WO2023135729A1 - Inference device, inference method, and program - Google Patents

Inference device, inference method, and program Download PDF

Info

Publication number
WO2023135729A1
WO2023135729A1 PCT/JP2022/001057 JP2022001057W WO2023135729A1 WO 2023135729 A1 WO2023135729 A1 WO 2023135729A1 JP 2022001057 W JP2022001057 W JP 2022001057W WO 2023135729 A1 WO2023135729 A1 WO 2023135729A1
Authority
WO
WIPO (PCT)
Prior art keywords
base station
data
traffic data
station traffic
user
Prior art date
Application number
PCT/JP2022/001057
Other languages
French (fr)
Japanese (ja)
Inventor
佑紀奈 高野
恵 竹下
Original Assignee
日本電信電話株式会社
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 日本電信電話株式会社 filed Critical 日本電信電話株式会社
Priority to PCT/JP2022/001057 priority Critical patent/WO2023135729A1/en
Publication of WO2023135729A1 publication Critical patent/WO2023135729A1/en

Links

Images

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04L—TRANSMISSION OF DIGITAL INFORMATION, e.g. TELEGRAPHIC COMMUNICATION
    • H04L43/00—Arrangements for monitoring or testing data switching networks
    • H04L43/08—Monitoring or testing based on specific metrics, e.g. QoS, energy consumption or environmental parameters
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D30/00—Reducing energy consumption in communication networks
    • Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks

Definitions

  • the present invention relates to an estimation device, an estimation method, and a program.
  • Non-Patent Document 1 proposes a method of obtaining an encrypted packet of a user and estimating the application used by the user from the packet.
  • Non-Patent Document 2 proposes a method of predicting the application to be used next time, based on information on the user's current application, place of use, and time of use.
  • Non-Patent Literature 3 treats the communication usage status of users in a base station collectively, It proposes a network management concept based on
  • Non-Patent Document 1 packets for each user are collected and used, but it is difficult for telecommunications carriers to collect such information directly due to privacy issues. Moreover, even if direct collection is possible, the amount of data will increase explosively as the number of users and the amount of communication increase, resulting in the problem of high data acquisition and storage costs. As for non-patent document 2, the same problem occurs because the user's usage application information itself is collected.
  • Non-Patent Document 3 Although the problems in Non-Patent Document 1 and Non-Patent Document 2 can be solved, it is a premise that the user set characteristics in the area corresponding to the base station can be directly acquired. There is still a problem with the method.
  • the present invention has been made in view of the above points, and an object of the present invention is to enable estimation of user set characteristics corresponding to a base station.
  • an estimation device includes an acquisition unit configured to acquire base station traffic data for a certain base station in a certain period; By inputting the base station traffic data acquired by the acquisition unit into a regression model learned based on a plurality of sets of aggregate characteristic data, user aggregate characteristic data for the certain base station in the certain period is estimated.
  • an estimator configured to:
  • FIG. 11 is a flowchart for explaining an example of a processing procedure of a regression model learning process
  • FIG. 4 is a diagram showing an example of UE number distribution of data rates
  • FIG. 10 is a diagram showing an example of the UE number distribution of the elapsed time of the Inactive Timer
  • FIG. 10 is a flowchart for explaining an example of a processing procedure for estimating user set characteristic data
  • an estimating apparatus 10 that estimates user aggregate characteristic data using base station traffic data as an input.
  • base station traffic data refers to statistical network traffic data that reflects the behavior of users' communication usage in the area covered by the base station. That is, the base station traffic data is data indicating the traffic status of the base station.
  • the user set characteristics data is data indicating characteristics of applications or terminals used by a user set under the control of a base station (a user set of terminals connected to a base station).
  • the estimating apparatus 10 first learns a regression model with base station traffic data as input and user set characteristic data as output, based on a plurality of sets of learning data each consisting of base station traffic data and user set characteristic data. .
  • the estimator 10 uses the learned regression model to estimate user aggregate characteristic data based on base station traffic data for a base station at a point in time.
  • FIG. 1 is a diagram showing a hardware configuration example of the estimation device 10 according to the embodiment of the present invention.
  • the estimating device 10 of FIG. 1 has a drive device 100, an auxiliary storage device 102, a memory device 103, a processor 104, an interface device 105, etc., which are connected to each other via a bus B, respectively.
  • a program that implements the processing in the estimation device 10 is provided by a recording medium 101 such as a CD-ROM.
  • a recording medium 101 such as a CD-ROM.
  • the program is installed from the recording medium 101 to the auxiliary storage device 102 via the drive device 100 .
  • the program does not necessarily need to be installed from the recording medium 101, and may be downloaded from another computer via the network.
  • the auxiliary storage device 102 stores installed programs, as well as necessary files and data.
  • the memory device 103 reads and stores the program from the auxiliary storage device 102 when a program activation instruction is received.
  • the processor 104 is a CPU or a GPU (Graphics Processing Unit), or a CPU and a GPU, and executes functions related to the estimation device 10 according to programs stored in the memory device 103 .
  • the interface device 105 is used as an interface for connecting to a network.
  • FIG. 2 is a diagram showing a functional configuration example of the estimation device 10 according to the embodiment of the present invention.
  • the estimation device 10 has a learning data collection unit 11, a model construction unit 12, a base station traffic data acquisition unit 13, and a user group characteristics estimation unit . Each of these units is implemented by processing that one or more programs installed in the estimation apparatus 10 cause the processor 104 to execute.
  • FIG. 3 is a flowchart for explaining an example of a processing procedure of a regression model learning process.
  • the learning data collection unit 11 collects user set characteristic data and base station traffic data corresponding thereto for a plurality of base stations at a plurality of timings.
  • the collected user set characteristic data and the corresponding base station traffic data may be user set characteristic data and base station traffic data generated by network simulation, or user set characteristic data obtained by crowdsourcing or the like. (information indicating the application currently in use) and base station traffic data corresponding to the user set.
  • the "corresponding base station traffic data” refers to the base station traffic data collected at the same timing as the collection timing of the user set characteristics data in relation to the same base station as the target base station for collecting the user set characteristics data. say.
  • the base station traffic data here refers to, for example, downlink (DL)/uplink (UL) data rate UE number distribution, Inactive Timer elapsed time UE number distribution, downlink (DL)/uplink Statistical network traffic data that reflects the behavior of users' communication usage, such as the resource block usage rate of (UL).
  • the base station traffic data may be data containing some (one or more) of these parameters, or may be data containing all of these parameters.
  • the DL/UL data rate UE number distribution corresponds to "DL scheduled IP throughput distribution” and "UL scheduled IP throughput distribution” respectively, and the UE number distribution of the Inactive Timer elapsed time corresponds to "Number of successful RRC connection setups in relation to the time between successful RRC connection setup and last RRCconnection release".
  • An example of the UE number distribution of the data rate is shown in FIG.
  • the DL/UL resource block usage rate corresponds to "DL Total PRB Usage” and "UL Total PRB Usage” and is a value of 0 to 100 [%].
  • base station traffic data depends on the vendor that manufactures the base station, but basically the base station has a data generation cycle, and data is generated at that cycle and timing. For example, if the data generation cycle is one minute, the data rate and resource block usage rate are collected as average values for that one minute, and the Inactive Timer elapsed time is collected as instantaneous values at the collection timing.
  • An example of user set characteristic data is the ratio of the number of users (UE ratio) for each application type under the target base station.
  • UE ratio the ratio of the number of users
  • the ratio of the number of users for each terminal type may be used as the user set characteristic data, or the ratio of the number of users for each combination of the application type and the terminal type may be used.
  • the population ratio is, for example, the value in the period corresponding to the collection period of the base station traffic data corresponding to the user set characteristics data, out of the generation period of the base station traffic data described above. is. For example, if the generation cycle of base station traffic data is one minute, the ratio of the number of users for each application type in one minute during which the base station traffic data corresponding to the user set characteristic data is collected becomes the user set characteristic data.
  • the inventor of the present application believes that the amount of DL data per user increases in areas where video applications are frequently used, the amount of DL data per user decreases in areas where web use is high, and in areas where web conference applications are frequently used
  • the DL/UL data volume per user increases, and in areas where voice usage is high, the DL/UL data volume per user decreases. thought to appear in data items. Therefore, the inventor of the present application considered that the UE number distribution of the DL/UL average data rate can be used as an indication of the DL/UL data amount. This value represents the amount of data that has flowed per subframe, ie the smallest scheduling time unit.
  • the inventor of the present application has another hypothesis that in areas where video is frequently used, connections occur every chunk period, and in areas where audio and web conferences are often used, connections occur frequently, and web use is frequent.
  • connection timing such as connections occurring at sparse timing in the area. Therefore, the inventor of the present application extracted the item of the UE number distribution of the Inactive Timer elapsed time, which represents the elapsed time from the last connection of each user.
  • the inventor of the present application believes that resource consumption is high in areas where there is a large amount of data used for video and web conferencing, and resource consumption is low in areas where there is a large amount of data used for voice and web.
  • resource block usage rate which represents the rate of resource consumption of the base station.
  • the model construction unit 12 uses, as learning data, a plurality of sets of the user set characteristic data and the corresponding base station traffic data collected by the learning data collection unit 11, and receives the base station traffic data as an input.
  • a regression model outputting the set characteristic data is learned (S102). Any regression method such as linear regression or random forest regression can be used as the regression method used here.
  • the model construction unit 12 outputs the learned regression model (for example, outputs the values of the learning parameters of the regression model) (S103).
  • FIG. 6 is a flowchart for explaining an example of the processing procedure for estimating user group characteristic data.
  • the base station traffic data acquisition unit 13 acquires base station traffic data for a period for which user group characteristic data is to be clarified and network base station traffic data in response to an input by the user of the estimation device 10 (S201).
  • the period and the network are specified by the user of the estimation device 10, for example.
  • the network is, for example, any network (or area) among networks (or areas) distinguished by base station unit, sector unit, sector ⁇ carrier (frequency band) unit, or the like.
  • the base station traffic data is composed of the same parameters as the parameters collected by the learning data collection unit 11 .
  • the user aggregate characteristics estimation unit 14 inputs the base station traffic data acquired by the base station traffic data acquisition unit 13 into the learned regression model, thereby estimating the user aggregate characteristics data for the period T of the network. (S202). That is, the user set characteristic estimation unit 14 acquires the user set characteristic data output by the regression model as an estimated value of the user set characteristic data for the period T of the network.
  • the user group characteristic estimation unit 14 outputs the user group characteristic data (S203).
  • the base station traffic data acquisition unit 13 is an example of an acquisition unit.
  • the user group characteristics estimating unit 14 is an example of an estimating unit.
  • the model construction unit 12 is an example of a learning unit.
  • estimation device 11 learning data collection unit 12 model construction unit 13 base station traffic data acquisition unit 14 user group characteristic estimation unit 100 drive device 101 recording medium 102 auxiliary storage device 103 memory device 104 processor 105 interface device B bus

Abstract

This inference device has: an acquisition unit configured so as to acquire base station traffic data in a certain period regarding a certain base station; and an inference unit configured to infer user cluster characteristic data in the certain period regarding the certain base station by receiving input of the base station traffic data acquired by the acquisition unit to a regression model obtained through learning on the basis of a plurality of sets of base station traffic data and user cluster characteristic data that are collected as learning data. As a result, the inference device can infer user cluster characteristics corresponding to the base station.

Description

推定装置、推定方法及びプログラムEstimation device, estimation method and program
　本発明は、推定装置、推定方法及びプログラムに関する。 The present invention relates to an estimation device, an estimation method, and a program.
　近年、スマートフォンの普及に伴い、モバイルトラヒックが急増している。そのため、ネットワーク事業者にとって、ユーザが満足する通信ネットワークを提供することが重要となっている。特に、近年は様々なアプリケーション及び利用端末等のユーザの通信利用状況においてモバイルネットワークが利用されているため、ユーザの通信利用状況を把握することは、ネットワーク設計や制御、ユーザエンゲージメントの向上等を行うために重要となる。一方で、ネットワーク管理にはネットワーク設備構築費（ＣＡＰＥＸ：CAPital EXpenditure）と保守運用費（ＯＰＥＸ：OPerating EXpense）といったコストがかかるため、これらをなるべく抑えつつユーザの満足度を向上させるといった、効率的なネットワーク管理を行うことが求められる。 In recent years, with the spread of smartphones, mobile traffic has increased rapidly. Therefore, it is important for network operators to provide communication networks that satisfy users. In particular, in recent years, mobile networks have been used in various applications and user communication usage situations such as terminals. important for On the other hand, network management involves costs such as network facility construction costs (CAPEX: CAPital EXpenditure) and maintenance and operation costs (OPEX: OPerating EXpense). Network management is required.
　ユーザの利用アプリや利用端末等の通信利用状況の情報の把握のため、その推定について、これまで様々な検討がなされている。 　In order to understand information on the communication usage status of the user's applications and devices, various studies have been conducted so far on the estimation.
　例えば、非特許文献１では、暗号化されたユーザのパケットを取得し、そのパケットからユーザの利用アプリケーションを推定する手法が提案されている。 For example, Non-Patent Document 1 proposes a method of obtaining an encrypted packet of a user and estimating the application used by the user from the packet.
　非特許文献２では、ユーザの現在の利用アプリ、利用場所、利用時間の情報に基づいて、次の時間の利用アプリケーションの予測を行う手法が提案されている。 Non-Patent Document 2 proposes a method of predicting the application to be used next time, based on information on the user's current application, place of use, and time of use.
　ユーザの利用アプリや利用端末等の通信利用状況の情報を扱ったネットワーク管理の検討について、非特許文献３では、基地局内のユーザの通信利用状況を集合的に扱い、基地局内のユーザ集合特性に基づいたネットワーク管理のコンセプトを提案している。 Regarding the study of network management that deals with communication usage information such as applications and terminals used by users, Non-Patent Literature 3 treats the communication usage status of users in a base station collectively, It proposes a network management concept based on
　非特許文献１では、ユーザ単位のパケットを収集し利用することとなっているが、通信事業者がそのような情報を直接収集することは、プライバシーの問題等から困難である。また、仮に直接収集することが可能だとしても、ユーザ数や通信量の増加に従いデータ量が爆発的に増加するため、データの取得や蓄積コスト等がかかるという問題が発生する。非特許文献２についても、ユーザの利用アプリ情報そのものを収集しているため、同様の問題が発生する。 In Non-Patent Document 1, packets for each user are collected and used, but it is difficult for telecommunications carriers to collect such information directly due to privacy issues. Moreover, even if direct collection is possible, the amount of data will increase explosively as the number of users and the amount of communication increase, resulting in the problem of high data acquisition and storage costs. As for non-patent document 2, the same problem occurs because the user's usage application information itself is collected.
　非特許文献３の検討では、非特許文献１や非特許文献２における課題を解決可能なものの、基地局に対応するエリア内のユーザ集合特性が直接取得可能な前提となっているため、その取得方法に課題が残る。 In the study of Non-Patent Document 3, although the problems in Non-Patent Document 1 and Non-Patent Document 2 can be solved, it is a premise that the user set characteristics in the area corresponding to the base station can be directly acquired. There is still a problem with the method.
　本発明は、上記の点に鑑みてなされたものであって、基地局に対応するユーザ集合特性を推定可能とすることを目的とする。 The present invention has been made in view of the above points, and an object of the present invention is to enable estimation of user set characteristics corresponding to a base station.
　そこで上記課題を解決するため、推定装置は、或る基地局に関する或る期間における基地局トラヒックデータを取得するように構成されている取得部と、学習データとして収集された基地局トラヒックデータ及びユーザ集合特性データの複数の組に基づいて学習された回帰モデルに前記取得部が取得した基地局トラヒックデータを入力することで、前記或る基地局に関する前記或る期間におけるユーザ集合特性データを推定するように構成されている推定部と、を有する。 Therefore, in order to solve the above problems, an estimation device includes an acquisition unit configured to acquire base station traffic data for a certain base station in a certain period; By inputting the base station traffic data acquired by the acquisition unit into a regression model learned based on a plurality of sets of aggregate characteristic data, user aggregate characteristic data for the certain base station in the certain period is estimated. an estimator configured to:
　基地局に対応するユーザ集合特性を推定可能とすることができる。 　It is possible to estimate the user set characteristics corresponding to the base station.
本発明の実施の形態における推定装置１０のハードウェア構成例を示す図である。It is a figure which shows the hardware structural example of the estimation apparatus 10 in embodiment of this invention. 本発明の実施の形態における推定装置１０の機能構成例を示す図である。It is a figure which shows the functional structural example of the estimation apparatus 10 in embodiment of this invention. 回帰モデルの学習処理の処理手順の一例を説明するためのフローチャートである。FIG. 11 is a flowchart for explaining an example of a processing procedure of a regression model learning process; FIG. データレートのＵＥ数分布の一例を示す図である。FIG. 4 is a diagram showing an example of UE number distribution of data rates; Ｉｎａｃｔｉｖｅ　Ｔｉｍｅｒ経過時間のＵＥ数分布の一例を示す図である。FIG. 10 is a diagram showing an example of the UE number distribution of the elapsed time of the Inactive Timer; ユーザ集合特性データの推定処理の処理手順の一例を説明するためのフローチャートである。FIG. 10 is a flowchart for explaining an example of a processing procedure for estimating user set characteristic data; FIG.
　以下、図面に基づいて本発明の実施の形態を説明する。本実施の形態では、基地局トラヒックデータを入力として、ユーザ集合特性データを推定する推定装置１０が開示される。ここで、基地局トラヒックデータとは、基地局がカバーするエリアにおけるユーザの通信利用による挙動が反映されている統計的なネットワークトラヒックデータをいう。すなわち、基地局トラヒックデータは、基地局のトラヒック状況を示すデータである。また、ユーザ集合特性データは、基地局配下のユーザ集合（基地局に接続する端末群のユーザ集合）が利用するアプリケーション又は端末に関する特性を示すデータをいう。 Embodiments of the present invention will be described below based on the drawings. In this embodiment, an estimating apparatus 10 is disclosed that estimates user aggregate characteristic data using base station traffic data as an input. Here, base station traffic data refers to statistical network traffic data that reflects the behavior of users' communication usage in the area covered by the base station. That is, the base station traffic data is data indicating the traffic status of the base station. Also, the user set characteristics data is data indicating characteristics of applications or terminals used by a user set under the control of a base station (a user set of terminals connected to a base station).
　推定装置１０は、まず、基地局トラヒックデータ及びユーザ集合特性データを一組とする複数組の学習データに基づき、基地局トラヒックデータを入力とし、ユーザ集合特性データを出力とする回帰モデルを学習する。その後、推定装置１０は、学習済みの回帰モデルを使用して、或る基地局に関する或る時点における基地局トラヒックデータに基づいてユーザ集合特性データを推定する。 The estimating apparatus 10 first learns a regression model with base station traffic data as input and user set characteristic data as output, based on a plurality of sets of learning data each consisting of base station traffic data and user set characteristic data. . The estimator 10 then uses the learned regression model to estimate user aggregate characteristic data based on base station traffic data for a base station at a point in time.
　図１は、本発明の実施の形態における推定装置１０のハードウェア構成例を示す図である。図１の推定装置１０は、それぞれバスＢで相互に接続されているドライブ装置１００、補助記憶装置１０２、メモリ装置１０３、プロセッサ１０４、及びインタフェース装置１０５等を有する。 FIG. 1 is a diagram showing a hardware configuration example of the estimation device 10 according to the embodiment of the present invention. The estimating device 10 of FIG. 1 has a drive device 100, an auxiliary storage device 102, a memory device 103, a processor 104, an interface device 105, etc., which are connected to each other via a bus B, respectively.
　推定装置１０での処理を実現するプログラムは、ＣＤ－ＲＯＭ等の記録媒体１０１によって提供される。プログラムを記憶した記録媒体１０１がドライブ装置１００にセットされると、プログラムが記録媒体１０１からドライブ装置１００を介して補助記憶装置１０２にインストールされる。但し、プログラムのインストールは必ずしも記録媒体１０１より行う必要はなく、ネットワークを介して他のコンピュータよりダウンロードするようにしてもよい。補助記憶装置１０２は、インストールされたプログラムを格納すると共に、必要なファイルやデータ等を格納する。 A program that implements the processing in the estimation device 10 is provided by a recording medium 101 such as a CD-ROM. When the recording medium 101 storing the program is set in the drive device 100 , the program is installed from the recording medium 101 to the auxiliary storage device 102 via the drive device 100 . However, the program does not necessarily need to be installed from the recording medium 101, and may be downloaded from another computer via the network. The auxiliary storage device 102 stores installed programs, as well as necessary files and data.
　メモリ装置１０３は、プログラムの起動指示があった場合に、補助記憶装置１０２からプログラムを読み出して格納する。プロセッサ１０４は、ＣＰＵ若しくはＧＰＵ（Graphics Processing Unit）、又はＣＰＵ及びＧＰＵであり、メモリ装置１０３に格納されたプログラムに従って推定装置１０に係る機能を実行する。インタフェース装置１０５は、ネットワークに接続するためのインタフェースとして用いられる。 The memory device 103 reads and stores the program from the auxiliary storage device 102 when a program activation instruction is received. The processor 104 is a CPU or a GPU (Graphics Processing Unit), or a CPU and a GPU, and executes functions related to the estimation device 10 according to programs stored in the memory device 103 . The interface device 105 is used as an interface for connecting to a network.
　図２は、本発明の実施の形態における推定装置１０の機能構成例を示す図である。図２において、推定装置１０は、学習データ収集部１１、モデル構築部１２、基地局トラヒックデータ取得部１３及びユーザ集合特性推定部１４を有する。これら各部は、推定装置１０にインストールされた１以上のプログラムが、プロセッサ１０４に実行させる処理により実現される。 FIG. 2 is a diagram showing a functional configuration example of the estimation device 10 according to the embodiment of the present invention. In FIG. 2, the estimation device 10 has a learning data collection unit 11, a model construction unit 12, a base station traffic data acquisition unit 13, and a user group characteristics estimation unit . Each of these units is implemented by processing that one or more programs installed in the estimation apparatus 10 cause the processor 104 to execute.
　以下、推定装置１０が実行する処理手順について説明する。図３は、回帰モデルの学習処理の処理手順の一例を説明するためのフローチャートである。 The processing procedure executed by the estimation device 10 will be described below. FIG. 3 is a flowchart for explaining an example of a processing procedure of a regression model learning process.
　ステップＳ１０１において、学習データ収集部１１は、ユーザ集合特性データ及びそれに対応する基地局トラヒックデータを、複数の基地局に関して複数のタイミングで収集する。その結果、複数組の学習データが収集される。例えば、収集されるユーザ集合特性データ及びそれに対応する基地局トラヒックデータは、ネットワークシミュレーションにより生成したユーザ集合特性データと基地局トラヒックデータであってもよいし、クラウドソーシング等で取得したユーザ集合特性データ（現在利用中のアプリケーションを示す情報）とそのユーザ集合に対応する基地局トラヒックデータであってもよい。なお、「それに対応する基地局トラヒックデータ」とは、ユーザ集合特性データの収集対象の基地局と同じ基地局に関し、当該ユーザ集合特性データの収集タイミングと同じタイミングに関して収集された基地局トラヒックデータをいう。 In step S101, the learning data collection unit 11 collects user set characteristic data and base station traffic data corresponding thereto for a plurality of base stations at a plurality of timings. As a result, multiple sets of learning data are collected. For example, the collected user set characteristic data and the corresponding base station traffic data may be user set characteristic data and base station traffic data generated by network simulation, or user set characteristic data obtained by crowdsourcing or the like. (information indicating the application currently in use) and base station traffic data corresponding to the user set. The "corresponding base station traffic data" refers to the base station traffic data collected at the same timing as the collection timing of the user set characteristics data in relation to the same base station as the target base station for collecting the user set characteristics data. say.
　ここでの基地局トラヒックデータとは、基地局内で得られる、例えば、下り（ＤＬ）／上り（ＵＬ）のデータレートのＵＥ数分布、Ｉｎａｃｔｉｖｅ　Ｔｉｍｅｒ経過時間のＵＥ数分布、下り（ＤＬ）／上り（ＵＬ）のリソースブロック使用率など、ユーザの通信利用による挙動が反映されている統計的なネットワークトラヒックデータをいう。基地局トラヒックデータは、これらのうちの一部（１以上）のパラメータを含むデータであってもよいし、全部のパラメータを含むデータであってもよい。 The base station traffic data here refers to, for example, downlink (DL)/uplink (UL) data rate UE number distribution, Inactive Timer elapsed time UE number distribution, downlink (DL)/uplink Statistical network traffic data that reflects the behavior of users' communication usage, such as the resource block usage rate of (UL). The base station traffic data may be data containing some (one or more) of these parameters, or may be data containing all of these parameters.
　これらのデータは、例えば、３ＧＰＰの標準化資料３ＧＰＰ　ＴＳ　３２．４２５（https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=2011）にて定められているＥ－ＵＴＲＡＮのＰｅｒｆｏｒｍａｎｃｅ　ｍｅａｓｕｒｅｍｅｎｔｓ項目に基づいて基地局が設計されていれば、自動的に取得可能である。なお、ＤＬ／ＵＬデータレートのＵＥ数分布は"DL scheduled IP throughput distribution"，"UL scheduled IP throughput distribution"にそれぞれ対応し、Ｉｎａｃｔｉｖｅ　Ｔｉｍｅｒ経過時間のＵＥ数分布は、"Number of successful RRC connection setups in relation to the time between successful RRC connection setup and lastRRCconnection release"に対応する。なお、データレートのＵＥ数分布の一例を図４に示し、Ｉｎａｃｔｉｖｅ　Ｔｉｍｅｒ経過時間のＵＥ数分布の一例を図５に示す。ＤＬ／ＵＬのリソースブロック使用率は、"DL Total PRB Usage"，"UL Total PRB Usage"にそれぞれ対応し、０～１００［％］の、値である。 These data are, for example, E-UTRAN Performance specified in 3GPP standardization document 3GPP TS 32.425 (https://portal.3gpp.org/desktopmodules/Specifications/SpecificationDetails.aspx?specificationId=2011) If the base station is designed based on the items of measurements, it can be obtained automatically. The DL/UL data rate UE number distribution corresponds to "DL scheduled IP throughput distribution" and "UL scheduled IP throughput distribution" respectively, and the UE number distribution of the Inactive Timer elapsed time corresponds to "Number of successful RRC connection setups in relation to the time between successful RRC connection setup and last RRCconnection release". An example of the UE number distribution of the data rate is shown in FIG. 4, and an example of the UE number distribution of the Inactive Timer elapsed time is shown in FIG. The DL/UL resource block usage rate corresponds to "DL Total PRB Usage" and "UL Total PRB Usage" and is a value of 0 to 100 [%].
　また、基地局トラヒックデータについては、基地局を製造するベンダ依存にはなるところもあるが、基本的に基地局側にデータ生成周期があり、その周期やタイミングにおいてデータが生成される。例えば、データ生成周期が１分周期であれば、データレートやリソースブロック使用率はその１分間の平均値が収集され、Ｉｎａｃｔｉｖｅ　Ｔｉｍｅｒ経過時間は収集のタイミングにおける瞬間的な値が収集される。 In addition, base station traffic data depends on the vendor that manufactures the base station, but basically the base station has a data generation cycle, and data is generated at that cycle and timing. For example, if the data generation cycle is one minute, the data rate and resource block usage rate are collected as average values for that one minute, and the Inactive Timer elapsed time is collected as instantaneous values at the collection timing.
　ユーザ集合特性データの一例は、対象基地局配下におけるアプリケーション種別ごとの利用人数比（ＵＥ数比）である。但し、これは一例であって、ユーザ集合特性データとしては、基地局配下のユーザ集合特性データを表す情報であれば任意の情報を用いることが可能である。例えば、ユーザ集合特性データとして、端末種別ごとの利用人数比が用いられてもよいし、アプリケーション種別と端末種別の組み合わせ毎の利用人数比が用いられてもよい。なお、いずれの場合であっても、人数比は、例えば、上記において説明した基地局トラヒックデータの生成周期のうち、ユーザ集合特性データに対応する基地局トラヒックデータの収集期間に対応する期間における値である。例えば、基地局トラヒックデータの生成周期が１分間であれば、ユーザ集合特性データに対応する基地局トラヒックデータの収集が行われる１分間におけるアプリケーション種別ごとの利用人数比がユーザ集合特性データとなる。 An example of user set characteristic data is the ratio of the number of users (UE ratio) for each application type under the target base station. However, this is only an example, and any information can be used as the user set characteristic data as long as it represents the user set characteristic data under the control of the base station. For example, the ratio of the number of users for each terminal type may be used as the user set characteristic data, or the ratio of the number of users for each combination of the application type and the terminal type may be used. In any case, the population ratio is, for example, the value in the period corresponding to the collection period of the base station traffic data corresponding to the user set characteristics data, out of the generation period of the base station traffic data described above. is. For example, if the generation cycle of base station traffic data is one minute, the ratio of the number of users for each application type in one minute during which the base station traffic data corresponding to the user set characteristic data is collected becomes the user set characteristic data.
　なお、本願発明者は、動画アプリの利用が多いエリアではユーザあたりのＤＬデータ量が多くなり、Ｗｅｂ利用が多いエリアではユーザあたりのＤＬデータ量が少なくなり、Ｗｅｂ会議アプリの利用が多いエリアではユーザあたりのＤＬ／ＵＬ共にデータ量が多くなり、音声利用が多いエリアではユーザあたりのＤＬ／ＵＬ共にデータ量が少なくなるといったように、アプリ利用の傾向が、データ量に関連する基地局のＫＰＩデータ項目に現れると考えた。そのため、本願発明者は、ＤＬ／ＵＬのデータ量を示すものとしてＤＬ／ＵＬの平均データレートのＵＥ数分布が利用可能と考えた。この値は、サブフレーム、すなわちスケジューリングの最小時間単位あたりに流れたデータ量を表す。また、本願発明者は、もう１つの仮説として、動画利用が多いエリアではチャンクの周期毎に接続が発生し、音声利用やＷｅｂ会議利用が多いエリアでは頻繁に接続が発生し、Ｗｅｂ利用が多いエリアではまばらなタイミングで接続が発生する等、アプリ利用の傾向は接続タイミングに関連する基地局のＫＰＩデータ項目に現れると考えた。そのため、本願発明者は、各ユーザの最後の接続からの経過時間を表すＩｎａｃｔｉｖｅ　Ｔｉｍｅｒ経過時間のＵＥ数分布の項目を抽出した。また、本願発明者は、動画利用とＷｅｂ会議利用が多いエリアではデータ量が多いためリソース消費が多くなり、音声利用やＷｅｂ利用が多いエリアではデータ量が少ないためリソース消費も少なくなることから、アプリ利用の傾向はリソース消費に関連する基地局のＫＰＩデータ項目に現れると考えた。そのため、本願発明者は、基地局のリソース消費の割合を表すリソースブロック使用率の項目を抽出した。 In addition, the inventor of the present application believes that the amount of DL data per user increases in areas where video applications are frequently used, the amount of DL data per user decreases in areas where web use is high, and in areas where web conference applications are frequently used The DL/UL data volume per user increases, and in areas where voice usage is high, the DL/UL data volume per user decreases. thought to appear in data items. Therefore, the inventor of the present application considered that the UE number distribution of the DL/UL average data rate can be used as an indication of the DL/UL data amount. This value represents the amount of data that has flowed per subframe, ie the smallest scheduling time unit. In addition, the inventor of the present application has another hypothesis that in areas where video is frequently used, connections occur every chunk period, and in areas where audio and web conferences are often used, connections occur frequently, and web use is frequent. We thought that the trend of application usage would appear in the KPI data items of the base station related to connection timing, such as connections occurring at sparse timing in the area. Therefore, the inventor of the present application extracted the item of the UE number distribution of the Inactive Timer elapsed time, which represents the elapsed time from the last connection of each user. In addition, the inventor of the present application believes that resource consumption is high in areas where there is a large amount of data used for video and web conferencing, and resource consumption is low in areas where there is a large amount of data used for voice and web. We thought that app usage trends would appear in the KPI data items of base stations related to resource consumption. Therefore, the inventor of the present application extracted the item of resource block usage rate, which represents the rate of resource consumption of the base station.
　続いて、モデル構築部１２は、学習データ収集部１１で収集した、ユーザ集合特性データ及びそれに対応する基地局トラヒックデータの複数の組を学習データとして用いて、基地局トラヒックデータを入力とし、ユーザ集合特性データを出力とした回帰モデルを学習する（Ｓ１０２）。ここで用いる回帰の手法は、線形回帰やランダムフォレスト回帰など、回帰であれば任意の手法を用いることができる。 Subsequently, the model construction unit 12 uses, as learning data, a plurality of sets of the user set characteristic data and the corresponding base station traffic data collected by the learning data collection unit 11, and receives the base station traffic data as an input. A regression model outputting the set characteristic data is learned (S102). Any regression method such as linear regression or random forest regression can be used as the regression method used here.
　続いて、モデル構築部１２は、学習済みの回帰モデルを出力（例えば、当該回帰モデルの学習パラメータの値を出力）する（Ｓ１０３）。 Subsequently, the model construction unit 12 outputs the learned regression model (for example, outputs the values of the learning parameters of the regression model) (S103).
　図６は、ユーザ集合特性データの推定処理の処理手順の一例を説明するためのフローチャートである。 FIG. 6 is a flowchart for explaining an example of the processing procedure for estimating user group characteristic data.
　基地局トラヒックデータ取得部１３は、例えば、推定装置１０のユーザによる入力に応じ、ユーザ集合特性データを明らかにしたい期間及びネットワークの基地局トラヒックデータを取得する（Ｓ２０１）。当該期間及び当該ネットワークは、例えば、推定装置１０のユーザによって指定される。当該ネットワークは、例えば、基地局単位、セクタ単位、又はセクタ×キャリア（周波数帯）単位等で区別されるネットワーク（又はエリア）のうちのいずれかのネットワーク（又はエリア）である。なお、当該基地局トラヒックデータは、学習データ収集部１１が収集するパラメータと同じパラメータで構成される。 For example, the base station traffic data acquisition unit 13 acquires base station traffic data for a period for which user group characteristic data is to be clarified and network base station traffic data in response to an input by the user of the estimation device 10 (S201). The period and the network are specified by the user of the estimation device 10, for example. The network is, for example, any network (or area) among networks (or areas) distinguished by base station unit, sector unit, sector×carrier (frequency band) unit, or the like. The base station traffic data is composed of the same parameters as the parameters collected by the learning data collection unit 11 .
　続いて、ユーザ集合特性推定部１４は、基地局トラヒックデータ取得部１３が取得した基地局トラヒックデータを学習済みの回帰モデルに入力することで、該当ネットワークの期間Ｔにおけるユーザ集合特性データを推定する（Ｓ２０２）。すなわち、ユーザ集合特性推定部１４は、当該回帰モデルが出力するユーザ集合特性データを、該当ネットワークの期間Ｔは、におけるユーザ集合特性データの推定値として取得する。 Subsequently, the user aggregate characteristics estimation unit 14 inputs the base station traffic data acquired by the base station traffic data acquisition unit 13 into the learned regression model, thereby estimating the user aggregate characteristics data for the period T of the network. (S202). That is, the user set characteristic estimation unit 14 acquires the user set characteristic data output by the regression model as an estimated value of the user set characteristic data for the period T of the network.
　続いて、ユーザ集合特性推定部１４は、当該ユーザ集合特性データを出力する（Ｓ２０３）。 Subsequently, the user group characteristic estimation unit 14 outputs the user group characteristic data (S203).
　上述したように、本実施の形態によれば、既にネットワーク管理で取得され幅広く利用されている統計的な基地局トラヒックデータから、ユーザの通信利用状況を反映されたデータ項目を利用し、機械学習の枠組みに基づき、基地局に対応するユーザ集合特性を推定可能とすることができる。これにより、プライバシーやコストの問題に対応しつつ、ユーザ集合特性データを得ることが可能となる。 As described above, according to the present embodiment, from the statistical base station traffic data already acquired in network management and widely used, data items reflecting the user's communication usage status are used to perform machine learning. framework, it may be possible to estimate user population characteristics corresponding to a base station. This makes it possible to obtain user set characteristic data while addressing privacy and cost issues.
　なお、本実施の形態において、基地局トラヒックデータ取得部１３は、取得部の一例である。ユーザ集合特性推定部１４は、推定部の一例である。モデル構築部１２は、学習部の一例である。 In addition, in the present embodiment, the base station traffic data acquisition unit 13 is an example of an acquisition unit. The user group characteristics estimating unit 14 is an example of an estimating unit. The model construction unit 12 is an example of a learning unit.
　以上、本発明の実施の形態について詳述したが、本発明は斯かる特定の実施形態に限定されるものではなく、請求の範囲に記載された本発明の要旨の範囲内において、種々の変形・変更が可能である。 Although the embodiments of the present invention have been described in detail above, the present invention is not limited to such specific embodiments, and various modifications can be made within the scope of the gist of the present invention described in the claims.・Changes are possible.
１０　　　　　推定装置
１１　　　　　学習データ収集部
１２　　　　　モデル構築部
１３　　　　　基地局トラヒックデータ取得部
１４　　　　　ユーザ集合特性推定部
１００　　　　ドライブ装置
１０１　　　　記録媒体
１０２　　　　補助記憶装置
１０３　　　　メモリ装置
１０４　　　　プロセッサ
１０５　　　　インタフェース装置
Ｂ　　　　　　バス 10 estimation device 11 learning data collection unit 12 model construction unit 13 base station traffic data acquisition unit 14 user group characteristic estimation unit 100 drive device 101 recording medium 102 auxiliary storage device 103 memory device 104 processor 105 interface device B bus

Claims (6)

  1. 　或る基地局に関する或る期間における基地局トラヒックデータを取得するように構成されている取得部と、
    　学習データとして収集された基地局トラヒックデータ及びユーザ集合特性データの複数の組に基づいて学習された回帰モデルに前記取得部が取得した基地局トラヒックデータを入力することで、前記或る基地局に関する前記或る期間におけるユーザ集合特性データを推定するように構成されている推定部と、
    を有することを特徴とする推定装置。     an acquisition unit configured to acquire base station traffic data for a base station over a period of time;
    By inputting the base station traffic data acquired by the acquisition unit into a regression model learned based on a plurality of sets of base station traffic data and user set characteristic data collected as learning data, an estimating unit configured to estimate user set characteristic data over the period of time;
    An estimation device characterized by comprising:
  2. 　前記学習データとして収集された基地局トラヒックデータ及びユーザ集合特性データの複数の組に基づいて前記回帰モデルを学習するように構成されている学習部、
    を有することを特徴とする請求項１記載の推定装置。     a learning unit configured to learn the regression model based on multiple sets of base station traffic data and user aggregate characteristic data collected as the learning data;
    The estimating device according to claim 1, characterized by comprising:
  3. 　前記基地局トラヒックデータは、下り／上りのデータレートのＵＥ数分布、Ｉｎａｃｔｉｖｅ　Ｔｉｍｅｒ経過時間のＵＥ数分布、及び下り／上りのリソースブロック使用率のうちの１以上を含む、
    ことを特徴とする請求項１又は２記載の推定装置。     The base station traffic data includes one or more of downlink/uplink data rate UE number distribution, Inactive Timer elapsed time UE number distribution, and downlink/uplink resource block usage rate,
    3. The estimating device according to claim 1 or 2, characterized in that:
  4. 　前記ユーザ集合特性データは、アプリケーション種別ごとの利用人数比若しくは端末種別ごとの利用人数比、又はアプリケーション種別ごとの利用人数比及び端末種別ごとの利用人数比を含む、
    ことを特徴とする請求項１乃至３いずれか一項記載の推定装置。     The user set characteristic data includes the ratio of the number of users for each application type or the ratio of the number of users for each terminal type, or the ratio of the number of users for each application type and the ratio of the number of users for each terminal type,
    4. The estimation device according to any one of claims 1 to 3, characterized in that:
  5. 　或る基地局に関する或る期間における基地局トラヒックデータを取得する取得手順と、
    　学習データとして収集された基地局トラヒックデータ及びユーザ集合特性データの複数の組に基づいて学習された回帰モデルに前記取得手順が取得した基地局トラヒックデータを入力することで、前記或る基地局に関する前記或る期間におけるユーザ集合特性データを推定する推定手順と、
    をコンピュータが実行することを特徴とする推定方法。     an acquisition procedure for acquiring base station traffic data for a base station over a period of time;
    By inputting the base station traffic data acquired by the acquisition procedure into a regression model trained based on a plurality of sets of base station traffic data and user aggregate characteristic data collected as learning data, an estimation procedure for estimating user set characteristic data over the period of time;
    A method of estimation characterized in that the computer executes the
  6. 　或る基地局に関する或る期間における基地局トラヒックデータを取得する取得手順と、
    　学習データとして収集された基地局トラヒックデータ及びユーザ集合特性データの複数の組に基づいて学習された回帰モデルに前記取得手順が取得した基地局トラヒックデータを入力することで、前記或る基地局に関する前記或る期間におけるユーザ集合特性データを推定する推定手順と、
    をコンピュータに実行させることを特徴とするプログラム。     an acquisition procedure for acquiring base station traffic data for a base station over a period of time;
    By inputting the base station traffic data acquired by the acquisition procedure into a regression model trained based on a plurality of sets of base station traffic data and user aggregate characteristic data collected as learning data, an estimation procedure for estimating user set characteristic data over the period of time;
    A program characterized by causing a computer to execute
PCT/JP2022/001057 2022-01-14 2022-01-14 Inference device, inference method, and program WO2023135729A1 (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
PCT/JP2022/001057 WO2023135729A1 (en) 2022-01-14 2022-01-14 Inference device, inference method, and program

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
PCT/JP2022/001057 WO2023135729A1 (en) 2022-01-14 2022-01-14 Inference device, inference method, and program

Publications (1)

Publication Number Publication Date
WO2023135729A1 true WO2023135729A1 (en) 2023-07-20

Family

ID=87278730

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/JP2022/001057 WO2023135729A1 (en) 2022-01-14 2022-01-14 Inference device, inference method, and program

Country Status (1)

Country Link
WO (1) WO2023135729A1 (en)

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2013178739A (en) * 2012-02-07 2013-09-09 Kddi Corp Terminal information estimation device for estimating software type information of terminal, dns server, program, and method
WO2019187296A1 (en) * 2018-03-29 2019-10-03 日本電気株式会社 Communication traffic analysis device, communication traffic analysis method, program, and recording medium

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2013178739A (en) * 2012-02-07 2013-09-09 Kddi Corp Terminal information estimation device for estimating software type information of terminal, dns server, program, and method
WO2019187296A1 (en) * 2018-03-29 2019-10-03 日本電気株式会社 Communication traffic analysis device, communication traffic analysis method, program, and recording medium

Similar Documents

Publication Publication Date Title
US10674388B2 (en) Wireless communication data analysis and reporting
US9578441B2 (en) Intelligent mobility application profiling tool
US10326640B2 (en) Knowledge base radio and core network prescriptive root cause analysis
US20090111382A1 (en) Methods for scheduling collection of key performance indicators from elements in a communications network
WO2016074446A1 (en) Method, apparatus, and system for energy conservation of frequency spectrum resource self-optimization
CN104126321A (en) Adaptation of mobility parameters based on user equipment measurement availability
WO2019101193A1 (en) Method and apparatus for predicting a quality of experience of a service in a wireless network
CN105052076B (en) Network element management system and network element management method based on cloud computing
CN110493789B (en) Network capacity expansion method and device
CN107371179B (en) Measurement result reporting method, measurement result receiving method, related equipment and system
WO2015184888A1 (en) Terminal, method, and system for switching networks
EP2673974B1 (en) System and method for determining effects of non-network elements on network resource utilization in a mobile network
WO2023135729A1 (en) Inference device, inference method, and program
WO2023045931A1 (en) Network performance abnormality analysis method and apparatus, and readable storage medium
US10721707B2 (en) Characterization of a geographical location in a wireless network
EP3860200A1 (en) Data transmission method and device
US11805043B2 (en) Method and system for real-time encrypted video quality analysis
EP2673973B1 (en) System and method for determining effects of non-network elements on the subscriber experience in a mobile network
Ickin et al. Qoe-based energy reduction by controlling the 3g cellular data traffic on the smartphone
WO2016070606A1 (en) Prediction method and device for system capacity
WO2019145171A1 (en) Method, apparatus and computer program for modification of admission control criteria
WO2021197463A1 (en) Energy saving indication method and device, and computer storage medium
CN114980148B (en) Network capability determining method and device
US20240147380A1 (en) Method for obtaining computing power information and related device
WO2023241320A1 (en) Method and device for determining indicators for network optimization function of autonomous system

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22920254

Country of ref document: EP

Kind code of ref document: A1