Enabling ensemble learning in wireless communication networks
The integration of a model learning framework within the network data analytics function to combine multiple machine learning models using ensemble learning strategies addresses the limitations of current systems, enhancing accuracy and performance by effectively utilizing multiple models in wireless communication networks.
Patent Information
- Application Number
- PCT/EP2024/053103
- Authority / Receiving Office
- WO · WO
- Patent Type
- Applications
- Current Assignee / Owner
- Priority Date
- 2024-01-25
- Filing Date
- 2024-02-07
- Publication Date
- 2024-11-28
AI Technical Summary
Current wireless communication systems, particularly 5G networks, lack the capability to effectively combine and utilize multiple machine learning models for ensemble learning, leading to performance degradation when multiple models are used instead of a single model for analytics services, as there is no provision for how the received models should be combined or how input data should be weighted, affecting accuracy and efficiency.
Implementing a method within the network data analytics function to determine a model learning framework for interconnecting multiple machine learning models, providing a group identifier and indication of the framework, and specifying input data preparation and output data combination logic, enabling the use of ensemble learning strategies such as bagging, stacking, or boosting to combine multiple models effectively.
This approach enhances analytics accuracy by allowing the combination of multiple machine learning models using ensemble learning strategies, improving performance and enabling the use of multiple models as a strong learner, rather than individual weak learners, thereby enhancing the overall analytics output.
Smart Images

Figure EP2024053103_28112024_PF_FP_ABST
Abstract
Description
ENABLING ENSEMBLE LEARNING IN WIRELESS COMMUNICATION NETWORKSTECHNICAL FIELD
[0001] The present disclosure relates to wireless communications, and more specifically to enabling ensemble learning in wireless communication networks.BACKGROUND
[0002] A wireless communications system may include one or multiple network communication devices, such as base stations, which may support wireless communications for one or multiple user communication devices, which may be otherwise known as user equipment (UE), or other suitable terminology. The wireless communications system may support wireless communications with one or multiple user communication devices by utilizing resources of the wireless communication system (e.g., time resources (e.g., symbols, slots, subframes, frames, or the like) or frequency resources (e.g., subcarriers, carriers, or the like). Additionally, the wireless communications system may support wireless communications across various radio access technologies including third generation (3G) radio access technology, fourth generation (4G) radio access technology, fifth generation (5G) radio access technology, among other suitable radio access technologies beyond 5G (e.g., sixth generation (6G)).
[0003] Some wireless communication systems, such as 5G systems, support network analytics and artificial intelligence / machine learning (AI / ML). Specifically, a 5G core network may support network analytics and AI / ML via a network data analytics function (NWDAF). The NWDAF may support one or more analytics identifiers (IDs). In some cases, the NWDAF may support one or more of an inference functionality (referred to as NWDAF AnLF (Analytics Logical Function) or a training functionality (referred to as NWDAF MTLF (Model Training Logical Function). An NWDAF AnLF that supports a specific analytics ID inference subscribes to a corresponding MTLF responsible for training.SUMMARY
[0004] An article “a” before an element is unrestricted and understood to refer to “at least one” of those elements or “one or more” of those elements. The terms “a,” “at least one,” “one or more,” and “at least one of one or more” may be interchangeable. As used herein, including in the claims, “or” as used in a list of items (e.g., a list of items prefaced by a phrase such as “at least one of’ or “one or more of’ or “one or both of’) indicates an inclusive list such that, for example, a list of at least one of A, B, or C means A or B or C or AB or AC or BC or ABC (i.e., A and B and C). Also, as used herein, the phrase “based on” shall not be constmed as a reference to a closed set of conditions. For example, an example step that is described as “based on condition A” may be based on both a condition A and a condition B without departing from the scope of the present disclosure. In other words, as used herein, the phrase “based on” shall be construed in the same manner as the phrase “based at least in part on. Further, as used herein, including in the claims, a “set” may include one or more elements.
[0005] Some implementations of the method and apparatuses described herein may further include a first network data analytics function comprising code for network analytics, the code comprising instructions executable by the first network data analytics function to: determine a model learning framework for interconnecting a group of machine learning models to establish a combined machine learning model; and output, to a second network data analytics function, one or more of a group identifier of the group of machine learning models and an indication of the model learning framework for interconnecting the group of machine learning models.
[0006] The instructions may be further executable by the first network data analytics function to determine the group of machine learning models based on one or more of a corresponding identifier associated with each machine learning model of the group of machine learning models, an analytics identifier, a context associated with the group of machine learning models, an accuracy indication, and a cost indication. The context may comprise one or more of an application context, a network topology, a network hardware, a software characteristic, a network load level, a network energy level, or a timing.
[0007] The instructions may be further executable by the first network data analytics function to receive the group identifier from the second network data analytics function anduse the group identifier to determine the group of machine learning models and / or each machine learning model to be used in combination with each other.
[0008] The indication of the model learning framework may include a private field enabling the indication of a specific framework not selected from a standardized list, and / or may include a public field chosen from a predefined lists of framework names. The indication of the model learning framework may include an indication of the role of each individual machine learning model in the group.
[0009] The instructions may be further executable by the first network data analytics function to output to the second network analytics function an indication of a preparation method of input data for input into each machine learning model in the group. The preparation of input data may include a logic for producing an input data set based on one or more of the following: a random distribution logic to prepare the input data set; a repetitive process logic that uses the same input data set; a weight-based logic to prepare the input data set; and an algorithmic or model-based logic to prepare the input data set.
[0010] The instructions may be further executable by the first network data analytics function to output to the second network analytics function an indication of consolidation of output data of each machine learning model in the group to create a unified output. The consolidation of output data may include a logic for producing a unified output based on one or more of the following: a weighted combination logic; an algorithm or model logic; a voting logic; and an aggregation logic.
[0011] The instructions may be further executable by the first network data analytics function to store and retrieve multiple machine learning models from a repository function in conjunction with at least one of a group identifier, an analytics identifier, a use case context, an accuracy indicator and a cost indicator.
[0012] The first network data analytics function may be configured to operate as an analytics model training function for providing training for a combined plurality of machine learning models, the first network data analytics function further comprising at least one memory, and at least one processor coupled with the at least one memory and configured tocause the first network data function to execute the instructions. The second network data analytic function may be an analytics inference function.
[0013] Some implementations of the method and apparatuses described herein may further include a second network analytics function for use in a wireless network, wherein an analytics output is derived based on a combined machine learning model established from a plurality of machine learning models, the second network analytics function comprising: at least one memory; and at least one processor coupled with the at least one memory and configured to cause the second network analytics function to: determine a group of machine learning models to be used in combination with each other; and send to a first analytics function a request for a unified model formed from a group of machine learning models, the request including at least one of a group identifier of the group of machine learning models, an identifier of each machine learning model in the group and an analytics service identifier.
[0014] The processor may be further configured to cause the second network analytics function to discover and select the first network analytics function from which to request the combined machine learning model based on at least one of a group model identifier, an identifier of each ML model in the group, analytics service identifier, a usage context, accuracy, and cost.
[0015] The processor may further be configured to cause the second network analytics function to receive from the first network analytics function the group identifier of the group of machine learning models and an indication of a model learning framework for interconnecting the group of machine learning models. The processor may further be configured to cause the second network analytics function to send to the first network analytics function an indication that the second network analytics functions supports ensemble learning.
[0016] The first network analytics function may be an analytics model training function, and the second network analytics function may be an analytics inference function.
[0017] Some implementations of the method and apparatuses described herein may further include a method performed by an analytics model training function that providestraining from a plurality of ML models, the method comprising: determining a model learning framework for interconnecting a group of machine learning models to establish a combined machine learning model; and sending to an analytics inference function one or more of a group identifier of the group of machine learning models and an indication of the model learning framework.
[0018] Some implementations of the method and apparatuses described herein may further include a processor in an analytics model training function for providing training for a combined plurality of machine learning models in a wireless network, the processor comprising at least one controller coupled with at least one memory and configured to cause the processor to: determine a model learning framework for interconnecting a group of machine learning models to establish a combined machine learning model; and output, to a second network data analytics function, one or more of a group identifier of the group of machine learning models and an indication of the model learning framework for interconnecting the group of machine learning models.BRIEF DESCRIPTION OF THE DRAWINGS
[0019] Figure 1 illustrates an example of a wireless communications system in accordance with aspects of the present disclosure.
[0020] Figure 2 illustrates an example of a network architecture in accordance with aspects of the present disclosure.
[0021] Figure 3 illustrates examples of ensemble learning techniques in accordance with aspects of the present disclosure.
[0022] Figure 4 illustrates a signaling diagram that supports model provisioning services in accordance with aspects of the present disclosure in accordance with aspects of the present disclosure.
[0023] Figure 5 illustrates an example of an analytics model training function in accordance with aspects of the present disclosure.
[0024] Figure 6 illustrates an example of a processor in accordance with aspects of the present disclosure.
[0025] Figure 7 illustrates an example of an analytics inference function in accordance with aspects of the present disclosure.
[0026] Figure 8 illustrates a flowchart of a method performed by an analytics model training function in accordance with aspects of the present disclosure.
[0027] Figure 9 illustrate a flowchart of a method performed by an analytics inference function in accordance with aspects of the present disclosure.DETAILED DESCRIPTION
[0028] A core network of a wireless communication system may include various network functions to support applications, services, analytics, and the like for various radio access technologies including 4G, 5G, among other suitable radio access technologies beyond 5G (e.g., 6G). Figure 2 illustrates a network architecture 200. In some implementations, an NWDAF 202 of the core network may support one or more of network analytics and learning models (e.g., AI / ML). The NWDAF 202 may include one or more of an AnLF 204 or an MTLF 206 as described herein. The NWDAF may one or more of receive information from input sources and transmit information to output sources (e.g., output consumers), which may include other network functions of the core network, application functions (AFs) 208, repositories 210 (e.g., network repository functions (NRF), user data managers (UDM), etc.), and operations, administration and maintenance (0AM) 212, which may be a management service (MnS) consumer or a management function (MF). In some implementations, one or more of the AnLF 204 or the MTLF 206 may exchange AI / ML models, for example, model provisioning as described in TS 23.288 clause 6.2A. Additionally, or alternatively, one or more of a data collection coordination functionality (DCCF) or messaging framework adapter function (MFAF) 214 may distribute and collect repeated data to or from various sources in the core network.
[0029] In some cases, analytics services may be provided dependent on whether each AnLF includes a single ML model for inference. However, there are cases where an AnLF may use multiple ML models to generate an analytics output result related to a specific analytics ID. The AnLF may select to use a specific ML model based on a use case context or other filter information, or alternatively use multiple ML models at the same time toobtain a prediction for a single analytics ID. Using multiple ML models may be referred to as ensemble learning, which may include a combination of several inter-connected ML models. Put another way, ensemble learning may use simpler ML models that complement each other. The concept being that a combination of several weak learners can become a strong learner. Ensemble learning (i.e., the combined use of multiple ML models) can be used as an alternative to a single ML model analytics by the NWDAF to provide any analytics ID services.
[0030] As described herein with reference to Figure 3, ensemble learning methods (e.g., strategies or frameworks) may include one or more of a bagging ensemble method, a stacking ensemble method, or a boosting ensemble method. The bagging ensemble method supports use of different ML models as separate decision trees on different samples of the same dataset and combines predictions, for example by averaging or voting. The stacking ensemble method supports use of different ML models on the same data and then a further ML model to learn how to combine them. The boosting ensemble method supports use of adding different models sequentially that correct the predictions made by prior models and outputs a combined weighted average of the predictions.
[0031] The use of multiple ML models focuses on introducing a unique identifier per ML model to enable an AnLF to request them from the respective MTLF. The respective pameters currently defined for multiple ML models aim to assist analytics accuracy enhancements. However, multiple ML models may be used not only to enhance accuracy but also as a group defining a different ML model based on a specified ensemble method (e.g., strategy). According to some techniques, once these ML models are received by a requesting AnLF, there is no information related to the ensemble learning method that should be employed. That is, there is no provision for how the received ML models are to be used, or for any other information related to, for example, combining the output data or introducing input data weights, or related to the role of each ML model (e.g., ensemble member or aggregator role).
[0032] Various aspects of the present disclosure relate to an apparatus and method that support ensemble learning to enable analytic services that considers use of multiple ML models, instead of a single ML model used in NWDAF Analytics ID services. Multiplemodels are combined or inter-connected following a chosen strategy. The combination or inter-connection of multiple models may be performed initially during a training phase, for example, performed by an NWDAF MTLF, which determines the input data preparation logic and the output data combination logic. The NWDAF MTLF provides to the AnLF information identifying attributes of the ML group model. This enables the AnLF to use the group ML model or the ensemble leaning model as it was trained.
[0033] Although aspects of the present disclosure are discussed in the context of an NWDAF AnLF and MWDAF MTLF, it will be appreciated that the principles can be applied to an analytics inference function and analytics model training function more generally.
[0034] Aspects of the present disclosure are described in the context of a wireless communications system.
[0035] Figure 1 illustrates an example of a wireless communications system 100 in accordance with aspects of the present disclosure. The wireless communications system 100 may include one or more NE 102, one or more UE 104, and a core network (CN) 106. The wireless communications system 100 may support various radio access technologies. In some implementations, the wireless communications system 100 may be a 4G network, such as an LTE network or an LTE- Advanced (LIE- A) network. In some other implementations, the wireless communications system 100 may be a NR network, such as a 5G network, a 5G- Advanced (5G-A) network, or a 5G ultrawideband (5G-UWB) network. In other implementations, the wireless communications system 100 may be a combination of a 4G network and a 5G network, or other suitable radio access technology including Institute of Electrical and Electronics Engineers (IEEE) 802.11 (Wi-Fi), IEEE 802.16 (WiMAX), IEEE 802.20. The wireless communications system 100 may support radio access technologies beyond 5G, for example, 6G. Additionally, the wireless communications system 100 may support technologies, such as time division multiple access (TDMA), frequency division multiple access (FDMA), or code division multiple access (CDMA), etc.
[0036] The one or more NE 102 may be dispersed throughout a geographic region to form the wireless communications system 100. One or more of the NE 102 describedherein may be or include or may be referred to as a network node, a base station, a network element, a network function, a network entity, a radio access network (RAN), a NodeB, an eNodeB (eNB), a next-generation NodeB (gNB), or other suitable terminology. An NE 102 and a UE 104 may communicate via a communication link, which may be a wireless or wired connection. For example, an NE 102 and a UE 104 may perform wireless communication (e.g., receive signaling, transmit signaling) over a Uu interface.
[0037] An NE 102 may provide a geographic coverage area for which the NE 102 may support services for one or more UEs 104 within the geographic coverage area. For example, an NE 102 and a UE 104 may support wireless communication of signals related to services (e.g., voice, video, packet data, messaging, broadcast, etc.) according to one or multiple radio access technologies. In some implementations, an NE 102 may be moveable, for example, a satellite associated with a non-terrestrial network (NTN). In some implementations, different geographic coverage areas 112 associated with the same or different radio access technologies may overlap, but the different geographic coverage areas may be associated with different NE 102.
[0038] The one or more UE 104 may be dispersed throughout a geographic region of the wireless communications system 100. A UE 104 may include or may be referred to as a remote unit, a mobile device, a wireless device, a remote device, a subscriber device, a transmitter device, a receiver device, or some other suitable terminology. In some implementations, the UE 104 may be referred to as a unit, a station, a terminal, or a client, among other examples. Additionally, or alternatively, the UE 104 may be referred to as an Internet-of- Things (loT) device, an Intemet-of-Everything (loE) device, or machine-type communication (MTC) device, among other examples.
[0039] A UE 104 may be able to support wireless communication directly with otherUEs 104 over a communication link. For example, a UE 104 may support wireless communication directly with another UE 104 over a device-to-device (D2D) communication link. In some implementations, such as vehicle-to-vehicle (V2V) deployments, vehicle-to-everything (V2X) deployments, or cellular-V2X deployments, the communication link 114 may be referred to as a sidelink. For example, a UE 104 may support wireless communication directly with another UE 104 over a PC5 interface.
[0040] An NE 102 may support communications with the CN 106, or with another NE 102, or both. For example, an NE 102 may interface with other NE 102 or the CN 106 through one or more backhaul links (e.g., SI, N2, N2, or network interface). In some implementations, the NE 102 may communicate with each other directly. In some other implementations, the NE 102 may communicate with each other or indirectly (e.g., via the CN 106. In some implementations, one or more NE 102 may include subcomponents, such as an access network entity, which may be an example of an access node controller (ANC). An ANC may communicate with the one or more UEs 104 through one or more other access network transmission entities, which may be referred to as a radio heads, smart radio heads, or transmission-reception points (TRPs).
[0041] The CN 106 may support user authentication, access authorization, tracking, connectivity, and other access, routing, or mobility functions. The CN 106 may be an evolved packet core (EPC), or a 5G core (5GC), which may include a control plane entity that manages access and mobility (e.g., a mobility management entity (MME), an access and mobility management functions (AMF)) and a user plane entity that routes packets or interconnects to external networks (e.g., a serving gateway (S-GW), a Packet Data Network (PDN) gateway (P-GW), or a user plane function (UPF)). In some implementations, the control plane entity may manage non-access stratum (NAS) functions, such as mobility, authentication, and bearer management (e.g., data bearers, signal bearers, etc.) for the one or more UEs 104 served by the one or more NE 102 associated with the CN 106.
[0042] The CN 106 may communicate with a packet data network over one or more backhaul links (e.g., via an SI, N2, N2, or another network interface). The packet data network may include an application server. In some implementations, one or more UEs 104 may communicate with the application server. A UE 104 may establish a session (e.g., a protocol data unit (PDU) session, or the like) with the CN 106 via an NE 102. The CN 106 may route traffic (e.g., control information, data, and the like) between the UE 104 and the application server using the established session (e.g., the established PDU session). The PDU session may be an example of a logical connection between the UE 104 and the CN 106 (e.g., one or more network functions of the CN 106).
[0043] In the wireless communications system 100, the NEs 102 and the UEs 104 may use resources of the wireless communications system 100 (e.g., time resources (e.g., symbols, slots, subframes, frames, or the like) or frequency resources (e.g., subcarriers, carriers)) to perform various operations (e.g., wireless communications). In some implementations, the NEs 102 and the UEs 104 may support different resource structures. For example, the NEs 102 and the UEs 104 may support different frame structures. In some implementations, such as in 4G, the NEs 102 and the UEs 104 may support a single frame structure. In some other implementations, such as in 5G and among other suitable radio access technologies, the NEs 102 and the UEs 104 may support various frame structures (i.e., multiple frame structures). The NEs 102 and the UEs 104 may support various frame structures based on one or more numerologies.
[0044] One or more numerologies may be supported in the wireless communications system 100, and a numerology may include a subcarrier spacing and a cyclic prefix. A first numerology (e.g., / r=0) may be associated with a first subcarrier spacing (e.g., 15 kHz) and a normal cyclic prefix. In some implementations, the first numerology (e.g., / r=0) associated with the first subcarrier spacing (e.g., 15 kHz) may utilize one slot per subframe. A second numerology (e.g., / r=l) may be associated with a second subcarrier spacing (e.g., 30 kHz) and a normal cyclic prefix. A third numerology (e.g., / r=2) may be associated with a third subcarrier spacing (e.g., 60 kHz) and a normal cyclic prefix or an extended cyclic prefix. A fourth numerology (e.g., / r=3) may be associated with a fourth subcarrier spacing (e.g., 120 kHz) and a normal cyclic prefix. A fifth numerology (e.g., / r=4) may be associated with a fifth subcarrier spacing (e.g., 240 kHz) and a normal cyclic prefix.
[0045] A time interval of a resource (e.g., a communication resource) may be organized according to frames (also referred to as radio frames). Each frame may have a duration, for example, a 10 millisecond (ms) duration. In some implementations, each frame may include multiple subframes. For example, each frame may include 10 subframes, and each subframe may have a duration, for example, a 1 ms duration. In some implementations, each frame may have the same duration. In some implementations, each subframe of a frame may have the same duration.
[0046] Additionally or alternatively, a time interval of a resource (e.g., a communication resource) may be organized according to slots. For example, a subframe may include a number (e.g., quantity) of slots. The number of slots in each subframe may also depend on the one or more numerologies supported in the wireless communications system 100. For instance, the first, second, third, fourth, and fifth numerologies (i.e., / r=0, jU=l , / r=2, jU=3, / r=4) associated with respective subcarrier spacings of 15 kHz, 30 kHz, 60 kHz, 120 kHz, and 240 kHz may utilize a single slot per subframe, two slots per subframe, four slots per subframe, eight slots per subframe, and 16 slots per subframe, respectively. Each slot may include a number (e.g., quantity) of symbols (e.g., OFDM symbols). In some implementations, the number (e.g., quantity) of slots for a subframe may depend on a numerology. For a normal cyclic prefix, a slot may include 14 symbols. For an extended cyclic prefix (e.g., applicable for 60 kHz subcarrier spacing), a slot may include 12 symbols. The relationship between the number of symbols per slot, the number of slots per subframe, and the number of slots per frame for a normal cyclic prefix and an extended cyclic prefix may depend on a numerology. It should be understood that reference to a first numerology (e.g., / r=0) associated with a first subcarrier spacing (e.g., 15 kHz) may be used interchangeably between subframes and slots.
[0047] In the wireless communications system 100, an electromagnetic (EM) spectrum may be split, based on frequency or wavelength, into various classes, frequency bands, frequency channels, etc. By way of example, the wireless communications system 100 may support one or multiple operating frequency bands, such as frequency range designations FR1 (410 MHz - 7.125 GHz), FR2 (24.25 GHz - 52.6 GHz), FR3 (7.125 GHz - 24.25 GHz), FR4 (52.6 GHz - 114.25 GHz), FR4a or FR4-1 (52.6 GHz - 71 GHz), and FR5 (114.25 GHz - 300 GHz). In some implementations, the NEs 102 and the UEs 104 may perform wireless communications over one or more of the operating frequency bands. In some implementations, FR1 may be used by the NEs 102 and the UEs 104, among other equipment or devices for cellular communications traffic (e.g., control information, data). In some implementations, FR2 may be used by the NEs 102 and the UEs 104, among other equipment or devices for short-range, high data rate capabilities.
[0048] FR1 may be associated with one or multiple numerologies (e.g., at least three numerologies). For example, FR1 may be associated with a first numerology (e.g., / r=0), which includes 15 kHz subcarrier spacing; a second numerology (e.g., / r=l), which includes 30 kHz subcarrier spacing; and a third numerology (e.g., / r=2), which includes 60 kHz subcarrier spacing. FR2 may be associated with one or multiple numerologies (e.g., at least 2 numerologies). For example, FR2 may be associated with a third numerology (e.g., / r=2), which includes 60 kHz subcarrier spacing; and a fourth numerology (e.g., / r=3), which includes 120 kHz subcarrier spacing.
[0049] As previously discussed, the present disclosure relates to an apparatus and method that employs ensemble learning for enabling analytic services that considers the use of multiple ML models instead of a single ML model currently used in NWDAF Analytics ID services. Multiple models are combined or inter-connected following a chosen strategy, or model learning framework. The combination or inter-connection of multiple models is performed initially in the training phase, i.e., carried out by an analytics model training function (which may be an NWDAF MTLF), which determines the input data preparation logic and the output data combination logic. The analytics model training function then provides to an analytics inference function (which may be an NWDAF AnLF) information identifying attributes of the ML group model. This enables the analytics inference function to use the group ML model or the ensemble Leaning Model as it was trained.
[0050] This disclosure focuses on introducing the information needed to allow the analytics inference function (e.g AnLF) to know how the combined multiple ML models may be used for inference considering the training preparation effort by the respective analytics model training function (e.g. MTLF). The following discussion is directed specifically to an arrangement with an AnLF and MTFL but it will be appreciated that it can apply to any network deploying network analytics and AI / ML. An MTFL may prepare, i.e. provide training for, for the same group of ML models or any combination of ML models employing different ensemble learning strategies, considering the desired accuracy, preparation cost (e.g., energy) and usage context or use case context.
[0051] Initially there is a need to allow the AnLF to know that the use of multiple ML models received should be combined following an ensemble learning strategy. This avoidsthe AnLF choosing a single ML model out of the multiple ones received, since this is an option that would result in performance degradation if the original use of the multiple ML models were combined. To achieve this an umbrella identifier for the group of multiple ML models is introduced to allow the AnLF to know that the specific ML Models received (each identified by a unique ML model ID) are to be used in a specific combined way, i.e. following the indicated ensemble learning strategy. An example of such an umbrella identifier is described herein for convenience as an “EL Group Model ID”: this is an identifier that indicates the grouping of the respective ML model members.
[0052] Furthermore, each ensemble learning group needs to employ an ensemble learning strategy that indicates how the group of multiple ML models are to be combined and used. The ensemble strategy field can be:• a private field that may allow the MTLF to indicate the specific strategy without the need to standardize a specific list of strategies including a name identifier, or alternatively,• a public field that contains a predefined set of strategy names as identifiers (e.g. Bagged Decision Trees, Stacked Models, Blending, AdaBoost, Stochastic Gradient Boosting, etc.) that can be selected to describe the ensemble learning strategy.
[0053] The usage context and / or desired accuracy or preparation cost can assist in discovering and selecting a specific EL Group Model ID and / or the ensemble learning strategy.
[0054] Each ensemble learning group may adopt a different method or mechanism to prepare the input data for each ML model member of the ensemble learning group, and to combine the output data from each different ML model of the ensemble learning group to create the output result towards the analytics consumer.
[0055] To prepare the input data for each ML model member of the ensemble learning group it will be necessary to introduce a new field that needs to hold a description of the preparation method. Examples of this may include:(i) random sampling to create a subset of the input data for each ML model member,(ii) an indication that the same input data should be used for all ML model members,(iii) the weights that may format and / or modify the input data to prepare it for each ML model member, or(iv) the weights that impact the sample input for each ML model member and the sequence of using each ML model member (applied for the case of boosting ensemble learning strategy).
[0056] This field can be private without standardizing the potential options for preparing the data input, or can be a public field with a standardized identifier (e.g., using a descriptor identifier) for each preparation option related to the input data.
[0057] To combine the output data from each different ML model of the ensemble learning group it will be necessary to introduce a new field to hold a description of how to combine the output of each ML model member to create the analytics output towards the consumer. Different options for combining the output of each ML model member may include, for example:(i) aggregation (e.g. can be used when the input data set is divided based on a specified logic and processed by each ML model separately),(ii) voting (e.g. can be used when the same input data is processed by each ML model and then the “best performing” option is chosen),(iii) an ML model including an identifier, which decides how to combine the output data,(iv) weights related to the output of each ML model member, etc.
[0058] Examples when weights or an ML model is used to combine the output data may include cases when the output of each ML model can be used to enhance the performance of the previous one, or when the role of each model focuses on a certain percentage of the input data or has a difference significance.
[0059] In a similar manner to the preparation process, this field can be private without standardizing the potential options or it can be a public field with a standardized identifier (e.g. using a descriptor identifier) for each combination option related to the input data. To avoid confusion in the case where another ML model is used to combine the output data, aparameter related to the role of each ML model may be included besides the “EL Group Model ID”. The role of each ML model may describe where each ML model should be used, e.g. in the initial phase, final phase or other, what type of input data it should consume and what type of output data it should produce.
[0060] Depending on usage context, i.e. use case, and / or on the network environment conditions, e.g. peak time or off-peak time, network congestion, energy saving, etc., there may be a different set of options for preparing the data input for each ML model member of the ensemble learning group and / or a different set of options for combining the output of ML model members to create the analytics output towards the consumer.
[0061] Hence, multiple options for preparing the input data and combining the output data may exist accompanied by a usage context field that indicates the conditions that may apply, which may include:• context attribute: slice or service parameter, QoS profile, a KPI, weight, time, etc., and• context condition: equal or greater / lower than a certain numeric value or a match to a certain state or profile and optionally a context value range of, for example, numeric values or identifiers or states or similar.
[0062] An embodiment of this disclosure may include the current 5G core network analytics, i.e., NWDAF. Each MTLF that supports ensemble learning needs to indicate this capability in a Network Repository Function (NRF) including the supported model IDs and / or the “EL Group Model ID” as well as the preferred accuracy, model interoperability information, the expected inference and training time, and other filter information related to attributes such as the time availability, area of interest, slice, etc. The “EL Group Model ID” and the ML model ID related to each member model may also be included in the Analytical Data Repository Function (ADRF), which is the network repository function that stores data retrieved and ML model information, for future usage.
[0063] An AnLF may be locally configured with a set of MTLF IDs and the respective Analytics ID(s) supported by each MTLF to retrieve trained ML models or may use the NWDAF discovery procedure specified in clause 5.2 TS 23.288 for discovering the desired MTLF(s). An NWDAF containing MTLF may determine that further training for anexisting ML model is needed when it receives the ML model subscription or the ML model request.
[0064] The parameters that can be provided by the AnLF (i.e., NWDAF service consumer) for subscribing, modifying, and requesting ML Model provisioning are listed in clause 6.2A.2 TS 23.288. In addition to the parameters listed there, the following parameters may enhance the ML model provisioning for the case of ensemble learning including:1. ML Model subscribe / unsubscribe or request:• Indication of supporting Ensemble Learning• Ensemble Learning Group Model Identifier (“EL Group Model ID”): Indicates the group of learning ensemble Models and optionally also contains information related to the supported Ensemble Learning strategy.• Indication of supporting Ensemble Learning strategy may alternatively be a separate parameter.• Indication of supporting ML model identifier for each model in the group may alternatively be provided instead of the Ensemble Learning Group Model Identifier.2. ML Model notify or response:• Indication of the method or mechanism that should be used to prepare the input data for each ML model member of an Ensemble Learning Group.• Indication of the method or mechanism to be used to combine the output data of each ML model member of an Ensemble Learning Group to create the output towards the consumer.• Indication of supporting Ensemble Learning strategy and / or indication of the role of each ML model in the group.
[0065] It should be noted that, if the AnLF has the capability to support multiple ML models and needs a “cheap” ML model solution, (i.e. an ML model with lower performance) then Ensemble Learning may also be the choice of the MTLF, with the selection of Ensemble Learning strategy based on the indicated preferred accuracy, training availability, expected time for inference and training or on other filter information.
[0066] Figure 4 illustrates a signalling flowchart for model provisioning services in accordance with aspects of the present disclosure.
[0067] For the present purposes we can assume that an NWDAF AnLF (z.e. NWDAF Service Consumer) uses for inference multiple ML models forming an Ensemble Learning ML model (i.e. with a specific group model ID “EL Group Model ID”, EL strategy, etc. as discussed above). The procedure illustrated in Figure 4 may be used by the NWDAF AnLF to subscribe / unsubscribe or request from a selected MTLF the desired ML Model Information for certain Analytics ID(s), using Nnwdaf MLModelProvision services as defined in clause 7.5 TS 23.288 and NnwdafJMLModellnfo services as defined in clause 7.6 TS 23.288. The ML Model Information can also be used by an AnLF to modify existing ML Model Subscription(s).
[0068] In step 400 an NWDAF MTLF 412 (analytics model training function) may optionally determine to store, in an ADRF 413, ML models owned by the MTLF 412. These models may include one or more group of ML models related to ensemble learning. Details related to a specified group ID “EL Group Model ID”, EL strategy, usage context, accuracy, preparation cost, etc. may be provided by the MTLF 412 to the ADRF 413 for future usage.
[0069] In step 401 an NWDAF AnLF 411 (analytics inference function) decides that it needs ML Model Information (e.g., ML model training) for certain Analytics ID(s). The AnLF may specify an EL Group Model ID to be used for inference. As discussed above, the EL Group Model ID is an umbrella identifier that indicates the grouping of the respective ML model members required by the AnLF 411.
[0070] In a first option the NWDAF AnLF 411 may discover and select an appropriate NWDAF MTLF 412 via NRF (using in the discovery process information related to EL Group Model ID, EL strategy, etc.).
[0071] In a second option the AnLF 411 may discover and select an appropriate NWDAF MTLF 412 via NRF using the multiple model IDs and the Analytics ID (including optionally the usage context).
[0072] Alternatively, the NWDAF AnLF 411 may be preconfigured with the information regarding which NWDAF MTLF 412 should be used, in which case no discovery is required.
[0073] In step 402 the NWDAF AnLF 411 subscribes to, modifies, or cancels a subscription related to a set of ML Models Information held by the NWDAF MTLF 412 by invoking Nnwdaf_MLModelProvision_Subscribe / Unsubscribe service operation, or requests ML Model(s) Information by invoking Nnwdaf MLModellnfo Request service operation. The NWDAF AnLF 411 includes in the subscription or request metadata related to ensemble learning.
[0074] In the first option the subscription or request is associated with at least one Analytics ID, EL Group Model ID, and / or the EL strategy related to ensemble learning.
[0075] In the second option the AnLF 411 includes for each model in the group an ID which considers all the models of the group together with the Analytics ID and optionally the usage context, accuracy, and preparation cost.
[0076] The parameters that can be provided by the NWDAF AnLF 411 are listed in clause 6.2A.2 TS 23.288. The additional parameters described above for supporting Ensemble Learning may also be included.
[0077] In step 403, when a subscription for a trained set of ML models (including the EL Group Model ID, EL strategy, etc. as in the first option, or including each model ID in the group, as in the second option) associated with an Analytics ID is received, the NWDAF MTLF 412 may determine the ML model action to be taken. This may include determining whether existing trained ML Models can be used for the subscription or request, or whether triggering further training for the existing trained ML models is needed for the subscription or request.
[0078] If the NWDAF MTLF 412 determines that further training is needed, it may initiate data collection from NFs, (e.g., AMF / DCCF / ADRF), UE Applications (via AF) or OAMs as described in clause 6.2 TS 23.288.
[0079] If the second option discussed above is employed, then the MTLF 412 may at this point determine the group model ID based on each model ID received in combination with the Analytics ID and optionally considering the usage context, accuracy, and preparation cost. The group model ID can then be used to determine if further training is needed or not as discussed above.
[0080] The NWDAF MTLF 412 can optionally retrieve trained ML Models from the ADRF 413 using the group model ID.
[0081] In step 404, the NWDAF MTLF 412 may optionally store the newly generated ML models for ensemble learning by invoking theNadrf MLModelManagement StorageRequest in the repository, i.e., ADRF 413, including the EL Group Model ID. In step 405 the ADRF 413 may send aNadrf MLModelManagement StorageRequest Response message to the NWDAF MTLF 412 including the ML Model storage result indication.
[0082] In step 406, if the NWDAF AnLF 411 subscribes to EL Group Model ID associated to a (set of) Analytics ID(s), the NWDAF MTLF 412 notifies the NWDAF AnLF 411 by invoking Nnwdaf_MLModelProvision_Notify service operation. Otherwise, the NWDAF MTLF responds to the NWDAF AnLF by invokingNnwdaf MLModellnfo Request response service operation. For both case the notify or response includes a set of pair(s) of unique ML Model Identifier and ML Model Information associated with each Analytics ID requested by the AnLF 411.
[0083] The content of ML Model Information that can be provided by the NWDAF MTLF 412 is specified in clause 6.2A.2 TS 23.288 and may also include the following information related to ensemble learning, e.g., EL Group Model ID, EL strategy, EL data preparation, combination, etc. The structure and format of the EL Group Model ID and its uniqueness should already have been determined in step 403.
[0084] In step 407, when the NWDAF MTLF 412 determines that the previously provided trained EL Group Model requires re-training, the NWDAF MTLF 412 invokes the Nnwdaf_MLModelProvision_Notify service operation to notify the AnLF 411 an available re-trained ML model (including Subscription Correlation ID and the information related toensemble learning, e.g., EL Group Model ID, EL strategy, EL data preparation, combination, etc.).
[0085] When step 402 is for a subscription modification (i.e., including Subscription Correlation ID), in step 407 the NWDAF MTLF 412 may provide either a new trained EL Group Model different to the previously provided one, or a re-trained EL Group Model by invoking the Nnwdaf_MLModelProvision_Notify service operation.
[0086] The approach described above thus relates to an apparatus and method that focuses on introducing the information needed to allow the AnLF to know how to combine the multiple ML models received from the respective MTLF and assure that the ML models received are used for inference considering the training preparation effort. This is provided by the introduction of a Model Group ID as an umbrella identifier of the group, a strategy to describe how to combine multiple ML models and other information related to input data preparation per ML Model and information on creating the output data for the NWDAF service consumer.
[0087] Figure 5 illustrates an example of an analytics model training function 500 (e.g. an NWDAF MTLF) in accordance with aspects of the present disclosure. The analytics model training function 500 may include a processor 502, a memory 504, a controller 506, and a transceiver 508. The processor 502, the memory 504, the controller 506, or the transceiver 508, or various combinations thereof or various components thereof may be examples of means for performing various aspects of the present disclosure as described herein. These components may be coupled (e.g., operatively, communicatively, functionally, electronically, electrically) via one or more interfaces.
[0088] The processor 502, the memory 504, the controller 506, or the transceiver 508, or various combinations or components thereof may be implemented in hardware (e.g., circuitry). The hardware may include a processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), or other programmable logic device, or any combination thereof configured as or otherwise supporting a means for performing the functions described in the present disclosure.
[0089] The processor 502 may include an intelligent hardware device (e.g., a general- purpose processor, a DSP, a CPU, an ASIC, an FPGA, or any combination thereof). In some implementations, the processor 502 may be configured to operate the memory 504. In some other implementations, the memory 504 may be integrated into the processor 502. The processor 502 may be configured to execute computer-readable instructions stored in the memory 504 to cause the analytics model training function 500 to perform various functions of the present disclosure.
[0090] The memory 504 may include volatile or non-volatile memory. The memory 504 may store computer-readable, computer-executable code including instructions when executed by the processor 502 cause the analytics model training function 500 to perform various functions described herein. The code may be stored in a non-transitory computer- readable medium such the memory 504 or another type of memory. Computer-readable media includes both non-transitory computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. A non-transitory storage medium may be any available medium that may be accessed by a general-purpose or special-purpose computer.
[0091] In some implementations, the processor 502 and the memory 504 coupled with the processor 502 may be configured to cause the analytics model training function 500 to perform one or more of the functions described herein (e.g., executing, by the processor 502, instructions stored in the memory 504). For example, the processor 502 may support wireless communication at the analytics model training function 500 in accordance with examples as disclosed herein. The analytics model training function 500 may be considered as a first network data analytics function and configured to support a means for determining a model learning framework for interconnecting a group of machine learning models to establish a combined machine learning model; and output, to a second network data analytics function (e.g. an analytics inference function), one or more of a group identifier of the group of machine learning models and an indication of the model learning framework for interconnecting the group of machine learning models.
[0092] The controller 506 may manage input and output signals for the analytics model training function 500. The controller 506 may also manage peripherals not integrated intothe analytics model training function 500. In some implementations, the controller 506 may utilize an operating system such as iOS®, ANDROID®, WINDOWS®, or other operating systems. In some implementations, the controller 506 may be implemented as part of the processor 502.
[0093] In some implementations, the analytics model training function 500 may include at least one transceiver 508. In some other implementations, the analytics model training function 500 may have more than one transceiver 508. The transceiver 508 may represent a wireless transceiver. The transceiver 508 may include one or more receiver chains 510, one or more transmitter chains 512, or a combination thereof.
[0094] A receiver chain 510 may be configured to receive signals (e.g., control information, data, packets) over a wireless medium. For example, the receiver chain 510 may include one or more antennas for receiving the signal over the air or wireless medium. The receiver chain 510 may include at least one amplifier (e.g., a low-noise amplifier (LNA)) configured to amplify the received signal. The receiver chain 510 may include at least one demodulator configured to demodulate the receive signal and obtain the transmitted data by reversing the modulation technique applied during transmission of the signal. The receiver chain 510 may include at least one decoder for decoding the processing the demodulated signal to receive the transmitted data.
[0095] A transmitter chain 512 may be configured to generate and transmit signals (e.g., control information, data, packets). The transmitter chain 512 may include at least one modulator for modulating data onto a carrier signal, preparing the signal for transmission over a wireless medium. The at least one modulator may be configured to support one or more techniques such as amplitude modulation (AM), frequency modulation (FM), or digital modulation schemes like phase-shift keying (PSK) or quadrature amplitude modulation (QAM). The transmitter chain 512 may also include at least one power amplifier configured to amplify the modulated signal to an appropriate power level suitable for transmission over the wireless medium. The transmitter chain 512 may also include one or more antennas for transmitting the amplified signal into the air or wireless medium.
[0096] Figure 6 illustrates an example of a processor 600 in accordance with aspects of the present disclosure. The processor 600 may be an example of a processor configured toperform various operations in accordance with examples as described herein. The processor 600 may include a controller 602 configured to perform various operations in accordance with examples as described herein. The processor 600 may optionally include at least one memory 604, which may be, for example, an L1 / L2 / L3 cache. Additionally, or alternatively, the processor 600 may optionally include one or more arithmetic-logic units (ALUs) 606. One or more of these components may be in electronic communication or otherwise coupled (e.g., operatively, communicatively, functionally, electronically, electrically) via one or more interfaces (e.g., buses).
[0097] The processor 600 may be a processor chipset and include a protocol stack (e.g., a software stack) executed by the processor chipset to perform various operations (e.g., receiving, obtaining, retrieving, transmitting, outputting, forwarding, storing, determining, identifying, accessing, writing, reading) in accordance with examples as described herein. The processor chipset may include one or more cores, one or more caches (e.g., memory local to or included in the processor chipset (e.g., the processor 600) or other memory (e.g., random access memory (RAM), read-only memory (ROM), dynamic RAM (DRAM), synchronous dynamic RAM (SDRAM), static RAM (SRAM), ferroelectric RAM (FeRAM), magnetic RAM (MRAM), resistive RAM (RRAM), flash memory, phase change memory (PCM), and others).
[0098] The controller 602 may be configured to manage and coordinate various operations (e.g., signaling, receiving, obtaining, retrieving, transmitting, outputting, forwarding, storing, determining, identifying, accessing, writing, reading) of the processor 600 to cause the processor 600 to support various operations in accordance with examples as described herein. For example, the controller 602 may operate as a control unit of the processor 600, generating control signals that manage the operation of various components of the processor 600. These control signals include enabling or disabling functional units, selecting data paths, initiating memory access, and coordinating timing of operations.
[0099] The controller 602 may be configured to fetch (e.g., obtain, retrieve, receive) instructions from the memory 604 and determine subsequent instruction(s) to be executed to cause the processor 600 to support various operations in accordance with examples as described herein. The controller 602 may be configured to track memory address ofinstructions associated with the memory 604. The controller 602 may be configured to decode instructions to determine the operation to be performed and the operands involved. For example, the controller 602 may be configured to interpret the instruction and determine control signals to be output to other components of the processor 600 to cause the processor 600 to support various operations in accordance with examples as described herein. Additionally, or alternatively, the controller 602 may be configured to manage flow of data within the processor 600. The controller 602 may be configured to control transfer of data between registers, arithmetic logic units (ALUs), and other functional units of the processor 600.
[0100] The memory 604 may include one or more caches (e.g., memory local to or included in the processor 600 or other memory, such RAM, ROM, DRAM, SDRAM, SRAM, MRAM, flash memory, etc. In some implementations, the memory 604 may reside within or on a processor chipset (e.g., local to the processor 600). In some other implementations, the memory 604 may reside external to the processor chipset (e.g., remote to the processor 600).
[0101] The memory 604 may store computer-readable, computer-executable code including instructions that, when executed by the processor 600, cause the processor 600 to perform various functions described herein. The code may be stored in a non-transitory computer-readable medium such as system memory or another type of memory. The controller 602 and / or the processor 600 may be configured to execute computer-readable instructions stored in the memory 604 to cause the processor 600 to perform various functions. For example, the processor 600 and / or the controller 602 may be coupled with or to the memory 604, the processor 600, the controller 602, and the memory 604 may be configured to perform various functions described herein. In some examples, the processor 600 may include multiple processors and the memory 604 may include multiple memories. One or more of the multiple processors may be coupled with one or more of the multiple memories, which may, individually or collectively, be configured to perform various functions herein.
[0102] The one or more ALUs 606 may be configured to support various operations in accordance with examples as described herein. In some implementations, the one or moreALUs 606 may reside within or on a processor chipset (e.g., the processor 600). In some other implementations, the one or more ALUs 606 may reside external to the processor chipset (e.g., the processor 600). One or more ALUs 606 may perform one or more computations such as addition, subtraction, multiplication, and division on data. For example, one or more ALUs 606 may receive input operands and an operation code, which determines an operation to be executed. One or more ALUs 606 be configured with a variety of logical and arithmetic circuits, including adders, subtractors, shifters, and logic gates, to process and manipulate the data according to the operation. Additionally, or alternatively, the one or more ALUs 606 may support logical operations such as AND, OR, exclusive-OR (XOR), not-OR (NOR), and not- AND (NAND), enabling the one or more ALUs 606 to handle conditional operations, comparisons, and bitwise operations.
[0103] The processor 600 may support wireless communication in accordance with examples as disclosed herein. The processor 600 may be configured to or operable to support a means for providing training for a combined plurality of machine learning models in a wireless network, the processor comprising at least one controller coupled with at least one memory and configured to cause the processor to: determine a model learning framework for interconnecting a group of machine learning models to establish a combined machine learning model; and output, to a second network data analytics function, one or more of a group identifier of the group of machine learning models and an indication of the model learning framework for interconnecting the group of machine learning models.
[0104] Figure 7 illustrates an example of an analytics inference function 700 (for example an NWDAF AnLF) in accordance with aspects of the present disclosure. The analytics inference function 700 may include a processor 702, a memory 704, a controller 706, and a transceiver 708. The processor 702, the memory 704, the controller 706, or the transceiver 708, or various combinations thereof or various components thereof may be examples of means for performing various aspects of the present disclosure as described herein. These components may be coupled (e.g., operatively, communicatively, functionally, electronically, electrically) via one or more interfaces.
[0105] The processor 702, the memory 704, the controller 706, or the transceiver 708, or various combinations or components thereof may be implemented in hardware (e.g.,circuitry). The hardware may include a processor, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), or other programmable logic device, or any combination thereof configured as or otherwise supporting a means for performing the functions described in the present disclosure.
[0106] The processor 702 may include an intelligent hardware device (e.g., a general- purpose processor, a DSP, a CPU, an ASIC, an FPGA, or any combination thereof). In some implementations, the processor 702 may be configured to operate the memory 704. In some other implementations, the memory 704 may be integrated into the processor 702. The processor 702 may be configured to execute computer-readable instructions stored in the memory 704 to cause the analytics inference function 700 to perform various functions of the present disclosure.
[0107] The memory 704 may include volatile or non-volatile memory. The memory 704 may store computer-readable, computer-executable code including instructions when executed by the processor 702 cause the analytics inference function 700 to perform various functions described herein. The code may be stored in a non-transitory computer- readable medium such the memory 704 or another type of memory. Computer-readable media includes both non-transitory computer storage media and communication media including any medium that facilitates transfer of a computer program from one place to another. A non-transitory storage medium may be any available medium that may be accessed by a general-purpose or special-purpose computer.
[0108] In some implementations, the processor 702 and the memory 704 coupled with the processor 702 may be configured to cause the analytics inference function 700 to perform one or more of the functions described herein (e.g., executing, by the processor 702, instructions stored in the memory 704). For example, the processor 702 may support wireless communication at the analytics inference function 700 in accordance with examples as disclosed herein. The analytics inference function 700 may be configured to support a means for determining a group of machine learning models to be used in combination with each other; and sending to an analytics model training function a request for a unified model formed from a group of machine learning models, the request includingat least one of a group identifier of the group of machine learning models, an identifier of each machine learning model in the group and an analytics service identifier.
[0109] The controller 706 may manage input and output signals for the analytics inference function 700. The controller 706 may also manage peripherals not integrated into the analytics inference function 700. In some implementations, the controller 706 may utilize an operating system such as iOS®, ANDROID®, WINDOWS®, or other operating systems. In some implementations, the controller 706 may be implemented as part of the processor 702.
[0110] In some implementations, the analytics inference function 700 may include at least one transceiver 708. In some other implementations, the analytics inference function 700 may have more than one transceiver 708. The transceiver 708 may represent a wireless transceiver. The transceiver 708 may include one or more receiver chains 710, one or more transmitter chains 712, or a combination thereof.
[0111] A receiver chain 710 may be configured to receive signals (e.g., control information, data, packets) over a wireless medium. For example, the receiver chain 710 may include one or more antennas for receive the signal over the air or wireless medium. The receiver chain 710 may include at least one amplifier (e.g., a low-noise amplifier (LN A)) configured to amplify the received signal. The receiver chain 710 may include at least one demodulator configured to demodulate the receive signal and obtain the transmitted data by reversing the modulation technique applied during transmission of the signal. The receiver chain 710 may include at least one decoder for decoding the processing the demodulated signal to receive the transmitted data.
[0112] A transmitter chain 712 may be configured to generate and transmit signals (e.g., control information, data, packets). The transmitter chain 712 may include at least one modulator for modulating data onto a carrier signal, preparing the signal for transmission over a wireless medium. The at least one modulator may be configured to support one or more techniques such as amplitude modulation (AM), frequency modulation (FM), or digital modulation schemes like phase-shift keying (PSK) or quadrature amplitude modulation (QAM). The transmitter chain 712 may also include at least one power amplifier configured to amplify the modulated signal to an appropriate power level suitablefor transmission over the wireless medium. The transmitter chain 712 may also include one or more antennas for transmitting the amplified signal into the air or wireless medium.
[0113] Figure 8 illustrates a flowchart of a method in accordance with aspects of the present disclosure. The operations of the method may be implemented by an analytics model training function as described herein. In some implementations, the analytics model training function may execute a set of instructions to control the function elements of the analytics model training function to perform the described functions.
[0114] At 802, the method may include determining a model learning framework for interconnecting a group of machine learning models to establish a combined machine learning model. The operations of 802 may be performed in accordance with examples as described herein. In some implementations, aspects of the operations of 802 may be performed by an analytics model training function as described with reference to Figure 5.
[0115] At 804, the method may include sending to an analytics inference function one or more of a group identifier of the group of machine learning models and an indication of the model learning framework. The operations of 804 may be performed in accordance with examples as described herein. In some implementations, aspects of the operations of 804 may be performed by an analytics training function as described with reference to Figure 5.
[0116] It should be noted that the method described herein describes a possible implementation, and that the operations and the steps may be rearranged or otherwise modified and that other implementations are possible.
[0117] Figure 9 illustrates a flowchart of a method in accordance with aspects of the present disclosure. The operations of the method may be implemented by an analytics inference function as described herein. In some implementations, the analytics inference function may execute a set of instructions to control the function elements of the analytics inference function to perform the described functions.
[0118] At 902, the method may include determining a group of machine learning models to be used in combination with each other. The operations of 902 may be performed in accordance with examples as described herein. In some implementations, aspects of theoperations of 902 may be performed by an analytics inference function as described with reference to Figure 7.
[0119] At 904, the method may include sending to an analytics model training function a request for a unified model formed from a group of machine learning models, the request including at least one of a group identifier of the group of machine learning models, an identifier of each machine learning model in the group and an analytics service identifier. The operations of 904 may be performed in accordance with examples as described herein. In some implementations, aspects of the operations of 904 may be performed by an analytics inference function as described with reference to Figure 7.
[0120] It should be noted that the method described herein describes a possible implementation, and that the operations and the steps may be rearranged or otherwise modified and that other implementations are possible.
[0121] The description herein is provided to enable a person having ordinary skill in the art to make or use the disclosure. Various modifications to the disclosure will be apparent to a person having ordinary skill in the art, and the generic principles defined herein may be applied to other variations without departing from the scope of the disclosure. Thus, the disclosure is not limited to the examples and designs described herein but is to be accorded the broadest scope consistent with the principles and novel features disclosed herein.
Claims
CLAIMSWhat is claimed is:
1. A first network data analytics function comprising code for network analytics, the code comprising instructions executable by the first network data analytics function to: determine a model learning framework for interconnecting a group of machine learning models to establish a combined machine learning model; and output, to a second network data analytics function, one or more of a group identifier of the group of machine learning models and an indication of the model learning framework for interconnecting the group of machine learning models.
2. The first network data analytics function of claim 1 , wherein the instructions are further executable by the first network data analytics function to: determine the group of machine learning models based on one or more of a corresponding identifier associated with each machine learning model of the group of machine learning models, an analytics identifier, a context associated with the group of machine learning models, an accuracy indication, and a cost indication.
3. The first network data analytics function of claim 2, wherein the context comprises one or more of an application context, a network topology, a network hardware, a software characteristic, a network load level, a network energy level, or a timing.
4. The first network data analytics function of claim 1, 2 or 3, wherein the instructions are further executable by the first network data analytics function to receive the group identifier from the second network data analytics function and use the group identifier to determine the group of machine learning models and / or each machine learning model to be used in combination with each other.
5. The first network data analytics function of any preceding claim, wherein the indication of the model learning framework includes a private field enabling the indication of a specific framework not selected from a standardized list.
6. The first network data analytics function of any of claims 1 to 4, wherein the indication of the model learning framework includes a public field chosen from a predefined lists of framework names.
7. The first network data analytics function of any preceding claim, wherein the indication of the model learning framework includes an indication of the role of each individual machine learning model in the group.
8. The first network data analytics function of any preceding claim, wherein the instructions are further executable by the first network data analytics function to output to the second network analytics function an indication of a preparation method of input data for input into each machine learning model in the group.
9. The first network data analytics function of claim 8, wherein the preparation of input data includes a logic for producing an input data set based on one or more of the following: a random distribution logic to prepare the input data set; a repetitive process logic that uses the same input data set; a weight-based logic to prepare the input data set; and an algorithmic or model-based logic to prepare the input data set.
10. The first network data analytics function of any preceding claim, wherein the instructions are further executable by the first network data analytics function to output to the second network analytics function an indication of consolidation of output data of each machine learning model in the group to create a unified output.
11. The first network data analytics function of claim 10, wherein the consolidation of output data includes a logic for producing a unified output based on one or more of the following: a weighted combination logic; an algorithm or model logic; a voting logic; and an aggregation logic.
12. The first network data analytics function of any preceding claim, wherein instructions are further executable by the first network data analytics function to store and retrieve multiple machine learning models from a repository function in conjunction with at least one of a group identifier, an analytics identifier, a use case context, an accuracy indicator and a cost indicator.
13. The first network data analytics function of any preceding claim configured to operate as an analytics model training function for providing training for a combined plurality of machine learning models, the first network data analytics function further comprising: at least one memory: and at least one processor coupled with the at least one memory and configured to cause the first network data function to execute the instructions.
14. A second network analytics function for use in a wireless network, wherein an analytics output is derived based on a combined machine learning model established from a plurality of machine learning models, the second network analytics function comprising: at least one memory; and at least one processor coupled with the at least one memory and configured to cause the second network analytics function to: determine a group of machine learning models to be used in combination with each other; and send to a first network analytics function a request for a unified model formed from a group of machine learning models, the request including at least one of a group identifier of the group of machine learning models, an identifier of each machine learning model in the group and an analytics service identifier, accuracy indicator and a cost indicator.
15. The second network analytics function of claim 14, wherein the processor is further configured to cause the second network analytics function to discover and select the first network analytics function from which to request the combined machine learningmodel based on at least one of a group model identifier, an identifier of each ML model in the group, analytics service identifier, a usage context, accuracy, and cost.
16. The second network analytics function of claim 14 or 15, wherein the processor is further configured to cause the second network analytics function to receive from the first network analytics function the group identifier of the group of machine learning models and an indication of a model learning framework for interconnecting the group of machine learning models.
17. The second network analytics function of claim 14, 15 or 16, wherein the processor is further configured to cause the second network analytics function to send to the first network analytics function an indication that the second network analytics function supports ensemble learning.
18. The second network analytics function of any of claims 14 to 17, which second network analytics function is an analytics inference function, and wherein the first network analytics function is an analytics model training function.
19. A method performed by an analytics model training function that provides training from a plurality of ML models, the method comprising: determining a model learning framework for interconnecting a group of machine learning models establish a combined machine learning model and sending to an analytics inference function one or more of a group identifier of the group of machine learning models and an indication of the model learning framework.
20. A processor in an analytics model training function for providing training for a combined plurality of machine learning models in a wireless network, the processor comprising at least one controller coupled with at least one memory and configured to cause the processor to: determine a model learning framework for interconnecting a group of machine learning models to establish a combined machine learning model; andoutput, to a second network data analytics function, one or more of a group identifier of the group of machine learning models and an indication of the model learning framework for interconnecting the group of machine learning models.
Citation Information
Patent Citations
Machine learning (ML) model management in 5g core network
WO2023099970A1
