Policies in network and enterprise domains

Abstract

A method performed by a node is provided A Radio Access Network (RAN) has a set of energy saving policies and provides connectivity to a plurality of User Equipments (UEs). The plurality of UEs are for performing one or more operational tasks in an enterprise domain. The method comprises obtaining, from the RAN, first information relating to the operation of the RAN. The method further comprises obtaining, from the enterprise domain, second information relating to the one or more operational tasks. Based on the first information and second information, the node selects one or more of the set of energy saving policies for use by the RAN and determines a configuration of the selected one or more energy saving policies.

Description

[0001] POLICIES IN NETWORK AND ENTERPRISE DOMAINS

[0002] Technical field

[0003] Embodiments of the present disclosure relate to communication networks, and particularly to policies in network and enterprise domains.

[0004] Mobile network operators implement Radio Access Network (RAN) energy-saving measures (also referred to herein as RAN energy-saving technologies or RAN energy-saving policies) to ensure the long-term sustainability of their networks and / or to reduce their carbon footprint. RAN energy-saving technologies can also provide lower operational costs and allow networks to operate more efficiently. Examples of RAN energy saving policies are given in a 3rd Generation Partnership Project (3GPP) Technical Report entitled “Technical Specification Group Radio Access Network; Study on network energy savings for NR (Release 18)” (TR 38.864 V18.1.0 (2023-03)), and include:

[0005] Dynamic Power Management;

[0006] Cell Activation / Deactivation;

[0007] Sleep Mode and Power Saving States;

[0008] - Traffic Control and Load Balancing;

[0009] Increasing the cell service area based on the number of users and their needs, helping to get the most out of the consumed energy; and Network Planning and Optimization

[0010] By implementing RAN energy saving policies, operators can lower energy costs, improve network efficiency, and / or enhance sustainability.

[0011] In order to analyze the problem of extreme energy consumption in RAN and provide a solution accordingly, Open Radio Access Network (O-RAN) WG1 recently initiated a work item regarding network energy saving use cases (see O-RAN Alliance document "O-RAN Network Energy Saving Use Cases Technical Report 2.0 (O-RAN. WG1. Network-Energy-Savings- Technical-Report-R003-v02.00),", 2023. [Available at This technical report discusses relevant counters and Key Performance Indicators (KPIs) for monitoring and reporting, requirements, key issues, use cases, solution deployment options, and potential impacts on interfaces. For each particular use case (e.g., carrier and cell switch off / on, Radio Frequency (RF) channel reconfiguration off / on, advanced sleep mode selection, O-Cloud resource energy saving mode), the applicability of solution proposals and potential enhancements are also introduced. For example, for carrier and cell switch off / on use case, the making of off / on decisions due to conflicting targets between performance and energy saving is discussed. It is proposed that a decision can be optionally made by an Artificial Intelligence (Al) / Machine Learning (ML) model, which can be deployed at Non-Real Time RAN Intelligent Controller (Non-RT RIC) or Near-RT RIC. In this use case, it is proposed to collect configurations and measurements (e.g., cell load, traffic) from other entities including Service Management and Orchestrator (SMO), E2 Nodes and Open Radio Units (O-RUs), and analyze them for determining the action.

[0012] To enhance energy efficiency in the RAN domain, a set of rApps may be provided in the network. rApp are software applications that are part of the O-RAN architecture. rApps can be used to collect relevant measurements and use analytics to process and generate insights on a network's performance.

[0013] In WO2023113674, a method is proposed to enhance the energy saving of a base station. The method involves an operation that is based on User Equipment (UE) configuration and behavior. To reduce energy consumption, the method enables the base station to dynamically adapt the downlink (DL) transmit power while maintaining correct UE behavior. Since reducing the power level of a base station's DL transmissions may cause inconsistent or unpredictable UE behavior, the method improves the energy saving methodology based on UE configuration.

[0014] Interest in energy conservation measures is not limited to wireless service providers, but extends to the enterprise domain where companies utilize intelligent / smart devices such as Internet of Things (loT) devices, Automated Guided Vehicles (AGVs), robots, etc.

[0015] Efficient wireless communication methods employed by smart devices aim to enhance the efficiency of their wireless communication technology by optimizing energy consumption.

[0016] Summary

[0017] Existing systems implement energy efficiency operations in RAN that are network-centric. That is, many of the current studies of RAN energy efficiency operations have a focus on communication system data and do not consider the constraints and expectations of other domains, such as operational enterprise domains (e.g., enterprise domains that utilise a (e.g., on-premise) RAN deployment). For example, current systems may try to improve energy efficiency in a network by utilising, in their decision making processes, measurements from lower-level RAN nodes (e.g., E2 nodes) and information regarding performance requirements of the RAN (e.g., Quality of Service (QoS) requirements). However, measurements from other operational domains (e.g., from sensors in operational domains) and information regarding features of production processes in operational domains (e.g., production task characteristics and expectations) are not incorporated or accounted for.

[0018] The issue of energy efficiency operations in RAN being network-centric is, in part, caused by the lack of 5G integration with Information Technology (IT) / Operational Technology (OT) systems for energy efficiency solutions.

[0019] It would therefore be beneficial to provide RAN energy efficiency solutions as an internal part of a common energy management framework, such that alignment and synchronization between 5G domains and enterprise domains may be achieved. This may be particularly beneficial for enterprise domains (see the use cases at the end of the description) with high levels of digitalization and / or stringent operational requirements. For such domains, not just network internal operations but also industrial requirements may result in potential conflicts between energy-saving policies that may be applied in the network. As a result, isolated actions implemented in the RAN for energy saving purposes may impact the QoS of production processes in an enterprise domain. For example, an rApp may decide to switch off a first cell to increase energy saving in a network whilst maintaining a certain level of network performance (e.g., an acceptable data rate). However, the first cell may be the only cell providing coverage and connectivity at a particular location to which an industrial application is steering a mobile robot to accomplish a manufacturing task. As such, switching off the first cell would impact the ability of the mobile robot to accomplish the manufacturing task.

[0020] Therefore, it would be beneficial for both network providers and users if energy saving policies could be configured such that conflicts not just within a network domain but across various domains may be handled.

[0021] To address these and other issues, embodiments of the present disclosure enable RAN energy-saving policies to be activated (i.e., configured) based on requirements and / or constraints of both network and industrial domains. As discussed above, this is beneficial, as existing systems may implement network-centric energy efficiency methodologies which do not consider the constraints and / or requirements of other domains (e.g., industry domains).

[0022] In embodiments of the present disclosure, a method may comprise aggregating / retrieving RAN energy efficiency policies in addition to operational constraints and / or requirements of an enterprise domain. This information may be retrieved from the (subsystems of) the RAN and industrial domains. The method utilises an energy saving mechanism that is capable of detecting and resolving potential conflicts between the RAN energy efficiency policies and industry domain communication expectations. That is, while attempting to optimize the energy saving policies to be applied in RAN, the method considers the operational tasks of the enterprise domain along with network policies and measurements. For example, the method may leverage AI / ML techniques using the aggregated information to make decisions and generate recommendations. As such, the method can be used to align and coordinate processes in the enterprise domain with the network domain's energy objectives.

[0023] According to a first aspect of the present disclosure, there is provided a method performed by a node. A RAN has a set of energy saving policies and provides connectivity to a plurality of UEs. The plurality of UEs are for performing one or more operational tasks in an enterprise domain. The method comprises obtaining, from the RAN, first information relating to the operation of the RAN. The method further comprises obtaining, from the enterprise domain, second information relating to the one or more operational tasks. The method further comprises, based on the first information and second information, selecting one or more of the set of energy saving policies for use by the RAN and determining a configuration of the selected one or more energy saving policies.

[0024] According to a second aspect of the present disclosure, there is provided a method performed by a node. A RAN provides connectivity to a plurality of UEs. The plurality of UEs are for performing one or more operational tasks in an enterprise domain. The method comprises obtaining, from the RAN, first information relating to the operation of the RAN. The method further comprises obtaining, from the enterprise domain, second information relating to the one or more operational tasks. The method further comprises determining, based on the first information and / or the second information, a cause of energy consumption in the RAN. The method further comprises, based on the determined cause, generating one or more operational policies for managing the one or more operational tasks to reduce the energy consumption in the RAN by using the first information and the second information.

[0025] According to a third aspect of the present disclosure, there is provided a method performed by a node. A RAN has a set of energy saving policies and provides connectivity to a plurality of UEs. The plurality of UEs are for performing one or more operational tasks in an enterprise domain. The method comprises performing a method according to the first aspect (or any embodiment thereof) and a method according to the second aspect (or any embodiment thereof). According to a fourth aspect of the present disclosure, there is provided a node. A RAN has a set of energy saving policies and provides connectivity to a plurality of UEs. The plurality of UEs are for performing one or more operational tasks in an enterprise domain. The node comprises processing circuitry configured to cause the node to obtain, from the RAN, first information relating to the operation of the RAN. The processing circuitry is further configured to cause the node to obtain, from the enterprise domain, second information relating to the one or more operational tasks. The processing circuitry is further configured to cause the node to, based on the first information and second information, select one or more of the set of energy saving policies for use by the RAN and determine a configuration of the selected one or more energy saving policies.

[0026] According to a fifth aspect of the present disclosure, there is provided a node. A RAN provides connectivity to a plurality of UEs. The plurality of UEs are for performing one or more operational tasks in an enterprise domain. The node comprises processing circuitry configured to cause the node to obtain, from the RAN, first information relating to the operation of the RAN. The processing circuitry is further configured to cause the node to obtain, from the enterprise domain, second information relating to the one or more operational tasks. The processing circuitry is further configured to cause the node to determine, based on the first information and / or the second information, a cause of energy consumption in the RAN. The processing circuitry is further configured to cause the node to, based on the determined cause, generate one or more operational policies for managing the one or more operational tasks to reduce the energy consumption in the RAN by using the first information and the second information.

[0027] According to a sixth aspect of the present disclosure, there is provided a node. A RAN has a set of energy saving policies and provides connectivity to a plurality of UEs. The plurality of UEs are for performing one or more operational tasks in an enterprise domain. The node comprises processing circuitry configured to cause the node to perform a method according to the first aspect (or any embodiment thereof) and a method according to the second aspect (or any embodiment thereof).

[0028] According to a seventh aspect of the present disclosure, there is provided a computer program product comprising a computer readable medium having computer readable code embodied therein, the computer readable code being configured such that, on execution by a suitable computer or processor, the computer or processor is caused to perform the method according to any of the above described aspects or any embodiment thereof. Certain embodiments may provide one or more of the following technical advantage(s):

[0029] 1 ) Enhanced integration of 5G network into the enterprise domain: methods of the present disclosure enable the integration of 5G network into an enterprise domain at an energy management level. The energy saving policies to be applied in RAN, which typically rely on network measurements, can be selected and configured according to the requirements and constraints defined by operational processes in the enterprise domain (e.g., so that the energy saving policies may be optimized). On the other hand, the operational processes (such as tasks performed at a factory floor) can be adjusted for further enhancement of energy saving in the RAN. Information regarding these energy saving policies in the enterprise ecosystem, device behavior at the factory floor, network conditions and other related information can be aggregated in a single Energy Efficiency Module (EEM) for more efficient application of RAN energy efficiency measures. Embodiments of the EEM in the present disclosure introduce a joint optimization process with respect to productivity expectations of an enterprise, performance requirements for device connectivity, and overall energy efficiency goals at an enterprise layer.

[0030] 2) Conflict handling mechanism with AI / ML techniques: embodiments of the present disclosure may incorporate AI / ML techniques to analyze information collected from different sources. For example, an AI / ML model can be used to determine an improved / optimal policy or set of policies to be recommended to the network while handling potential conflicts with the expectations and constraints of the enterprise domain. For example, enhanced energy efficiency in RAN might lead to a decreased performance in production in the enterprise domain. Embodiments of the present disclosure are able to enable energy saving while meeting the expectations of prioritized production processes.

[0031] 3) Joint optimization of the energy saving policies: Higher energy efficiency in the communication networks, lower overall energy consumption, longer battery life for the user equipment, and / or enhanced energy sustainability can be achieved through utilising embodiments of the present disclosure, which combine the network and vertical (enterprise) domain or UE energy optimization policies and executes a joint decision-making process.

[0032] Brief description of the drawings

[0033] For a better understanding of examples of the present disclosure, and to show more clearly how the examples may be carried into effect, reference will now be made, by way of example only, to the following drawings in which: Figure 1 is a schematic diagram illustrating an energy efficiency module according to embodiments of the disclosure;

[0034] Figure 2 illustrates an ML model utilised in embodiments of the disclosure for determining configurations of energy saving policies;

[0035] Figure 3 illustrates a message flow for determining configurations of energy saving policies according to embodiments of the disclosure;

[0036] Figure 4 illustrates an ML model utilised in a first use case implementing embodiments of the disclosure;

[0037] Figure 5 is a schematic diagram illustrating an operational policy generator according to embodiments of the disclosure;

[0038] Figure 6 illustrates a message flow for generating operational policies according to embodiments of the disclosure;

[0039] Figure 7 is a schematic diagram illustrating a 5G Asset Administration Shell (AAS) according to embodiments of the disclosure;

[0040] Figure 8 is a flowchart showing a method in accordance with embodiments of the disclosure;

[0041] Figure 9 is a flowchart showing a method in accordance with embodiments of the disclosure;

[0042] Figure 10 is a schematic diagram of an apparatus according to embodiments of the disclosure; and

[0043] Figure 11 is a schematic diagram of a virtualization environment in which embodiments of this disclosure can be implemented.

[0044] Detailed description

[0045] Some of the embodiments contemplated herein will now be described more fully with reference to the accompanying drawings. Other embodiments, however, are contained within the scope of the subject matter disclosed herein. The disclosed subject matter should not be construed as limited to only the embodiments set forth herein; rather, these embodiments are provided by way of example to convey the scope of the subject matter to those skilled in the art.

[0046] Embodiments of the present disclosure relate to methods for improving energy efficiency in a RAN. For example, some embodiments allow network energy saving policies to be optimised with the support of data provided by the enterprise domain (also referred to herein as being, or comprising, vertical domains, industrial domains, operational domains, etc). By handling specified communication constraints, energy saving policies can be configured such that they simultaneously meet the constraints of production processes in the enterprise domain and network performance requirements.

[0047] Embodiments of the present disclosure enable energy efficiency solutions of an enterprise 5G network to be integrated into an energy management framework of an enterprise domain.

[0048] Methods according to embodiments of the present disclosure are performed by a node. For example, methods according to embodiments of the present disclosure may be implemented by a node referred to as an “Energy Efficiency Module” (EEM). The EEM may be deployed in a RAN (e.g., as an individual network node or as part of O-RAN apparatus) or it may be deployed as an individual / separate network node outside of (but connected to) the RAN. As an overview, embodiments of the EEM: aggregate energy-related information from a network domain and an enterprise domain (e.g., a 5G system and an Operational Technology (OT) systems). For example, the EEM may retrieve information from both network and enterprise domains through exposures, and store this information as historical data; and propose energy-saving policies to be applied in the RAN. For example, the EEM may then use AI / ML techniques to analyze the aggregated data (e.g., measurements, constraints, requirements) and to recommend candidate energy policies to be applied in the network.

[0049] Additionally or alternatively to the above discussed embodiments, the EEM may use a rulebased system to generate a set of operational policies (e.g., recommended operational policies) to be applied in the enterprise domain.

[0050] Figure 1 illustrates the operational framework of an EEM 100 according to embodiments of the present disclosure. The EEM 100 may comprise an energy policy generator 102 and / or an operational policy generator 104. The functionality of the energy policy generator 102 and the operational policy generator 104 is discussed in more detail below.

[0051] The EEM 100 acts as a framework integrating separate domains. Each domain defines its own requirements and constraints. In Figure 1 , the domains are illustrated as being a 5G RAN 106 and an enterprise domain in the form of industrial application 108. However, it should be appreciated the teachings of the present disclosure can be applied to various types of network domains and enterprise domains (e.g., Information Technology (IT) / OT domains). Building a framework integrating different domains typically involves an interface to connect and establish communication via exposure capabilities. For example, in embodiments of the present disclosure, the network domain and enterprise domain provide data and services (e.g., measurements, configuration services) via a set of Application Programming Interfaces (APIs), which are consumed / aggregated by the EEM 100. For further analysis and energy policy generation, the EEM 100 may store aggregated data in a database. The operation of the energy policy generator 102 is discussed below.

[0052] The EEM 100 uses the energy policy generator 102 to generate energy saving policies (that may be implemented in the 5G RAN 106) that meet the constraints and expectations of both the 5G RAN 106 and the industrial application 108.

[0053] The energy policy generator 102 comprises a first data collection sub-module 110, a data preprocessing sub-module 112, and an AI / ML-based policy generation sub-module 114. The AI / ML-based policy generation sub-module 114 comprises an ML model that can be used to generate the energy saving policies for the 5G RAN 106.

[0054] The first data collection sub-module 110 obtains, from the 5G RAN 106, information relating to the operation of the RAN. This information may be referred to herein as “first information”. The first data collection sub-module 110 also obtains, from the industrial application 108, information relating to one or more operational tasks associated with the industrial application 108. This information may be referred to herein as “second information”. The first and second information may be retrieved from different sources (e.g., rApps, the industrial application 108, etc.). The first and second information may be acquired via API requests.

[0055] The methods used for requesting the first and second information (which may include data and reports in a structured format) depend on the capabilities exposed by the respective domains. For example, the EEM 100 may subscribe to events and services exposed by an energy efficiency rApp 122 such that the EEM 100 may address the energy efficiency requirements in the 5G RAN 106.

[0056] The energy efficiency rApp 122 may be part of a Service Management and Orchestration (SMO) entity 124 in the 5G RAN 106. The SMO entity 124 may include a Non-Real-Time Radio Intelligent Controller (Non-RT RIC) for running rApps with the objective of leveraging RAN automation. Events, services, and other related data may be shared with EEM 100 (which is assumed to be an authorized external consumer) through exposures provided by proprietary rApps and the SMO framework. The energy efficiency rApp 122 may be able to collect measurements from lower layers (e.g., near-RT RIC).

[0057] Using AI / ML methodologies, the energy policy generator 102 may implement an autonomous approach to determine the energy saving policies to be activated / deactivated in the 5G RAN 106. The first information aquired by the first data collection sub-module 110, which may include measurements from the 5G RAN 106 and / or a list of available policies supported in the 5G RAN 106 (e.g., at a cell or radio level) may be exposed to the EEM 100 for further operations and to maintain integration of 5G energy management into production processes.

[0058] As stated in the O-RAN Non-RT RIC architecture, the Non-RT RIC framework may support functionality that allows external sources to inject RAN intents and / or suspend / resume / check rApps. Based on the requirements and constraints of the industrial application 108, the energy policy generator 102 may propose policies to be processed by the energy efficiency rApp 122.

[0059] Through external terminations and "data management and exposure functions" in the Non-RT RIC framework, the EEM 100 may interact with the energy efficiency rApp 122 in both directions: (1) the energy efficiency rApp 122 may expose energy saving policies and relevant data to the EEM 100, and (2) the EEM 100 may provide new policy proposals for enhanced energy efficiency in RAN to the energy efficiency rApp 122.

[0060] In O-RAN, there is no standard way of developing exposure to provide this information to external entities. However, there exist relevant interfaces (e.g., in 3GPP) to expose relevant data such as cell locations, performance measurements, resource configurations, etc.

[0061] To determine new policy proposals for enhanced energy efficiency in RAN, an energy efficiency rApp may use a variety of data including one or more of the following:

[0062] • Real-time and historical data on the energy consumption from various network elements, such as base stations, transceivers, and other RAN components

[0063] • KPIs related to the performance of the network (e.g., throughput, latency, packet loss, etc)

[0064] • Network traffic patterns (e.g., peak usage times, locations with high demand, variations in traffic throughout the day, etc)

[0065] • Environmental factors and / or geographic information

[0066] • Number and / or types of devices connected to the network, and optionally user behavior patterns • Network resource usage

[0067] • Distribution of network load across various network components

[0068] • Parameters and / or configurations based on predefined policies and guidelines for energy efficiency.

[0069] The first information obtained by the first data collection sub-module 110 may include any one or more of the above list of information used by energy efficiency rApps.

[0070] The EEM 100 may also implement relevant messaging and interaction capabilities to collect information from industrial applications 108 (i.e., the second information). Through APIs provided by the industrial application 108, the first data collection sub-module 110 may obtain energy comsumption data related to industrial application 108. Some examples of energy- related data include:

[0071] • Device characteristics and condition monitoring: e.g., UE ID, device task, task priority, device capabilities, battery status, battery lifetime, current position, etc

[0072] • Specific Energy Consumption (SEC), measuring the amount of energy used per device for 5G / 6G communication

[0073] • Power Factor, measuring the efficiency of electrical power usage for communication, indicates how effectively power is converted into useful data transfer

[0074] • Environmental data (e.g., conditions, temperature, etc.)

[0075] • UE wireless traffic pattern and mapping to industrial usage

[0076] • Findings from regular energy audits of the industrial application 108

[0077] The second information obtained by the first data collection sub-module 110 may include any one or more of the above list of of energy-related data.

[0078] The APIs made available by industrial domains are mostly proprietary. Since there are different applications providing data at various layers (e.g., Supervisory Control and Data Acquisition (SCADA)), the API set provided by the applications is associated with the function in EEM 100 that is responsible for subscribing to the corresponding services.

[0079] An example of how to exchange information between EEM 100 and industrial application 108 is to use Open Platform Communications (OPC) Unified Architecture (UA), which allows integration of different platforms and the exchanging of standardized data over multiple protocols. In this example, the industrial application 108 may behave as a publisher, while EEM 100 may be a subscriber to retrieve the requested data (i.e., the second information). By making use of the information obtained from both the energy efficiency rApps 122 and the industrial application 108, collaborative processing among different network domains and elements is enabled.

[0080] Once the first data collection sub-module 110 has obtained the first and second information (e.g., through an API), the data preprocessing sub-module 112 may preprocess the data and / or store the data in a database.

[0081] In some embodiments, the data preprocessing sub-module 112 may process the obtained information in order to generate input data that is suitable for use by the M L model of the Al / M L- based policy generation sub-module 114.

[0082] The data preprocessing sub-module 112 may perform any one or more of the following tasks: handling missing values in datasets, data normalization, handling outliers, feature selection, noise removal, etc. The skilled person would understand how such tasks could be performed, depending on the specific data types obtained from the 5G RAN 106 and the industrial application 108.

[0083] For example, the first infromation obtained from the 5G RAN 106 and the industrial application 108 may be in different formats and / or languages. Therefore, in some embodiments, the data preprocessing module 112 may translate the information obtained by the first data collection sub-module 110 into a common format. The translated information may then be stored in the EEM 100 (e.g., in a database) in a structured manner.

[0084] One example of a common format is tabular data, in which data is stored as rows and columns. For example, each row may represent a particular sample, and each column may represent an input feature or output variable.

[0085] For example, policies and related information provided by the 5G RAN 106 may be stored in the EEM 100 with one or more of the following attributes:

[0086] • Source: String

[0087] • Policy Name: String

[0088] • Associated Nodes: Array

[0089] • Allowed Values: Array An example of polcies and related information provided by the 5G RAN 106 in tabular format is as follows: Another type of data that may be provided by the 5G RAN 106 is information associated with measurements collected by the SMO entity 124. These measurements may be exposed to the EEM 100 for generating the energy policy recommendations and may be stored in the EEM 100 with one or more of the following attributes:

[0090] • Cell identifier: String • Cell is on or off: Boolean

[0091] • Physical Location: Coordinates

[0092] • Mean Transmit Power: Integer

[0093] • Reference Signal Received Power (RSRP): Integer

[0094] • Energy consumption: Integer

[0095] An example of measurements provided by the 5G RAN 106 in tabular format is as follows: Regarding information provided by the industrial application 108, this information may be stored in the EEM 100 with one or more of the following attributes:

[0096] • UE ID: String

[0097] • Task: String

[0098] • Task Priority: Integer

[0099] • Device Capabilities: Array

[0100] • Battery Lifetime: Double

[0101] • Available Battery: Double

[0102] • Current Position: Coordinates

[0103] • Timestamp: Date / Time

[0104] An example of information provided by the industrial application 108 in tabular format is as follows:

[0105] To incorporate AI / ML capabilities into the EEM 100, the AI / ML-based generation policy submodule 114 comprises an ML model that can generate energy policies (e.g., as recommendations) for the 5G RAN 106.

[0106] This ML model leverages AI / ML techniques for performing any one or more of the following tasks: data analysis, predictive modelling, decision-making, logical inference, and optimization strategies. The ML model may be trained using supervised learning on data obtained by the first data collection sub-module 110. The training data may be labelled and preprocessed by the data preprocessing sub-module 112.

[0107] To enhance the ML model's performance, minimize the training cost, and / or maximize the overall efficiency, a feature selection methodology may be employed. For example, among a large set of candidate input features, a feature selection approach may maintain the most relevant features and remove redundant features from the input data of the ML model. In one example, the feature selection approach may conclude that task priority is relevant to energy saving policies, while device capabilities are not.

[0108] Examples of feature selection methodologies include: filter-based methodologies (e.g., Chi- squared), wrapper methodologies (e.g., forward selection); and embedded / intrinsic methodologies (e.g., Lasso regression). The optimal choice of the feature selection algorithm may depend on the data characteristics of input data.

[0109] A set of data stored in tabular format (e.g., the example tabular formats discussed above) may be used for both training the ML model and testing the ML model. For example, 80% of the historical data may be used as training data set while the rest can be used to evaluate the performance.

[0110] An example of an ML model that may be used by the AI / ML-based policy generation submodule 114 is illustrated in Figure 2. The ML model is a deep neural network (DNN) 200 that can be used as a regression model to determine energy saving policies that may be implemented in a RAN. The DNN 200 comprises an input layer 202, a number of hidden layers 204, and an output layer 206.

[0111] The input layer 202 (i.e., features) of the ML model 200 comprises the first and second information collected from the 5G RAN 106 and industrial application 108.

[0112] The ML model 200 aims to recommend appropriate energy policies to be applied in the RAN, wherein the energy policies are based on the input data (which is passed through the hidden layers 204). The output variable of the ML model 200 (in the output layer 206) represents the configuration of one or more energy policies to be recommended based on, for example, a given policy and target node. In a first example, the ML model 200 may be provided with an input policy “cellSwitchOnOff” and a target node “cell1”. The output of the ML model 200 provides information about whether celll should be switched on or off (i.e., the ML model 200 provides 0 or 1 as the output value).

[0113] In a second example, the ML model 200 may be provided with an input policy relating to antenna transmission power reconfiguration. For a given target antenna, the ML model 200 determines the optimal transmission power (e.g., 90 dBm) of the target antenna based on, for example, device locations, requirements, task priorities, and / or other relevant information.

[0114] By incorporating AI / ML capabilities into the energy saving generator 102, the energy saving generator 102 can adapt to complex energy patterns, optimize energy consumption, and make informed decisions to improve the overall energy efficiency of a RAN.

[0115] It should be noted that, whilst the ML model 200 of Figure 2 is a DNN, other AI / ML techniques may be employed by the AI / ML-based policy generation sub-module 114. For example, the AI / ML-based policy generation sub-module 114 may utilise Reinforcement Learning (e.g., Deep-Q-Learning). In such embodiments, a reward function could be designed based on, for example, a state of the 5G RAN and a production process of the industrial application 108. For example, the reward function could reward an action that leads to a state where energy consumption in the RAN is decreased while meeting all the expectations of the industrial application 108.

[0116] Figure 3 illustrates a message flow according to embodiments of the present disclosure. In the message flow, an EEM 302 is deployed as a network node and is capable of interacting with both a 5G network (in particular, and SMO 306 of the 5G network in which an energy control rApp 308 is deployed) and an industrial application 304.

[0117] In step 310, the EEM 302 sends, to the industrial application 304, a request for information relating to devices and / or processes in the industrial application 304, such as associated tasks and priority (i.e., second information).

[0118] In step 312, the EEM 302 receives, from the industrial application 304, the requested information relating to devices and / or processes in the industrial application (i.e., the second information). In steps 314, the EEM 302 sends, to the SMO 306, a request for network-related information (i.e. , first information). For example, the information may be obtained from the energy control rApp 308 deployed in the SMO 306.

[0119] In steps 316 and 318, the SMO 306 retrieves the requested network-related information (i.e., the second information). For example, in Figure 3, the SMO 306: sends, to the energy control rApp 308, a request for network-related information (step 316); and receives, from the energy control rApp 308, the requested network-related information. The requested network-related information may comprise any one or more of: available energy saving policies, resource configurations (e.g., cells), and performance measurements (e.g., RSRP in the 5G network, energy consumption in the 5G network).

[0120] In step 320, the SMO 306 sends the requested network-related information to the EEM 302.

[0121] In step 322, after the EEM 302 has received the first and second information from the industrial application 304 and the 5G network, the EEM 302 executes an AI / ML model (e.g., the ML model 200 of Figure 2) to determine a configuration for one or more of the energy saving policies supported by the 5G network / provided by the rApp 308. The configurations may be considered optimal configurations.

[0122] In step 326, the EEM 302 sends the determined configuration of the one or more energy saving policies to the energy control rApp 308 for potential application in a RAN. The determined configurations may be sent via the SMO 306 (step 324).

[0123] A first use case of an energy saving policy generator according to embodiments of the present disclosure is discussed below. The energy saving policy generator may be the energy saving policy generator 102 of Figure 1 and is used to determine a configuration of energy saving policies for potential implementation in a RAN.

[0124] The first use case relates to smart manufacturing involving a mission critical loT service, such as remote control of UEs. For such a service, it may be critical that a RAN provides “always- on time” critical communication with ultra reliability to the remote controlled UEs.

[0125] For the first use case, performance measurements collected from lower layer nodes and connectivity requirements of the (smart manufacturing) production processes should be taken into consideration when determining the configuration of energy saving policies for application in the RAN. That is, in order to enable end-to-end automation for the smart manufacturing process, there should be smooth integration of the production processes and the 5G network, and energy saving policies in the RAN should not interfere with the production and asset management services of the smart manufacturing process.

[0126] For remote control services, one or more of the following constraints may apply: remote controlled devices should always be under coverage (i.e. , connected to the RAN), the latency should be minimized, and / or the reliability of communication (e.g., successful packet transmission without loss) should be maximized.

[0127] In the first use case, a robot A and robot B controlled remotely by a human operator may have critical responsibility in the smart manufacturing process. The energy saving policies implemented in the RAN may need to be activated / deactivated / configured according to the tasks and features of these robots. For example, the cells that provide connectivity for these robots may not be switched off if the production tasks associated with robots A and B are marked as “HIGH” priority. This is because, when a cell is switched off, the remote controlled UEs incorporated in robots A and B may be forced to handover, potentially causing short-term service disruptions due to unsuccessful handover. This can lead to operational failures, which is not acceptable in mission-critical scenarios.

[0128] The proposed energy policy generator therefore gathers information related to robots A and B and their tasks (e.g., task priority, task schedule, and / or current location of the robots) so that the trained ML model can generate an output that recommends an appropriate configuration of energy saving policies for implementation in the RAN. In the case that the robots A and B have high priority tasks, the configuration of the energy saving policies may result in the cells corresponding to the location of robots A and B not being switched off.

[0129] On the other hand, some of the devices or tasks performed by robots in the smart manufacturing process may be identified as low priority, such as software updates for a particular robot. For example, a software update may need to be accomplished under the right conditions, but it may not be a time-critical task that takes priority over the energy efficiency of the RAN.

[0130] For this reason, the ML model of the energy policy generator may recommend energy saving policies to the RAN that result in decreased transmission (Tx) power of, or even to switch off, a cell corresponding to the location of a robot when input data indicates that robots covered by this cell have been assigned tasks with low priority. This scenario is exemplified in Figure 4, which illustrates an ML model 400 (an embodiment of ML model 200 of Figure 2) outputting, in an output layer 402, a determined configuration of an energy saving policy when the input layer 404 of the ML model 400 comprises information indicating a task priority. For example, the input layer 404 may comprise information indicating a policy ID “switchOnOff’, a target RAN node “ru1”, a device capability “drill”, and a task priority “low”.

[0131] After passing the input data through a number of hidden layers 406 of the ML model 400, a configuration indicating “0” (i.e. , “switch off”) is output.

[0132] The first use case can be extended to consider different asset management tasks across the value chain in an Industry 4.0 environment, and should not be limited to considering only the performance expectations (e.g., latency) and / or priorities (e.g., 5G QoS Identifier (5QI)) assigned to traffic flow in an enterprise domain. Rather, the energy saving generator of embodiments of the present disclosure is also capable of considering business level expectations and other critical information about the enterprise domain.

[0133] Returning to Figure 1 , the functionality of the operational policy generator 104 is discussed below. It should be appreciated that, in embodiments of Figure 1 , the operational policy generator may be implemented in addition to the energy policy generator 102. In other embodiments, the operational policy generator 104 is implemented without the energy policy generator 102.

[0134] The EEM 100 uses the operational policy generator 104 to generate candidate operational policies that can be applied in the industrial application 108. The operational policy generator 104 comprises a second data collection sub-module 116, a data preprocessing and root cause analysis (RCA) sub-module 118, and a rule-based policy generation sub-module 120. The operational policy generator 104 uses these sub-modules to process collected data, execute a RCA process when an excessive energy consumption in RAN is detected, and propose operational policies that can be implemented by the industrial application 108 to increase energy saving in the 5G RAN 106 and / or industrial application 108.

[0135] The second data collection sub-module 118 operates in a similar manner to the first data collection sub-module 118 of the energy policy generator 102. That is, the second data collection sub-module 118 retrieves first and second information from the 5G RAN 106 and industrial application 108, which is then consumed by the operational policy generator 104. The operational policy generator 104 may store this information in a database. The data processing and root-cause analysis sub-module 118 aims to detects if there is excessive energy consumption in the 5G RAN 106. To achieve this, the data processing and root-cause analysis sub-module 118 processes the first and second information collected by the second data collection sub-module 116. The processing of the first and second information may be performed in a similar manner to the data preprocessing sub-module 112.

[0136] The processing of the data processing and root-cause analysis sub-module 118 may be periodically triggered to evaluate whether the total energy consumption in the 5G RAN 106 fulfils a criteria (e.g., it exceeds an energy consumption threshold, or it exceeds an energy consumption threshold for at least a required period of time). If any abnormal behavior (i.e., excessive energy consumption) in the 5G RAN 106 is detected, the data processing and rootcause analysis sub-module 118 investigates the cause of the problem by using RCA in both the 5G RAN 106 and industrial application 108 domains.

[0137] Figure 5 illustrates the functionality of the operational policy generator 104 in more detail. The operational policy generator 104 comprises a data collecting and preprocessing function 502, an ML model 504 for problem detection, a timeline correlation function 506, a RCA function 508, and a rule-based operation policy generation function 510.

[0138] Data from a 5G system 512 (i.e., first information) and data from an industrial domain 514 (i.e., second information) is input into the operational policy generator 104. In some examples, the data from the 5G system 512 may comprise antenna measurements, policies, and / or configuration settings. In some examples, the data from the industrial domain 514 may comprise device characteristics, UE traffic patterns, and / or UE’s task-data relation.

[0139] Operational policies 516 are output from the operational policy generator 104, wherein the operational policies 516 may be implemented in the industrial application 108 in order to address a cause of energy consumption in the 5G RAN 106 and / or the industrial application 108.

[0140] The operational policy generator 104 may perform one or more of the following steps:

[0141] 1. Problem detection: The ML model 504 may detect that a problem has occurred via an alarm or via detecting when a criterion has been fulfilled (e.g., when an expected threshold value of a KPI / measurement element has / has not been met).

[0142] 2. Collection and Processing of Data: The data collecting and preprocessing function 502 may collect information relating to the detected problem, including information relating to the environment, equipment, processes, and individuals involved. This may involve reviewing logs, records, and / or reports.

[0143] 3. Finding of the root cause of the problem: the ML model 504 may be used to examine whether there are any other problems that are associated with the detected problem. For example, similar parameters and KPI values related to the detected problem may be examined using the ML model 504 to find out whether any abnormal behavior occurred or is occurring in the 5G RAN 106. Afterwards, the timeline correlation function 506 may find temporal correlations between each of the detected findings. To determine the root cause of the detected problem, the RCA function 508 reaches a conclusion with the findings (e.g., using unsupervised learning methods such as K-Means Clustering, Isolation forest and / or One Class Support Vector Machine). If the concluded root cause is satisfactory, this information is forwarded to the rule-based operation policy generator 510. If the conclusion is not satisfactory, the operational policy generator 104 repeats the operations of the ML model 504, the timeline correlation function 506, and the RCA function 508. Analysis tools and techniques, such as cause-and-effect diagrams or AI / ML methods, may be used by the operational policy generator 104 to identify the core causes of the problem.

[0144] 4. Operational policy generation: After the reason / root cause of the detected problem has been determined, operational policy generator 104 may create and prioritize a list of potential operational actions / policies that enable the detected problem to be avoided. For example, in some embodiments, a rule-based algorithm may be utilised, particularly when predefined problems and solutions are known to the rule-based operation policy generator 510. In other embodiments, an ML method that maps a solution to a problem may be used.

[0145] 5. Sharing of the operational policies 510: The generated operational policies 510 output from the operational policy generator 104 are shared with the industry application 108. The industry application 108 may be advised to implement the operational policies 510. For example, in some embodiments, the generated operational policies may relate to data link usage recommendations for the industrial application 108.

[0146] Figure 6 illustrates a message flow according to embodiments of the present disclosure. In the message flow, the EEM 302 deployed as a network node interacts with both the 5G network (particularly, and SMO 306 of the 5G network in which the energy control rApp 308 may be deployed) and the industrial application 304. In some embodiments, the message flow of Figure 5 may be implemented in addition to or alternatively to the message flow of Figure 3.

[0147] Steps 600 and 601 of Figure 6 correspond to steps 310 and 312 of Figure 3, in which: the EEM 302 sends, to the industrial application 304, a request for information relating to devices and / or processes in the industrial application 304, such as associated tasks and priority (i.e., second information); and the EEM 302 receives, from the industrial application 304, the requested information relating to devices and / or processes in the industrial application (i.e., the second information).

[0148] Steps 602-608 of Figure 6 correspond to steps 314-320 of Figure 3, where the EEM 302 retrieves network-related information (i.e., first information). The network-related information may be obtained from the energy control rApp 308 deployed in the SMO 306. In particular, the EEM 302 retrieves information relating to an energy consumption in the 5G network.

[0149] If the energy consumption in the 5G network is determined to fulfil a criteria (e.g., it meets or exceeds a threshold), the EEM 302 performs the following steps: processing aggregated / obtained network and industrial information (i.e., first and second information) (step 610). This step may correspond to actions performed by the data processing and root-cause analysis sub-module 118 of Figure 1 and / or the data collecting and preprocessing function 502 of Figure 5; performing a RCA process (step 612). This step may correspond to actions performed by the data processing and root-cause analysis sub-module 118 of Figure 1 and / or the ML model 504, timeline correlation function 506 of Figure 5, and / or the RCA function 508 of Figure 5; generating an operational policy (e.g., as a recommendation) (step 614). This step may correspond to steps performed by the rule-based policy generation sub-module 120 of Figure 1 and / or the rule-based operation policy generator 510 of Figure 5; and sending the operational policy recommendation to the industrial application 308 (step 616).

[0150] A second use case is discussed below, in which the EEM 100 uses an operational policy generator according to embodiments of the present disclosure to detect periodic power peaks in network provider data (occurring for some unknown reason) and to examine the power peaks relationship with the enterprise domain.

[0151] A trained ML model of the operational policy generator monitors the network and captures the occurrence of the periodic peaks. The ML model then investigates whether there is a reason for the power peaks, and if so, whether there is a solution to reducing them.

[0152] In the second use case, EEM 100 collects energy-related policy information (from the network and enterprise domains) and network performance-related information, such as Received Signal Strength Indication (RSSI), Tx Power, Signal-to-Noise Ratio (SNR), etc., that has been processed from the 5G RAN domain (i.e. , first and second information). Furthermore, the EEM 100 may request UE data if UE energy consumption exceeds specified limitations set expressly for enterprise purposes. Similarly, enterprise energy policy information includes information on communication energy expectations, limitations, and user equipment energy measurements in the enterprise domain.

[0153] If more data is needed from the network and enterprise domains, the EEM 100 may request additional relevant information (e.g., periodic tasks, upgrade, log upload / download, repair and fail) from the relevant domains. The data may be examined first for a short period of time and then for a long period of time and a relationship may be established using AI / ML techniques.

[0154] To find the root cause of the increase of energy consumption in the RAN domain, value sets of related metrics may be examined by the ML model, and their relationships may be extracted (e.g., to generate a correlation matrix). The ML model may also try to find periodic network behavior in the enterprise domain and its correlation with RAN behavior in the time domain. If it converges to a predefined problem, its RCA is focused on. A series of defined rules can be used to track down the root cause using a graph-based approach to identify the dependencies.

[0155] In the second use case, the root cause of an increase in energy consumption stems from numerous smart devices periodically transmitting data simultaneously, resulting in increased Tx / Receive (Rx) power and increased SNR due to the synchronization.

[0156] Once this issue is identified, the operational policy generator assesses whether it can formulate an operational policy to address it. In this example, the operational policy generator generates an optimal transformation policy related to the scheduling of smart devices within the enterprise domain. For example, a policy proposal is suggested that configures devices in the enterprise domain to transmit data with a defined time difference between each device to reduce energy consumption.

[0157] Implementing a data transformation schedule to reduce SNR and Tx Power offers several benefits, including:

[0158] 1. By reducing interference and optimizing the data transmission process, SNR levels are increased, resulting in enhanced signal quality and decreased packet loss.

[0159] 2. The adaptive transformation reduces unneeded power consumption by dynamically adjusting the Tx Power levels in response to the current environmental conditions. 3. The optimized data transmission scheme increases dependability because the transmitted packets are better adapted for the prevailing SNR levels, thereby decreasing the probability of errors and retransmissions.

[0160] 4. By minimizing interference and enhancing SNR, the network is able to support more simultaneous connections, resulting in increased capacity and an enhanced user experience.

[0161] The EEM 100 then communicates information relating to the problem and its resolution (i.e. , the data transformation schedule) to both the enterprise and network domains. It offers advice on the generated policy and seeks feedback from the service provider and enterprise domain regarding the solution report, advice, and root cause of the problem.

[0162] If the suggested operational policy is applied in the enterprise domain, the EEM 100 may initiate a monitoring phase. For example, The EEM 100 assesses whether the network issue has been resolved or if it has led to any new complications.

[0163] If the desired outcomes are not attained, the EEM may restart the process discussed in relation to Figures 5 and 6 in order to generate new solutions and recommendations through continued communication with both domains.

[0164] In some embodiments, embodiments of the present disclosure may be implemented using the principles of Asset Administration Shell (AAS). AAS is a digital twin concept currently being standardized by the Industry 4.0 platform. AAS is a framework that can be used to describe an asset in the virtual domain. As a result of the ongoing standardization effort in different organizations (e.g., International Electrotechnical Commission (IEC)), the AAS representation and interactions can be handled in a standardized manner. Thus, AAS can enable interoperability on the factory floor.

[0165] One of the key properties implemented by AAS is the concept of submodels, which characterize the asset by describing its aspects in different domains (e.g., communication, engineering, lifecycle status, asset functions). An AAS instance includes active and passive parts. While the passive AAS includes various sub-models describing the asset itself, the active AAS implements relevant functions to interact with other AAS instances as well as make decisions based on the interactions.

[0166] Figure 7 illustrates an EEM implementation in form of AAS. Based on the capabilities of AAS, EEM can be implemented within an AAS representing the 5G Network in the virtual world (e.g., a digital twin of the 5G network). This implementation is illustrated in Figure 7. By default, 5G AAS is expected to accommodate relevant information about the 5G system, such as Data Network Name (DNN) and frequency bands supported by the network (i.e., “Network Identity Submodel” in this example).

[0167] In addition to such information, the data collected from RAN and industrial applications can be stored as separate submodels in 5G AAS. Each data type provided by energy control rApp is stored as separate submodel elements in “Network Energy Submodel”. Cell identifiers, measurements, cell configurations and all relevant information are structured as particular submodel elements in “Network Energy Submodel”. Similar structure is applied to the “Industrial Process Submodel” as well, which consists of the submodel elements representing the information collected from industrial applications. For example, the list of devices, associated tasks, task characteristics, and battery lifetime are structured as separate submodel elements in “Industrial Process Submodel”. Lastly, the candidate policies generated by AAS, which are sent to either RAN or industrial domain, are represented in the “Policy Submodel” with relevant information.

[0168] The relevant capabilities to retrieve information via APIs exposed by network and industrial applications can be implemented as an active part function. A typical AAS implementation is expected to have an “Interaction Model” in the active part, which is responsible of communicating with the physical twin and other AAS instances. In embodiments of the present disclosure, the “Interaction Model” is extended to interact with an energy control rApp and industrial application, subscribing to the events, and fetching and storing relevant information in the corresponding submodels.

[0169] Along with the interaction model, in the active part 5G AAS can incorporate the energy policy generator 102 and operational policy generator 104 that use the information stored in a passive part (i.e., submodels) and generates output in the form of a recommended set of policies to the associated domains.

[0170] As discussed above, the AAS implementation can be considered as a digital twin of the 5G network, which can be extended to utilize as a recommendation mechanism for RAN domain. It brings the required capabilities to integrate disjoint domains (e.g., network, industry), aggregate the data from different sources, execute AI / ML capabilities to generate energysaving policies, and send them to the RAN domain as recommendations. Due to its inherent features, standardized AAS communication language named Industry 4.0, and JavaScript Object Notation (JSON) serialized way of defining the submodels, AAS is a promising tool to implement AAS for Industry 4.0 use cases. Figure 8 illustrates a method performed by a node according to embodiments of the present disclosure. A RAN has a set of energy saving policies and provides connectivity to a plurality of UEs. The plurality of UEs are for performing one or more operational tasks in an enterprise domain (e.g., industrial application 108, 304). The node may be part of the RAN or the enterprise domain.

[0171] The method begins at step 802, with obtaining, from the RAN, first information relating to the operation of the RAN (e.g., via steps 314-320 of figure 3). The first information may include one or more of: performance requirements of the RAN and / or the one or more UEs; and measurements collected in the RAN. For example, the first information may include one or more of: real-time data on an energy consumption of the RAN or part of the RAN; historical data on an energy consumption of the RAN or part of the RAN; measurements of performance parameters for the RAN; information on network traffic patterns; information on environmental factors; information on a geographic location of the RAN; a number of UEs connected to the RAN; types of UEs connected to the RAN; information on user behavior patterns in the RAN; information on network resource usage; and information on a distribution of network load across the RAN.

[0172] At step 804, the node obtains, from the enterprise domain, second information relating to the one or more operational tasks (e.g., via step 310 and 312 of figure 3). The second information may relate to one or more of: characteristics of the one of more UEs; conditions of the one or more UEs; a priority of the one or more operational tasks performed by the one or more UEs; battery information of the one or more UEs; functional capabilities of the one or more UEs; an energy consumption of the one or more UEs; a power efficiency of the one or more UEs; data relating to the environment in which the one or more UEs are operating; a UE traffic pattern; and an energy audit of the enterprise domain. The method may further comprise obtaining policy information relating to the set of energy saving policies. For example, the policy information may comprise a configuration of any one of the energy saving policies currently supported by the RAN.

[0173] At step 806, based on the first information and second information, the node selects one or more of the set of energy saving policies for use by the RAN and determines a configuration of the selected one or more energy saving policies.

[0174] For example, the one or more of the set of energy saving policies may be selected, and / or the configuration of the selected one or more energy saving policies may be determined, using a machine learning model (e.g., a deep neural network). In such embodiments, the method may comprise a step of pre-processing the first information and / or the second information before using the machine learning model.

[0175] In some embodiments, the node may implement the selected one or more energy saving policies and the respective configurations in the RAN (e.g., if it is part of the RAN).

[0176] In other embodiments, the node may send information identifying the selected one or more energy saving policies and the respective configurations to the RAN (e.g., if the node is part of the enterprise domain).

[0177] Figure 9 illustrates a method performed by a node according to embodiments of the present disclosure. A RAN provides connectivity to a plurality of UEs. The plurality of UEs are for performing one or more operational tasks in an enterprise domain (e.g., industrial application 108, 304). The node may be part of the RAN or the enterprise domain.

[0178] The method may begin at step 902, with obtaining, from the RAN, first information relating to the operation of the RAN (e.g., via steps 602-608 of figure 6). The first information may include one or more of: performance requirements of the RAN and / or the one or more UEs; and measurements collected in the RAN. For example, the first information may include one or more of: real-time data on the energy consumption of the RAN or part of the RAN; historical data on the energy consumption of the RAN or part of the RAN; measurements of performance parameters for the RAN; information on network traffic patterns; information on environmental factors; information on a geographic location of the RAN; a number of UEs connected to the RAN; types of UEs connected to the RAN; information on user behavior patterns in the RAN; information on network resource usage; and information on a distribution of network load across the RAN.

[0179] At step 904, the node obtains, from the enterprise domain, second information relating to the one or more operational tasks (e.g., via steps 600 and 601 of Figure 6). The second information may relate to one or more of: characteristics of the one of more UEs; conditions of the one or more UEs; a priority of the one or more operational tasks performed by the one or more UEs; battery information of the one or more UEs; functional capabilities of the one or more UEs; an energy consumption of the one or more UEs; a power efficiency of the one or more UEs; data relating to the environment in which the one or more UEs are operating; a UE traffic pattern; and an energy audit of the enterprise domain.

[0180] At step 906, based on the first information and / or the second information, the node determines a cause of energy consumption in the RAN. This determining may be performed using a trained machine learning model. In such embodiments, a step of pre-processing the first information and / or the second information may be performed before using the trained machine learning model.

[0181] To determine the cause of energy consumption in the RAN, the method may comprise determining (e.g., periodically) whether the energy consumption in the RAN satisfies a criterion. The method may then comprise analysing the first information and / or the second information to determine the cause of energy consumption in the RAN (e.g., responsive to determining that the energy consumption in the RAN satisfies the criterion).

[0182] Determining the cause of energy consumption in the RAN may comprise determining a temporal correlation between one or more events in the enterprise domain and an increase in energy consumption in the RAN. At step 908, based on the determined cause, the node generates one or more operational policies for managing the one or more operational tasks to reduce the energy consumption in the RAN by using the first information and the second information. For example, the node may generate the one or more operational policies using a rule-based algorithm.

[0183] In some embodiments, the node may implement at least one of the one or more operational policies in the enterprise domain (e.g., if the node is part of the enterprise domain).

[0184] In other embodiments, the node may send information for the one or more operational policies to a node in the enterprise domain (e.g., if the node is part of the RAN domain).

[0185] Following at least one of the one or more operational policies being implemented in the enterprise domain, the node may receive feedback from the enterprise domain. The feedback may indicate whether the energy consumption in the RAN and / or the enterprise domain was reduced.

[0186] In some embodiments, a node may perform methods according to embodiments of both figures 8 and 9.

[0187] Figure 10 is a schematic diagram of an apparatus 1000. The apparatus comprises memory 1002, processing circuitry 1004, and interface(s) 1006. The processing circuitry 1004 may be configured such that the apparatus 1000 may be operable to perform the methods discussed in relation to Figures 1-9. For example, the apparatus 1000 may be the node performing the methods described in relation to figures 8 and / or 9.

[0188] It will be appreciated that the apparatus 1000 may comprise one or more virtual machines running different software and / or processes. The apparatus 1000 may therefore comprise, or be implemented in or as one or more servers, switches and / or storage devices and / or may comprise cloud computing infrastructure that runs the software and / or processes.

[0189] The processing circuitry 1004 controls the operation of the apparatus 1000 to implement the relevant part of the methods described herein. The processing circuitry 1004 can comprise one or more processors, processing units, multi-core processors or modules that are configured or programmed to control the apparatus 1000 in the manner described herein. In particular implementations, the processing circuitry 1004 can comprise a plurality of software and / or hardware modules that are each configured to perform, or are for performing, individual or multiple steps of the method described herein in relation to the apparatus 1000. The interface(s) / communications interface(s) 1006 is for use in enabling communications with other apparatus, network nodes, computers, servers, etc. For example, the communications interface 1006 can be configured to transmit to and / or receive from other apparatus or nodes requests, acknowledgements, information, data, signals, or similar. The communications interface 1006 can use any suitable communication technology.

[0190] The processing circuitry 1004 may be configured to control the communications interface 1006 to transmit to and / or receive from other nodes, etc. requests, acknowledgements, information, data, signals, or similar, according to the methods described herein.

[0191] In some embodiments, the memory 1002 can be configured to store program code that can be executed by the processing circuitry 1004 to perform the methods described herein. Alternatively or in addition, the memory 1002 can be configured to store any requests, acknowledgements, information, data, signals, or similar that are described herein. The processing circuitry 1004 may be configured to control the memory 1002 to store such information therein.

[0192] Figure 11 is a block diagram illustrating a virtualization environment 1100 in which functions implemented by some embodiments may be virtualized. In the present context, virtualizing means creating virtual versions of apparatuses or devices which may include virtualizing hardware platforms, storage devices and networking resources. As used herein, virtualization can be applied to any device described herein, or components thereof, and relates to an implementation in which at least a portion of the functionality is implemented as one or more virtual components. Some or all of the functions described herein may be implemented as virtual components executed by one or more virtual machines (VMs) implemented in one or more virtual environments 1100 hosted by one or more of hardware nodes, such as a hardware computing device that operates as a network node, UE, core network node, or host. Further, in embodiments in which the virtual node does not require radio connectivity (e.g., a core network node or host), then the node may be entirely virtualized. In some embodiments, the virtualization environment 1100 includes components defined by the O-RAN Alliance, such as an O-Cloud environment orchestrated by a Service Management and Orchestration Framework via an O-2 interface.

[0193] Applications 1102 (which may alternatively be called software instances, virtual appliances, network functions, virtual nodes, virtual network functions, etc.) are run in a virtualization environment to implement some of the features, functions, and / or benefits of some of the embodiments disclosed herein.

[0194] Hardware 1104 includes processing circuitry, memory that stores software and / or instructions executable by hardware processing circuitry, and / or other hardware devices as described herein, such as a network interface, input / output interface, and so forth. Software may be executed by the processing circuitry to instantiate one or more virtualization layers 1106 (also referred to as hypervisors or virtual machine monitors (VMMs)), provide VMs 1108a and 1108b (one or more of which may be generally referred to as VMs 1108), and / or perform any of the functions, features and / or benefits described in relation with some embodiments described herein. The virtualization layer 1106 may present a virtual operating platform that appears like networking hardware to the VMs 1108.

[0195] The VMs 1108 comprise virtual processing, virtual memory, virtual networking or interface and virtual storage, and may be run by a corresponding virtualization layer 1106. Different embodiments of the instance of a virtual appliance 1102 may be implemented on one or more of VMs 1108, and the implementations may be made in different ways. Virtualization of the hardware is in some contexts referred to as network function virtualization (NFV). NFV may be used to consolidate many network equipment types onto industry standard high volume server hardware, physical switches, and physical storage, which can be located in data centers, and customer premise equipment.

[0196] In the context of NFV, a VM 1108 may be a software implementation of a physical machine that runs programs as if they were executing on a physical, non-virtualized machine. Each of the VMs 1108, and that part of hardware 1104 that executes that VM, be it hardware dedicated to that VM and / or hardware shared by that VM with others of the VMs, forms separate virtual network elements. Still in the context of NFV, a virtual network function is responsible for handling specific network functions that run in one or more VMs 1108 on top of the hardware 1104 and corresponds to the application 1102.

[0197] Hardware 1104 may be implemented in a standalone network node with generic or specific components. Hardware 1104 may implement some functions via virtualization. Alternatively, hardware 1104 may be part of a larger cluster of hardware (e.g. such as in a data center or CPE) where many hardware nodes work together and are managed via management and orchestration 1110, which, among others, oversees lifecycle management of applications 1102. In some embodiments, hardware 1104 is coupled to one or more radio units that each include one or more transmitters and one or more receivers that may be coupled to one or more antennas. Radio units may communicate directly with other hardware nodes via one or more appropriate network interfaces and may be used in combination with the virtual components to provide a virtual node with radio capabilities, such as a radio access node or a base station. In some embodiments, some signaling can be provided with the use of a control system 1112 which may alternatively be used for communication between hardware nodes and radio units.

[0198] Although the computing devices described herein (e.g. the apparatus 1000) may include the illustrated combination of hardware components, other embodiments may comprise computing devices with different combinations of components. It is to be understood that these computing devices may comprise any suitable combination of hardware and / or software needed to perform the tasks, features, functions and methods disclosed herein. Determining, calculating, obtaining or similar operations described herein may be performed by processing circuitry, which may process information by, for example, converting the obtained information into other information, comparing the obtained information or converted information to information stored in the network node, and / or performing one or more operations based on the obtained information or converted information, and as a result of said processing making a determination. Moreover, while components are depicted as single boxes located within a larger box, or nested within multiple boxes, in practice, computing devices may comprise multiple different physical components that make up a single illustrated component, and functionality may be partitioned between separate components. For example, a communication interface may be configured to include any of the components described herein, and / or the functionality of the components may be partitioned between the processing circuitry and the communication interface. In another example, non-computationally intensive functions of any of such components may be implemented in software or firmware and computationally intensive functions may be implemented in hardware.

[0199] It should be noted that the above-mentioned examples illustrate rather than limit the disclosure, and that those skilled in the art will be able to design many alternative embodiments without departing from the scope of the appended embodiments. The word “comprising” does not exclude the presence of elements or steps other than those listed in an embodiment or claim, “a” or “an” does not exclude a plurality, and a single processor or other unit may fulfil the functions of several units recited in the embodiments. Any reference signs in the claims shall not be construed so as to limit their scope.

Claims

CLAIMS1. A method (800) performed by a node (1000), wherein a Radio Access Network, RAN, has a set of energy saving policies and provides connectivity to a plurality of User Equipments, UEs, wherein the plurality of UEs are for performing one or more operational tasks in an enterprise domain, the method comprising: obtaining (802), from the RAN, first information relating to the operation of the RAN; obtaining (804), from the enterprise domain, second information relating to the one or more operational tasks; and based on the first information and second information, selecting (806) one or more of the set of energy saving policies for use by the RAN and determining (806) a configuration of the selected one or more energy saving policies.

2. The method of claim 1, wherein the second information relates to one or more of: characteristics of the one of more UEs; conditions of the one or more UEs; a priority of the one or more operational tasks performed by the one or more UEs; battery information of the one or more UEs; functional capabilities of the one or more UEs; an energy consumption of the one or more UEs; a power efficiency of the one or more UEs; data relating to the environment in which the one or more UEs are operating; a UE traffic pattern; and an energy audit of the enterprise domain.

3. The method of any of claims 1-2, wherein the method further comprises: obtaining policy information relating to the set of energy saving policies.

4. The method of claim 3, wherein the policy information comprises a configuration of any one of the energy saving policies currently supported by the RAN.

5. The method of any of claims 1-4, wherein the first information includes one or more of: performance requirements of the RAN and / or the one or more UEs; and measurements collected in the RAN.

6. The method of any of claims 1-5, wherein the first information includes one or more of:real-time data on an energy consumption of the RAN or part of the RAN; historical data on an energy consumption of the RAN or part of the RAN; measurements of performance parameters for the RAN; information on network traffic patterns; information on environmental factors; information on a geographic location of the RAN; a number of UEs connected to the RAN; types of UEs connected to the RAN; information on user behavior patterns in the RAN; information on network resource usage; and information on a distribution of network load across the RAN.

7. The method of any of claims 1-6, wherein the one or more of the set of energy saving policies are selected, and / or the configuration of the selected one or more energy saving policies are determined, using a machine learning model.

8. The method of claim 7, wherein the machine learning model is a deep neural network.

9. The method of any of claims 7-8, the method further comprising: pre-processing the first information and / or the second information before using the machine learning model.

10. The method of any of claims 1-9, the method further comprising implementing the selected one or more energy saving policies and the respective configurations in the RAN.

11. The method of any of claims 1-10, wherein the node is part of the RAN.

12. The method of any of claims 1-9, wherein the node is part of the enterprise domain.

13. The method of any of claims 1-9 and 12, the method further comprising: sending information identifying the selected one or more energy saving policies and the respective configurations to the RAN.

14. A method (900) performed by a node (1000), wherein a Radio Access Network, RAN, provides connectivity to a plurality of User Equipments, UEs, wherein the plurality ofUEs are for performing one or more operational tasks in an enterprise domain, the method comprising: obtaining (902), from the RAN, first information relating to the operation of the RAN; obtaining (904), from the enterprise domain, second information relating to the one or more operational tasks; determining (906), based on the first information and / or the second information, a cause of energy consumption in the RAN; and based on the determined cause, generating (908) one or more operational policies for managing the one or more operational tasks to reduce the energy consumption in the RAN by using the first information and the second information.

15. The method of claim 14, wherein the method further comprises determining whether the energy consumption in the RAN satisfies a criterion.

16. The method of claim 15, wherein the method further comprises analysing the first information and / or the second information to determine the cause of energy consumption in the RAN.

17. The method of claim 16, wherein the step of analysing occurs responsive to determining that the energy consumption in the RAN satisfies the criterion.

18. The method of any of claims 15 -17, wherein the step of determining whether the energy consumption in the RAN satisfies a criterion is performed periodically.

19. The method of any of claims 14-18, wherein the second information relates to one or more of: characteristics of the one of more UEs; conditions of the one or more UEs; a priority of the one or more operational tasks performed by the one or more UEs; battery information of the one or more UEs; functional capabilities of the one or more UEs; an energy consumption of the one or more UEs; a power efficiency of the one or more UEs; data relating to the environment in which the one or more UEs are operating; a UE traffic pattern; and an energy audit of the enterprise domain.

20. The method of any of claims 14-19, wherein the first information includes one or more of: performance requirements of the RAN and / or the one or more UEs; and measurements collected in the RAN.

21. The method of any of claims 14-20, wherein the first information includes one or more of: real-time data on the energy consumption of the RAN or part of the RAN; historical data on the energy consumption of the RAN or part of the RAN; measurements of performance parameters for the RAN; information on network traffic patterns; information on environmental factors; information on a geographic location of the RAN; a number of UEs connected to the RAN; types of UEs connected to the RAN; information on user behavior patterns in the RAN; information on network resource usage; and information on a distribution of network load across the RAN.

22. The method of any of claims 14-21 , wherein determining the cause of energy consumption in the RAN comprises determining a temporal correlation between one or more events in the enterprise domain and an increase in energy consumption in the RAN.

23. The method of any of claims 14-22, wherein determining the cause of energy consumption in the RAN is performed using a trained machine learning model.

24. The method of claim 23, the method further comprising: pre-processing the first information and / or the second information before using the trained machine learning model.

25. The method of any of claims 14-24, wherein the one or more operational policies are generated using a rule-based algorithm.

26. The method of any of claims 14-25, the method further comprising implementing at least one of the one or more operational policies in the enterprise domain.

27. The method of any of claims 14-26, wherein the node is part of the enterprise domain.

28. The method of any of claims 14-25, wherein the node is part of the RAN29. The method of any of claims 14-25 and 28, the method further comprising: sending information for the one or more operational policies to a node in the enterprise domain.

30. The method of any of claims 14-29, the method further comprising: following at least one of the one or more operational policies being implemented in the enterprise domain, receiving feedback from the enterprise domain, wherein the feedback indicates whether the energy consumption in the RAN and / or the enterprise domain was reduced.

31. A method (800, 900) performed by a node (1000), wherein a Radio Access Network, RAN, has a set of energy saving policies and provides connectivity to a plurality of User Equipments, UEs, wherein the plurality of UEs are for performing one or more operational tasks in an enterprise domain, the method comprising: performing the method of any of claims 1-13; and performing the method of any of claims 14-30.

32. A node (1000), wherein a Radio Access Network, RAN, has a set of energy saving policies and provides connectivity to a plurality of User Equipments, UEs, wherein the plurality of UEs are for performing one or more operational tasks in an enterprise domain, the node comprising processing circuitry (1004) configured to cause the node to: obtain (802), from the RAN, first information relating to the operation of the RAN; obtain (804), from the enterprise domain, second information relating to the one or more operational tasks; and based on the first information and second information, select (806) one or more of the set of energy saving policies for use by the RAN and determine (806) a configuration of the selected one or more energy saving policies.

33. The node of claim 32, wherein the processing circuitry is further configured to cause the node to perform the method of any of claims 2-13.

34. A node (1000), wherein a Radio Access Network, RAN, provides connectivity to a plurality of User Equipments, UEs, wherein the plurality of UEs are for performing one or more operational tasks in an enterprise domain, the node comprising processing circuitry (1004) configured to cause the node to: obtain (902), from the RAN, first information relating to the operation of the RAN; obtain (904), from the enterprise domain, second information relating to the one or more operational tasks; determine (906), based on the first information and / or the second information, a cause of energy consumption in the RAN; and based on the determined cause, generate (908) one or more operational policies for managing the one or more operational tasks to reduce the energy consumption in the RAN by using the first information and the second information.

35. The node of claim 34, wherein the processing circuitry is further configured to perform the method of any of claims 15-30.

36. A node (1000), wherein a Radio Access Network, RAN, has a set of energy saving policies and provides connectivity to a plurality of User Equipments, UEs, wherein the plurality of UEs are for performing one or more operational tasks in an enterprise domain, the node comprising processing circuitry (1004) configured to: perform the method of any of claims 1-13; and perform the method of any of claims 14-30.

37. A computer program product comprising a computer readable medium having computer readable code embodied therein, the computer readable code being configured such that, on execution by a suitable computer or processor, the computer or processor is caused to perform the method of any of claims 1-31.

Images