. 8. In S101, handover conditions are set in the connection destination switching processing unit 130 of the user terminal 100. The handover conditions are set, for example, as follows: n +HO offset,s,n >M s The condition is that the trigger continues for a certain period of time (TTT: Time to Trigger) or more. offset,s,n and TTT values.

[0041] In the flow of Fig. 8, steps S103 and S104 are performed for each cell (serving cell and each neighboring cell). The determination in step S105 is performed for each pair of a serving cell and a neighboring cell. For ease of explanation, the following description will be given assuming that cell 1 (the cell of base station 200-1) is the serving cell and cell 2 (the cell of base station 200-2) is the neighboring cell.

[0042] In S103, the data acquisition unit 110 of the user terminal 100 observes the received power of the serving cell and the received power of the neighboring cells, and acquires the observed value of the received power for each cell.

[0043] In S104, the prediction processing unit 120 sequentially receives the observed values ​​of received power for each cell, and predicts the future received power for each cell. For example, if the current time is t, the prediction processing unit 120 predicts the received power k seconds after t. k may be 1, 2, 3, 4, 5, 6, 7, 8, 9, or 10, or may be any other value.

[0044] In S105, the connection destination switching processing unit 130 of the user terminal 100 n +HO offset,s,n >M s It is determined whether the time has continued for more than TTT. s is the predicted received power of serving cell s, and M n is the predicted value of the received power of neighboring cell n. If the determination in S105 is Yes, proceed to S106, and if No, proceed to S102. If proceeding to S102, time t is updated, and the processes of S103 to S105 are performed for the next time t.

[0045] When the process proceeds to S106, the connection switch processing unit 130 of the user terminal 100 transmits an MR to the base station 200-1, which is the connection destination base station. The MR includes the cell ID of the base station 200-2, which is the handover destination. As a result, handover processing is performed in S107. An example of the handover processing is as described with reference to FIG. 3.

[0046] As described above, the connection destination switching processing unit 130 can perform control related to handover using the predicted value of the received power k seconds from the current time t. More specifically, any one or more of the processes in Examples 1 to 3 below may be performed.

[0047] <Example 1> In Example 1, the connection destination switch processing unit 130 identifies in advance the connection destination base station to which handover should be performed. By predicting the handover in advance, high wireless communication quality can be maintained.

[0048] For example, assume that there is a serving cell and neighboring cells A, B, and C. In S105 of Fig. 8, when the connection destination switching processing unit 130 determines that only the received power of neighboring cell B satisfies the condition from the predicted values ​​of the received power of each cell, it specifies the base station of neighboring cell B as the destination base station to which handover should be performed. By transmitting an MR early, it is thereby possible to perform handover early and maintain wireless communication quality.

[0049] <Example 2> In Example 2, the connection destination switching processing unit 130 narrows down the connection destination base stations for which handover conditions should be observed in advance. Narrowing down the base stations for which handover should be performed in advance reduces the calculation load related to handover. Note that "identifying" in Example 1 is an example of "narrowing down."

[0050] For example, assume that there is a serving cell and neighboring cells A, B, C, and D. For example, at a certain time, in S104 of Fig. 8, if the connection destination switching processing unit 130 detects that, among the predicted values ​​of the received power of neighboring cells A, B, C, and D, the predicted value of the received power of neighboring cells A and B has become very low (for example, smaller than a certain threshold value) and the predicted value of the received power of neighboring cells C and D is not low (for example, larger than a certain threshold value), the connection destination switching processing unit 130 narrows down the neighboring cells to be observed thereafter and the targets for determining handover conditions to only neighboring cells C and D.

[0051] Narrowing down the observation targets and handover condition determination targets reduces the processing load on the user terminal 100. In addition, the number of neighboring cells reported to the base station in S106 also decreases, reducing the processing load on the base station for determining the handover destination.

[0052] <Example 3> In Example 3, the connection destination switching processing unit 130 identifies a base station to which a handover process will occur again immediately after a handover process, and excludes the base station from candidates for the connection destination base station. This makes it possible to avoid frequent handover processes and maintain high wireless communication quality.

[0053] For example, assume that there is a serving cell and neighboring cells A, B, C, and D. In this example, the prediction processing unit 120 calculates, for each cell, a predicted value of received power k seconds after the current time t, and also calculates a predicted value of received power (k+r) seconds after the current time t. r is a short time, and may be, for example, 1 second, shorter than 1 second, or longer than 1 second.

[0054] The connection destination switching processing unit 130 performs processing to identify a base station to which the currently connected base station will be handed over using the predicted value of the received power k seconds after the current time t (predicted value for each cell) in the steps of S102 to S105 shown in FIG. 8.

[0055] In S105, the connection destination switching processing unit 130 regards the base station of the adjacent cell that is determined to satisfy the handover conditions based on the predicted value of received power after k seconds as the new connection destination base station (referred to as connection destination base station X), and regards the other base stations as base stations of the adjacent cells. When this is done, the connection destination switching processing unit 130 makes the determination in S105 using the calculated predicted value of received power after (k+r) seconds, and if there is a base station that is determined to satisfy the handover conditions for the connection destination base station X, it excludes the connection destination base station X from the candidates for the handover destination base station, and continues the determinations in S102 to S105 based on the predicted value of received power after k seconds from the current time t.

[0056] On the other hand, when the judgment in S105 is made using the predicted value of the received power (k+r) seconds after the current time t, and there is no base station that is judged to satisfy the handover conditions, the destination base station X is determined as the base station to handover to, and in S106, an MR having the cell ID of the destination base station X as the ID of the adjacent cell is transmitted to the current destination base station.

[0057] <Effects of Example 1> The technology described in the first embodiment makes it possible to avoid situations where smooth handover processing is not possible (radio quality is significantly reduced before handover) or handover processing occurs frequently, even when the cell size becomes smaller and cell overlap increases.

[0058] In addition, it is possible to reduce the computational load involved in handover processing by narrowing down the destination base stations whose handover conditions should be observed in advance, or by identifying destination base stations for which handover processing will occur again immediately after handover processing, and excluding those destination base stations from the candidates for destination base stations.

[0059] Example 2 Next, a description will be given of Example 2. It is assumed that the technology according to Example 2 is used in combination with Example 1. However, the technology according to Example 2 may be used independently without being combined with Example 1.

[0060] <Explanation of background and issues regarding Example 2> Various methods for predicting time-series information have been proposed as methods for predicting future information based on information observed from the past to the present. For example, as described in Example 1, there are various methods using deep learning, such as time-series prediction using a recurrent neural network (RNN) and methods based on the RNN, such as a gated recurrent unit (GRU) or a long short-term memory (LSTM).

[0061] In particular, in the field of wireless communications, these prediction methods are used for parameters such as the amplitude and phase information of received signals. For example, these methods are used for signal processing of transmitted signals by predicting future parameters from received signals, such as amplitude and phase information required for signal processing in MIMO (Multiple-Input Multiple-Output) systems that use multiple transmitting and receiving antennas. These predictions are made on the order of μsec to msec, and are focused on predicting instantaneous fluctuations in signals.

[0062] As explained in the first embodiment, when making a prediction on the order of seconds based on the observed value of received power, if the prediction is made using the data of the observed value having instantaneous fluctuations as is using a DNN using LSTM or the like, the time correlation of the received signal may be reduced due to the influence of the instantaneous fluctuations, and the observed value, which is the source information of the prediction, and the information to be predicted may become uncorrelated, making the prediction difficult. Alternatively, when attempting to predict information containing such uncorrelated instantaneous fluctuations, such fluctuations may be regarded as noise in the data, resulting in a result with low prediction accuracy being output.

[0063] Figure 9 shows an example in which a user terminal 100, while moving, predicts the received power at one point later using a DNN with LSTM based on the received power at the previous 50 points. The horizontal axis in Figure 9 indicates the observation point, and the vertical axis indicates the received power. In the example in Figure 9, the instantaneous fluctuations of Nakagami-Rice fading (K factor = 10 dB) are shown as observed values, with predicted values ​​indicated in white. As shown in the lower right, errors occur in the predicted values. In other words, signal fluctuations cannot be predicted, and the fluctuations become noise, making stable predictions impossible.

[0064] <Outline of prediction method in Example 2> In the second embodiment, a DNN having an LSTM or the like as shown in Fig. 6 is used as a prediction model, as in the first embodiment. In order to solve the above problem, in the second embodiment, statistical values ​​(median, average, etc.) at a sequence length that take into account the Doppler frequency (frequency or moving speed) of the received signal are used for the input data and output data to the prediction model, thereby enabling prediction that eliminates the influence of instantaneous fluctuations in the received signal.

[0065] <Device configuration example> The system configuration in the second embodiment is the same as that in the first embodiment, as shown in Fig. 1. Fig. 10 shows a functional configuration diagram of a user terminal 100 in the second embodiment. In the second embodiment, the above-mentioned statistical values ​​(values ​​obtained by pre-processing observed values) are used in predicting the received power for each cell in the first embodiment. Therefore, as shown in Fig. 10, the user terminal 100 in the second embodiment has a configuration in which a data pre-processing unit 160 is added to the user terminal 100 in the first embodiment.

[0066] The data pre-processing unit 160 reads out the observed values ​​stored in the prediction input data holding unit 140, performs pre-processing, calculates statistical values, and inputs the statistical values ​​to the prediction processing unit 120 (prediction model).

[0067] For example, when predicting the received power one second after a certain time t, a predetermined number (e.g., 100) of statistical values ​​from a predetermined period (e.g., 10 seconds) before time t to time t are input to the prediction processing unit 120 (prediction model). The prediction processing unit 120 (prediction model) outputs a predicted value of the received power at time t+1 based on the series of input statistical values. The processing of the data acquisition unit 110 and the connection destination switching processing unit 130 is as described in the first embodiment.

[0068] The prediction method in the second embodiment can be applied to other purposes besides prediction of received power for handover processing described in the first embodiment. That is, the configuration of the user terminal 100 in the second embodiment may be the configuration shown in FIG. 11 , which does not include the connection destination switching processing unit 130. In the case of the configuration in FIG. 11 , the prediction result obtained by the prediction processing unit 120 is output from the prediction result output unit 170.

[0069] <Operation overview> An outline of the operation of the user terminal 100 in the second embodiment regarding the prediction of received power will be described with reference to FIGS.

[0070] Figure 12 shows the movement trajectory of the user terminal 100. Figure 12 shows a situation in which the user terminal 100 moves in a straight line from a lattice point ● to another ●. When the user terminal 100 reaches a certain ●, it changes direction and moves on to another ◯.

[0071] Furthermore, the user terminal 100 moves, for example, 0.1 m every 0.1 seconds, and the data acquisition unit 110 measures the received power every 0.1 seconds. In this situation, the data acquisition unit 110 of the user terminal 100 acquires raw data (data having instantaneous fluctuations) shown in the image of FIG. 13 and stores the acquired data in the prediction input data holding unit 140.

[0072] For example, the data pre-processing unit 160 calculates the median of data for a 10-second interval (i.e., 100 points) while shifting the data by one point (one sample) at a time. That is, by shifting the data by one sample at a time, the moving median (or a moving average value) is calculated. For example, the data pre-processing unit 160 sequentially inputs the moving medians of 200 points from a point 200 points before the current point to the prediction processing unit 120, and the prediction processing unit 120 predicts the received power in the future (e.g., one second from now) relative to the current point. This processing can reduce the influence of instantaneous fluctuations, enabling handover processing that is not overly sensitive to instantaneous fluctuations.

[0073] <Detailed operation example> Next, an example of operation (input / output data) of the data pre-processing unit 160 and the prediction processing unit 120 (prediction model) will be described with reference to Figs. 14 to 17. Here, it is assumed that the information to be predicted is received power. In Figs. 14 to 17, t indicates the current time, N indicates the number of past series (number of data) of time-series data used for prediction, and M indicates the number of time series to be predicted (i.e., predicting received power at a time (or point) M ahead of t).

[0074] In any of the examples, the prediction model can be learned using an existing method such as backpropagation by using statistical values ​​calculated from actual measurements as correct answer data. In the following examples, the median is used as the statistical value, but this is just an example, and a value other than the median (for example, the average value) may also be used as the statistical value.

[0075] For reference, Fig. 14 shows input / output data in the prior art. N+1 data (non-preprocessed instantaneous fluctuation data) at each point in time (each time) from tN to t are input to the prediction model, and a predicted value at time t+M (predicted value of instantaneous fluctuation data) is output.

[0076] FIG. 15 shows Example 1 of input / output data in Example 2. In Example 1, prediction is performed using both input data and output data as the median value of a predetermined interval. The predetermined interval may be a time (e.g., 10 seconds) or a travel distance (e.g., 10 m) of the user terminal 100. The same applies to the "predetermined interval" below.

[0077] Instantaneous fluctuation data is input to the data pre-processing unit 160. The data pre-processing unit 160 calculates the interval median while moving the interval by one piece of time-series data, and sequentially inputs the calculated interval median values ​​to the prediction model. The prediction model outputs a predicted value at time t+M (predicted value of the predetermined interval median) based on the input data.

[0078] 16 shows Example 2 of input / output data in Example 2. In Example 2, the input data is data that includes instantaneous fluctuations (data that has not been pre-processed), and prediction is performed using the output data as the median value of a predetermined interval.

[0079] N+1 data (instantaneous fluctuation data that has not been pre-processed) at each point (each time) from tN to t are input into the prediction model, and a predicted value at the point t+M (predicted value of the median value of a specified interval) is output.

[0080] FIG. 17 shows Example 3 of input / output data in Example 2. In Example 3, the input data is a plurality of data series, such as data including instantaneous fluctuations, first interval median data, and second interval median data, and the output data is predicted using, for example, the second interval median. For example, the first interval may be 5 m and the second interval may be 10 m, or the first interval may be 5 seconds and the second interval may be 10 seconds. Also, the number of data series in the plurality of data series is, as mentioned above, three, which is one example. The number of data series in the plurality of data series may be two or a number greater than three.

[0081] Instantaneous fluctuation data is input to the data pre-processing unit 160. The data pre-processing unit 160 calculates a first interval median while moving one time-series data unit through a first interval (e.g., interval length P). The data pre-processing unit 160 also calculates a second interval median while moving one time-series data unit through a second interval (e.g., interval length Q). The data pre-processing unit 160 sequentially inputs the instantaneous fluctuation data, the first interval median, and the second interval median into a prediction model. The prediction model outputs a predicted value at time t+M (a predicted value of the predetermined interval median) based on the input data.

[0082] In the above examples 1 to 3, the processing interval for statistical values ​​may be an interval length that takes into account the environment in the radio wave propagation field and short-term fluctuations for each frequency (for example, a 10 m interval at 4.5 GHz), or the interval length may be dynamically set according to the location and moving speed of the user terminal. Also, for example, if the moving speed of the user terminal 100 is equal to or less than a predetermined threshold, the interval length may be set to a time unit such as 10 seconds.

[0083] <Processing flow> The processing of the data pre-processing unit 160 and the prediction processing unit 120 will be described with reference to FIGS.

[0084] 18 corresponds to Examples 1 and 3. In S202, the data pre-processing unit 160 checks whether the time-series data required for statistical value processing is stored in the prediction input data holding unit 140, and if it is stored, proceeds to S203. If it is not stored, wait until it is stored before proceeding to S203.

[0085] In S203, the data pre-processing unit 160 reads time-series data from the prediction input data holding unit 140 and performs statistical processing (median, average, etc.) on the time-series data. In S204, the data pre-processing unit 160 inputs the statistically processed data to the prediction processing unit 120 and performs prediction. In S205, the prediction processing unit 120 obtains and outputs a predicted value.

[0086] 19 corresponds to Example 2. In S302, the prediction processing unit 120 checks whether time-series data required for prediction is stored in the prediction input data holding unit 140, and if it is stored, proceeds to S303. If it is not stored, wait until it is stored and then proceeds to S303.

[0087] In S303, the prediction processing unit 120 receives time-series data (instantaneous fluctuation data) and performs prediction. In S304, the prediction processing unit 120 obtains and outputs predicted values ​​of statistical values ​​(median, average, etc.).

[0088] <Effects of Example 2> According to the second embodiment, by using statistical values ​​(median or average) according to radio wave propagation characteristics as data to be predicted, it is possible to eliminate the influence of high-speed instantaneous fluctuations and achieve improved prediction accuracy. For example, since accurate prediction on the order of seconds is possible, it becomes possible to utilize the method for the handover process described in the first embodiment. Note that, for prediction on the order of seconds, there are methods that incorporate external information such as camera information in addition to parameters of received signals, and the technology according to the second embodiment can also be applied to such methods.

[0089] (Summary of the embodiment) This specification discloses at least the connection destination switching control method, communication device, and program described in the following items. (Section 1) A connection destination switching control method executed by a communication device, a data acquisition step of acquiring an observed value of the received power of a signal transmitted from a base station; a prediction processing step of predicting future received power by using the observed values ​​acquired in the data acquisition step as input data for a prediction model; a connection destination switching processing step for executing control for handover based on the future received power predicted by the prediction processing step; A connection destination switching control method comprising: (Section 2) In the connection switching processing step, the base stations to which the handover is to be connected are narrowed down based on the future reception power. 2. The connection destination switching control method according to claim 1. (Section 3) In the connection destination switching processing step, a connection destination base station where a handover process will occur again immediately after the handover process is identified based on the future reception power, and the connection destination base station is excluded from candidates for the connection destination base station, thereby narrowing down the connection destination base stations for the handover destination. 3. A connection destination switching control method according to claim 1 or 2. (Section 4) The statistical values ​​obtained by pre-processing the observed values ​​acquired in the data acquisition step are used as input data for the prediction model. 4. A connection destination switching control method according to any one of claims 1 to 3. (Section 5) The statistical value is the median or average value of the observed values ​​in a predetermined interval. 5. A connection destination switching control method according to claim 4. (Section 6) a data acquisition unit that acquires an observed value of the received power of a signal transmitted from a base station; a prediction processing unit that predicts future received power by using the observed value acquired by the data acquisition unit as input data for a prediction model; a connection destination switching processing unit that executes control for handover based on the future received power predicted by the prediction processing unit; A communication device comprising: (Section 7) A program for causing a computer to function as each unit in the communication device described in item 6.

[0090] Although the present embodiment has been described above, the present invention is not limited to such a specific embodiment, and various modifications and changes are possible within the scope of the gist of the present invention described in the claims. [Explanation of symbols]

[0091] 100 user terminals 110 Data Acquisition Unit 120 Prediction processing unit 130 Connection destination switching processing unit 140 Prediction input data storage unit 150 Prediction result data storage unit 160 Data preprocessing section 170 Prediction result output section 1000 Drive Device 1001 Recording media 1002 Auxiliary storage 1003 Memory device 1004 CPU 1005 Interface device 1006 Display device 1007 Input Device

Claims

1. A connection destination switching control method executed by a communication device, a data acquisition step of acquiring an observed value of the received power of a signal transmitted from a base station; a prediction processing step of predicting future received power by using statistical values ​​obtained by pre-processing the observed values ​​in a predetermined section acquired in the data acquisition step as input data for a prediction model; a connection destination switching processing step of executing control for handover based on the future received power predicted by the prediction processing step, The statistical value is the median or average value of the observed values ​​in the predetermined interval. Connection destination switching control method.

2. In the connection switching processing step, the base stations to which the handover is to be connected are narrowed down based on the future reception power. The connection destination switching control method according to claim 1 .

3. In the connection destination switching processing step, a connection destination base station where a handover process will occur again immediately after the handover process is identified based on the future reception power, and the connection destination base station is excluded from candidates for the connection destination base station, thereby narrowing down the connection destination base stations for the handover destination.

3. The connection destination switching control method according to claim 1 or 2.

4. a data acquisition unit that acquires an observed value of the received power of a signal transmitted from a base station; a prediction processing unit that predicts future received power by using statistical values ​​obtained by performing pre-processing on the observed values ​​in a predetermined section acquired by the data acquisition unit as input data for a prediction model; and a connection destination switching processing unit that executes control for handover based on the future received power predicted by the prediction processing unit, The statistical value is the median or average value of the observed values ​​in the predetermined interval. Communication equipment.

5. A program for causing a computer to function as each unit in the communication device according to claim 4.

Images