Communication control device, control method for communication control device, and control program for communication control device

Abstract

A communication control device in a communication system for controlling a radio access network including a virtualized base station using artificial intelligence according to one embodiment of the present invention comprises: a storage unit that, regarding a plurality of different models using artificial intelligence and aimed at power saving of the radio access network, stores model information including at least a power consumption amount when the models are executed and the expected value of an improvement amount of the state of the radio access network by the execution of the models; an index acquisition unit that regularly acquires, from the virtualized base station, a plurality of types of indexes relating to the state of the radio access network; a selection unit that selects, on the basis of the indexes, a model to be executed from among the plurality of different models; and a determination unit that determines, on the basis of the model information, whether or not to execute the model selected by the selection unit.

Description

Communication control device, control method for communication control device, and control program for communication control device

[0001] The present invention relates to a communication control device, a control method for a communication control device, and a control program for a communication control device.

[0002] In recent years, development has been progressing on communication systems that control radio access networks (RANs), including virtualized base stations, using artificial intelligence (AI). For example, in 5G-Advanced, starting with 3GPP® (Third Generation Partnership Project) Release 18, efforts are being made to realize optimal network-wide orchestration using AI and machine learning (ML), as well as efforts to improve RAN performance using AI / ML. Furthermore, AI / ML is being introduced into existing networks to improve the efficiency of RAN operation and automate parameter setting (for example, Non-Patent Document 1).

[0003] "The Evolution of RAN through AI ~ AI for RAN ~", [online], February 26, 2024, SoftBank Corp., [Retrieved November 5, 2024], Internet <URL: https: / / www.softbank.jp / corp / technology / research / story-event / 040 / >

[0004] A communication control device according to an embodiment of the present invention is a communication control device in a communication system that controls a radio access network (RAN) including a virtual base station using artificial intelligence (AI). The communication control device includes a storage unit that stores model information including at least the power consumption when a model is executed and the expected value of the improvement amount of the state of the radio access network when the model is executed for a plurality of different models using artificial intelligence and aiming at power saving of the radio access network; an index acquisition unit that periodically acquires a plurality of types of indexes related to the state of the radio access network from the virtual base station; a selection unit that selects a model to be executed from among the plurality of different models based on the indexes; and a determination unit that determines whether to execute the model selected by the selection unit based on the model information.

[0005] In the communication control device according to an embodiment of the present invention, the storage unit may store an evaluation index based on the result when the model is executed under a predetermined condition as the expected value of the improvement amount related to the model.

[0006] The communication control device according to an embodiment of the present invention further includes an evaluation measurement unit that acquires an output when a predetermined data set is input to the model, and the storage unit may store an evaluation index based on the output as the expected value of the improvement amount related to the model.

[0007] In the communication control device according to an embodiment of the present invention, the storage unit stores, as model information, the relationship between the respective power consumptions when a plurality of models are executed and the improvement amount of the state of the radio access network for a plurality of models that improve the state of the radio access network indicated by one of the plurality of types of indexes, and the determination unit may determine a model to be executed from among the plurality of models according to the state (load) of the radio access network indicated by one of the indexes.

[0008] A control method for a communication control device in a communication system that controls a Radio Access Network (RAN) including a virtualized base station using artificial intelligence (AI), according to one embodiment of the present invention, involves the communication control device storing model information for a plurality of different models using artificial intelligence, which includes at least the power consumption when the model is executed and the expected amount of improvement in the state of the radio access network as a result of executing the model; periodically obtaining a plurality of indicators regarding the state of the radio access network from the virtualized base station; selecting a model to execute from among the plurality of different models based on the indicators; and deciding whether or not to execute the model selected in the selection step based on the model information.

[0009] A control program for a communication control device in a communication system that controls a Radio Access Network (RAN) including a virtualized base station using artificial intelligence (AI), according to one embodiment of the present invention, provides the communication control device with the following functions: a function to store model information for multiple different models using artificial intelligence, which includes at least the power consumption when the model is executed and the expected value of the improvement in the state of the radio access network as a result of executing the model; a function to periodically acquire multiple types of indicators regarding the state of the radio access network from the virtualized base station; a function to select a model to execute from among multiple different models based on the indicators; and a function to determine whether or not to execute the model selected by the selection function based on the model information.

[0010] Figure 1 is an example of a system configuration including a wireless access network according to one embodiment of the present invention. Figure 2 is an example of a model information table according to one embodiment of the present invention. Figure 3 is an example of a sequence diagram between a communication control device, a model, and a distributed station DU according to one embodiment of the present invention. Figure 4 is an example of a block configuration of a communication control device according to one embodiment of the present invention. Figure 5 is a flowchart of an example of a control method for a communication control device according to one embodiment of the present invention.

[0011] Hereinafter, an embodiment of the invention described herein (also referred to as the present invention) will be explained using the figures. Note that the figures are examples only, and the present invention is not limited to what is shown in the figures. For example, the number of communication control devices, RUs (Radio Units), DUs (Distributed Units), CUs (Central Units), data tables (data sets), flowcharts, and sequence diagrams shown are examples only, and the present invention is not limited to these.

[0012] As mentioned above, in recent years, there has been progress in the development of communication systems that control radio access networks (RANs), including virtualized base stations, using artificial intelligence (AI). The applicant defines the entire range of functions that aim to maximize the performance of RANs by applying AI and machine learning (ML) as "AI-for-RAN" and is conducting research and development accordingly.

[0013] For example, AI-based control of RANs (Range Ranging Networks) aims to reduce power consumption. This involves reducing the transmission power of the RU (Radio Unit) or reducing power consumption by changing the performance and operating settings of the processor in the CU (Controller Unit) or DU (Digital Unit). However, when implementing faster and more accurate power reduction using AI, there is a problem in that power consumption occurs due to the operation of the processor (GPU / CPU). In contrast, according to one embodiment of the present invention, when optimizing (reducing power consumption) of the RAN using AI, the optimization may also take into account the power consumption of operating the AI ​​itself. This makes it possible to reduce unnecessary power consumption caused by AI processing and achieve accurate power saving.

[0014] <System Configuration> Figure 1 is a schematic diagram showing an example configuration of a radio access network (RAN) according to one embodiment of the present invention. The RAN 400 is a RAN to which the specifications defined by the industry group O-RAN Alliance (Open Radio Access Network Alliance) are applied, and may include base stations consisting of a radio station RU 10 equipped with an antenna (not shown), a distributed station DU 20, and a centralized station CU 30. The interfaces between each RU 10, DU 20, and CU 30 may be standardized and open according to the specifications defined by the O-RAN Alliance. Furthermore, the RAN 400 may be implemented by vRAN (virtual RAN), which realizes its functions by software on general-purpose hardware. Here, for the sake of clarity, three RU 10s and one each of DU 20 and CU 30 are shown in Figure 1, but there may be multiple of these. In this configuration, each DU 20 and each CU 30 may be connected in a mesh configuration via midhaul, and each DU 20 and each RU 10 may be connected in a mesh configuration via fronthaul. Furthermore, each CU 30 may be connected to the core network (CN) via backhaul.

[0015] The communication control device 100 may be an information processing device that optimizes the power consumption of the RAN 400, taking into account the power consumption of operating the AI ​​itself. In one embodiment of the present invention, the communication control device 100 may be an SMO (Service Management and Orchestration) on which a RIC (RAN Intelligent Controller) is implemented.

[0016] Although only one communication control device 100 is shown in Figure 1, it is not limited to this. In other words, each function described as being provided by the communication control device 100 may be implemented by multiple servers. Furthermore, the communication control device 100 may be, for example, a distributed server system that operates cooperatively by communicating over a network, or a so-called cloud server. That is, the communication control device 100 may include not only physical servers but also virtual servers created by software.

[0017] RAN 400 may include multiple models (Model A 40A, Model B 40B, Model C 40C) aimed at reducing the power consumption of RAN 400. The number of models is not limited to those shown in the figure, and unless there is a need to distinguish them, they are collectively referred to as "Model 40". Model 40 may be a mathematical function or program constructed for the purpose of solving a predetermined problem using artificial intelligence or machine learning. In one embodiment of the present invention, Model 40 may be aimed at reducing the power consumption of RAN 400. For example, Model 40 may aim to predict traffic fluctuations according to geographical conditions or time of day and appropriately allocate base station resources, dynamically adjust the power required by each base station, predict interference and select the optimal frequency band or channel, thereby reducing power consumption. Models related to power saving are not limited to these. Also, in Figure 1, for clarity, multiple Model 40 are shown as independent blocks, but multiple Model 40 may be stored in a predetermined database not shown.

[0018] The model information database 300 may store model information for multiple models 40, including at least the power consumption when model 40 is executed and the expected amount of improvement in the state of the wireless access network due to the execution of model 40, i.e., the implementation of the improvement measures by model 40. These may be measured and stored in advance, and Figure 2 shows an example of model information stored in the model information database 300. The model information table TB10 may store each model in association with a model ID (IDentifier: a type of identification information), which is identification information that uniquely identifies each model, and the power consumption when the model is executed and the expected amount of improvement. Note that the figure is just an example, and the information stored is not limited to this. Also, although the model information database 300 is shown separately from the RAN 400 in Figure 1, the model information database 300 may be included in the RAN 400.

[0019] The power consumption when running Model 40 may be monitored using predetermined commands or tools related to GPU management, or tools that monitor the power consumption of specific hardware such as server racks. Alternatively, predetermined software for estimating processor power consumption may be used.

[0020] Furthermore, the "expected value of improvement" may refer to a predicted value of the degree of improvement in the state or performance of the target system as a result of running the model. For example, the expected value of improvement may be an evaluation index based on the results when the model is run under predetermined conditions. In other words, the expected value of improvement may be a benchmark measured when the model is run against defined test cases. The benchmark may be, but is not limited to, the ratio of reduced power consumption to the power consumption before the model was run.

[0021] In one embodiment of the present invention, whether or not to implement power saving for the RAN 400 may be determined based on indicators related to the state of the RAN 400 obtained from the base station. There may be multiple types of indicators related to the state of the RAN, such as KPI (Key Performance Indicators), Sounding Reference Signal (SRS), Channel State Information Reference Signal (CSI-RS), etc., but are not limited to these. The power consumption of the RAN 400 may be estimated from these indicators, for example, from indicators related to communication load and resource usage, such as traffic volume, resource utilization rate, handover success rate, signal-to-noise ratio (SINR), channel quality indicator (CQI), and resource block utilization rate.

[0022] In one embodiment of the present invention, the communication control device 100 may select a model to execute from among a plurality of different models based on the above indicators. For example, the model information table TB10 may associate a model ID with information about the indicators to be executed by the model identified by the model ID. The communication control device 100 may select a model associated with an indicator from the acquired indicators for which a reduction in power consumption is determined to be necessary, and determine whether or not to execute the model based on the power consumption when the model is executed and the expected value of the improvement. That is, the communication control device 100 predicts how much the power consumption estimated based on the indicators will improve when the model is executed from the expected value of the improvement, and if the predicted improvement exceeds the power consumption when the model is executed, the model may be executed. Conversely, if the predicted improvement is less than the power consumption when the model is executed, the model does not need to be executed.

[0023] Thus, according to one embodiment of the present invention, the amount of power consumed when the model is executed and the expected amount of improvement when the model is executed are taken into consideration when determining whether or not to use an AI model for RAN power saving. Therefore, it is possible to provide a communication system in which unnecessary power is not consumed by the execution of the model.

[0024] Furthermore, the expected value of the improvement amount is not limited to being stored in advance, but may also be measured when model 40 is deployed. That is, at the time of deployment, the output result of running model 40 with a dataset (a predetermined dataset) for test cases transmitted from the communication control device 100 as input may be stored as the expected value of the improvement amount. An example of this sequence is shown in Figure 3. First, when deploying multiple models 40A to 40C, a usage request (subscription) may be made to the communication control device 100 (step T1). In response, the communication control device 100 may send a dataset corresponding to the test cases to each model and request measurement (step T2). Each model 40A to 40C may send the output result (measurement result) with the dataset as input to the communication control device 100 (step T3). The communication control device 100 may store the acquired measurement result as model information (step T4). Subsequently, the communication control device 100 may periodically or upon request acquire the above-mentioned indicator from DU 20 (step T5). Furthermore, the communication control device 100 may select a model and decide whether or not to execute the model based on the acquired indicators (step T6). If the model is to be executed, the DU 20 may be notified of the model to be executed (step T7). Subsequently, the DU 20 may establish communication with the model to be executed (set the execution environment) (step T8). This makes it possible to generate model information that is more in line with the execution environment.

[0025] In one embodiment of the present invention, the model information database 300 may store, as a model information table TB10, the relationship between the power consumption of each model and the degree of improvement in the state of the wireless access network, for multiple models that improve the state of the wireless access network indicated by one of several indicators. The communication control device 100 may then determine which model to execute from among the multiple models, according to the state (load) of the wireless access network indicated by one indicator. In other words, in one embodiment of the present invention, multiple models may be provided for the target of improvement, each having at least one different expected improvement amount and power consumption. For example, several types of models may be provided, such as one with low power consumption but a low expected improvement amount, one with moderate power consumption and a moderate expected improvement amount, and one with high power consumption but a high expected improvement amount, and the model to execute may be selected according to the state of the target of improvement. This makes it possible to further reduce unnecessary power consumption.

[0026] Next, the configuration and functions of the communication control device will be explained using Figure 4.

[0027] <Hardware configuration of the communication control device> The communication control device 100 may include a control unit 110, a communication unit 120, an input / output unit 130, and a storage unit 170.

[0028] The control unit 110 is typically a processor, and may include a central processing unit (CPU), a microprocessing unit (GPU), a graphics processing unit (GPU), a microprocessor, etc., and may be implemented by logic circuits (hardware) or dedicated circuits formed on an integrated circuit (IC (Integrated Circuit) chip, LSI (Large Scale Integration)), etc.

[0029] The communication unit 120 may be implemented as hardware such as a NIC (Network Interface Card), a network adapter, communication software, or a combination thereof. The communication unit 120 may send and receive various data with the DU 20, CU 30, and model information database 300 via the network 500.

[0030] The input / output unit 130 may include an input device for inputting various operations to the communication control device 100, and an output device for outputting processing results processed by the communication control device 100. The input device may include, for example, hardware keys such as a touch panel, touch display, or keyboard, a pointing device such as a mouse, a camera, or a microphone. The output device may output processing results processed by the control unit 110. The output device may include, for example, a display, touch panel, or speaker.

[0031] The storage unit 170 stores various programs and data necessary for the operation of the communication control device 100. The storage unit 170 may include, for example, an HDD (Hard Disk Drive), an SSD (Solid State Drive), flash memory, etc. The storage unit 170 may also include memory that provides a working area for the control unit 110.

[0032] <Functional Configuration of Communication Control Device> The communication control device 100 may include, as functions realized by the control unit 110, an index acquisition unit 111, a selection unit 112, a determination unit 113, an evaluation and measurement unit 114, and a judgment unit 115.

[0033] The indicator acquisition unit 111 may periodically acquire multiple types of indicators related to the status of the wireless access network from the virtualized base station, which consists of RU 10, DU 20, and CU 30. These multiple types of indicators may be KPIs, etc., as described above.

[0034] The selection unit 112 may select a model to execute from among several different models based on the indicators acquired by the indicator acquisition unit 111. That is, depending on the indicators, it may select the necessary model (for example, appropriate resource allocation, antenna beamforming, adjustment of power usage for each base station, etc.).

[0035] The decision unit 113 may decide whether or not to execute the model selected by the selection unit 112, based on the model information. That is, if the power consumption when executing the selected model is greater than the expected degree of improvement of the target to be improved (redundant), the model may not be executed.

[0036] The evaluation measurement unit 114 may acquire the output when a predetermined dataset is input to the model. The determination unit 115 may perform various determination processes.

[0037] <Control Flowchart of the Communication Control Device> The control method of the communication control device 100 described above will be explained using the flowchart in Figure 5. First, the model information database 300 stores model information for multiple different models using artificial intelligence, each model aimed at saving power in the wireless access network, including at least the power consumption when the model is executed and the expected value of the improvement in the state of the wireless access network due to the execution of the model (Step S11). Next, the indicator acquisition unit 111 of the communication control device 100 may periodically acquire multiple types of indicators related to the state of the wireless access network from the virtualized base station (Step S12). The selection unit 112 may select a model to execute from among multiple different models based on the indicators (Step S13). The decision unit 113 may decide whether or not to execute the selected model based on the model information (Step S14).

[0038] The present invention has been described based on various drawings and embodiments, but it should be noted that those skilled in the art will find it easy to make various modifications and alterations based on this disclosure. Therefore, it should be noted that these modifications and alterations are within the scope of the present invention. For example, the functions included in each component, step, etc., can be rearranged in a logically consistent manner, and multiple components or steps, etc., can be combined into one or divided. Furthermore, the configurations shown in the above embodiments may be combined as appropriate.

[0039] For example, the above description focused on a model aimed at reducing power consumption in RAN. However, the present invention may be applicable not only to power saving but also to models aimed at improving the functionality and performance of RAN using AI.

[0040] The programs of each embodiment of this disclosure may be provided stored in a storage medium readable by the information processing device. The storage medium is a “non-temporary tangible medium” capable of storing programs. The programs include, for example, software programs and information processing device programs. When each functional unit of the communication control device 100 as an information processing device is implemented by software, the communication control device 100 functions as an index acquisition unit 111, a selection unit 112, a determination unit 113, an evaluation measurement unit 114, and a judgment unit 115 by having the processor execute a program loaded into memory.

[0041] The storage medium may, where appropriate, include one or more semiconductor-based or other integrated circuits (ICs) (e.g., field-programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), hard disk drives (HDDs), hybrid hard drives (HHDs), optical disks, optical disk drives (ODDs), magneto-optical disks, magneto-optical drives, floppy diskettes, floppy disk drives (FDDs), magnetic tapes, solid-state drives (SSDs), RAM drives, secure digital cards or drives, any other suitable storage medium, or two or more suitable combinations thereof. The storage medium may, where appropriate, be volatile, non-volatile, or a combination of volatile and non-volatile.

[0042] Further, the program of the present disclosure may be provided to the communication control device 100 via any transmission medium (such as a communication network or a broadcast wave) capable of transmitting the program.

[0043] Further, each embodiment of the present disclosure can also be realized in the form of a data signal embedded in a carrier wave, in which the program is embodied by electronic transmission. Note that the program of the present disclosure may be implemented using, for example, script languages such as JavaScript (registered trademark), Python (registered trademark), C language, Go language, Swift (registered trademark), Kotlin (registered trademark), Java (registered trademark), etc.

[0044] According to each aspect of the present disclosure described above, it is possible to further promote power saving in the wireless access network, thus contributing to the achievement of Sustainable Development Goal (SDG) 9, "Build the infrastructure for industry and innovation."

[0045] 10 Wireless Unit (RU) 20 Distributed Unit (DU) 30 Centralized Unit (CU) 40 (40A, 40B, 40C) Model 100 Communication Control Device (Server) 110 Control Unit 111 Index Acquisition Unit 112 Selection Unit 113 Decision Unit 114 Evaluation Measurement Unit 115 Judgment Unit 120 Communication Unit 130 Input / Output Unit 170 Storage Unit 300 Model Information Database 400 Wireless Access Network (RAN) CN Core Network

Claims

1. A communication control device for a communication system that controls a radio access network (RAN) including a virtualized base station using artificial intelligence (AI), comprising: a storage unit that stores model information including at least the power consumption when a model is executed and the expected amount of improvement in the state of the radio access network as a result of executing the model, for a plurality of different models using the artificial intelligence, the model aimed at saving power of the radio access network; an indicator acquisition unit that periodically acquires a plurality of indicators relating to the state of the radio access network from the virtualized base station; a selection unit that selects a model to be executed from the plurality of different models based on the indicators; and a decision unit that determines whether or not to execute the model selected by the selection unit based on the model information.

2. The communication control device according to claim 1, wherein the storage unit stores an evaluation index based on the results obtained when the model is executed under predetermined conditions, as an expected value of the amount of improvement with respect to the model.

3. The communication control device further comprises an evaluation measurement unit that acquires an output when a predetermined dataset is input to the model, and the storage unit stores an evaluation index based on the output as an expected value of the improvement amount for the model, according to claim 1.

4. The storage unit stores, as model information, the relationship between the power consumption resulting from executing each of the multiple models for improving the state of the wireless access network, indicated by one of the multiple indicators, and the amount of improvement in the state of the wireless access network; and the determination unit determines which of the multiple models to execute according to the state of the wireless access network indicated by the one indicator, the communication control device according to claim 1.

5. A method for controlling a communication control device in a communication system that controls a radio access network (RAN) including a virtualized base station using artificial intelligence (AI), the method comprising: storing model information for a plurality of different models using the artificial intelligence, which includes at least the power consumption when the model is executed and the expected value of the improvement in the state of the radio access network as a result of executing the model; periodically obtaining a plurality of indicators relating to the state of the radio access network from the virtualized base station; selecting a model to be executed from the plurality of different models based on the indicators; and determining, based on the model information, whether or not to execute the model selected in the selection step.

6. A control program for a communication control device in a communication system that controls a radio access network (RAN) including a virtualized base station using artificial intelligence (AI), the control program for a communication control device that enables the following functions: a function to store model information for a plurality of different models using the artificial intelligence, including at least the power consumption when the model is executed and the expected value of the improvement in the state of the radio access network as a result of executing the model; a function to periodically acquire a plurality of indicators relating to the state of the radio access network from the virtualized base station; a function to select a model to execute from the plurality of different models based on the indicators; and a function to determine whether or not to execute the model selected by the selection function based on the model information.

Images