User equipment data privacy preservation during machine learning model training

Abstract

Techniques are disclosed for preserving privacy for user equipment data during machine learning model training in a communication network environment. In one example, from a user equipment perspective, a method comprises sending, from the user equipment, a selected privacy level to a communication network with which the user equipment is participating in a machine learning model training process. The method further comprises receiving, at the user equipment, one or more privacy parameters corresponding to the selected privacy level. The method further comprises applying, by the user equipment, the one or more privacy parameters to data to be sent to the communication network. In addition to the user equipment-side solution, a communication network-side solution is provided for preserving privacy for user equipment data.

Description

[0001] USER EQUIPMENT DATA PRIVACY PRESERVATION DURING MACHINE LEARNING MODEL TRAINING

[0002] Field

[0003] The field relates generally to communication networks, and more particularly, but not exclusively, to security management in such communication networks.

[0004] Background

[0005] This section introduces aspects that may be helpful in facilitating a better understanding of the inventions. Accordingly, the statements of this section are to be read in this light and are not to be understood as admissions about what is in the prior art or what is not in the prior art.

[0006] Fourth generation (4G) wireless mobile telecommunications technology, also known as Long Term Evolution (LTE) technology, was designed to provide high-capacity mobile multimedia with high data rates particularly for human interaction. Next generation or fifth generation (5G) technology is intended to be used not only for human interaction, but also for machine type communications in so-called Internet of Things (loT) networks.

[0007] While 5G networks are intended to enable massive loT services (e.g., very large numbers of limited capacity devices) and mission-critical loT services (e.g., requiring high reliability), improvements over legacy mobile communication services are supported in the form of enhanced mobile broadband (eMBB) services providing improved wireless Internet access for mobile devices.

[0008] In an example communication system, user equipment (5G UE in a 5G network or, more broadly, a UE) such as a mobile terminal (subscriber) communicates over an air interface with a base station or access point of an access network referred to as a 5G AN in a 5G network. The access point (e.g., gNB) is illustratively part of an access network of the communication system.

[0009] For example, in a 5G network, the access network referred to as a 5G AN is described in 5G Technical Specification (TS) 23.501, entitled “Technical Specification Group Services and System Aspects; System Architecture for the 5G System,” and TS 23.502, entitled “Technical Specification Group Services and System Aspects; Procedures for the 5G System (5GS),” the disclosures of which are incorporated by reference herein in their entireties. In general, the access point (e.g., gNB) provides access for the UE to a core network (CN or 5GC), which then provides access for the UE to other UEs and / or a data network such as a packet data network (e.g., Internet).

[0010] TS 23.501 goes on to define a 5G Service-Based Architecture (SBA) which models services as network functions (NFs) that communicate with each other using representational state transfer application programming interfaces (Restful APIs).

[0011] Furthermore, TS 33.501, entitled “Technical Specification Group Services and System Aspects; Security Architecture and Procedures for the 5G System,” the disclosure of which is incorporated by reference herein in its entirety, further describes security management details associated with a 5G network.

[0012] Security management is an important consideration in any communication network environment. However, due to continuing attempts to improve the architectures and protocols associated with a 5G network in order to increase network efficiency and / or subscriber convenience, security management issues associated with data exchanged in the communication network environment can present a significant challenge. For example, implementing data privacy algorithms in the communication network environment is a technical challenge.

[0013] Summary

[0014] Illustrative embodiments provide techniques for preserving privacy for user equipment data during machine learning model training in a communication network environment.

[0015] In one illustrative embodiment, from a user equipment perspective, a method comprises sending, from the user equipment, a selected privacy level to a communication network with which the user equipment is participating in a machine learning model training process. The method further comprises receiving, at the user equipment, one or more privacy parameters corresponding to the selected privacy level. The method further comprises applying, by the user equipment, the one or more privacy parameters to data to be sent to the communication network.

[0016] In another illustrative embodiment, from a communication network entity perspective, a method comprises receiving, at the communication network entity, a selected privacy level from a user equipment with which the communication network entity is participating in a machine learning model training process. The method further comprises determining, by the communication network entity, one or more privacy parameters corresponding to the selected privacy level.

[0017] In yet another illustrative embodiment, from a machine learning model training management function perspective, a method comprises receiving, at the machine learning model training management function, one or more privacy parameters, from a communication network entity, corresponding to a selected privacy level associated with a user equipment participating in a machine learning model training process. The method further comprises receiving, at the machine learning model training management function, data from the user equipment. The method further comprises applying, by the machine learning model training management function, the one or more privacy parameters to the data received from the user equipment.

[0018] Further illustrative embodiments are provided in the form of a non-transitory computer readable medium having embodied therein executable program code that when executed by a processor causes the processor to perform the above and / or other steps, operations, and the like. Still further illustrative embodiments comprise an apparatus with a processor and a memory configured to perform the above and / or other steps, operations, and the like. Some illustrative embodiments comprise a system configured to perform the above and / or other steps, operations, and the like. Further, some illustrative embodiments comprise an apparatus or a system comprising means for performing the above and / or other steps, operations, and the like.

[0019] Advantageously, illustrative embodiments provide user equipment data privacy preserving techniques during various machine learning model training scenarios wherein the leakage of privacy sensitive data might occur.

[0020] These and other features and advantages of embodiments described herein will become more apparent from the accompanying drawings and the following detailed description.

[0021] Brief Description of the Drawings

[0022] FIG. 1 illustrates a communication network environment with which one or more illustrative embodiments may be implemented.

[0023] FIG. 2 illustrates user equipment and entities with which one or more illustrative embodiments may be implemented. FIG. 3 illustrates a procedure for preserving privacy of user equipment data during training of a machine learning model in a communication network environment according to an illustrative embodiment.

[0024] FIG. 4 illustrates a procedure for preserving privacy of user equipment data during training of a machine learning model in a communication network environment according to another illustrative embodiment.

[0025] Detailed Description

[0026] Embodiments will be illustrated herein in conjunction with example communication systems and associated techniques for security management in communication systems. It should be understood, however, that the scope of the claims is not limited to particular types of communication systems and / or processes disclosed. Embodiments can be implemented in a wide variety of other types of communication systems, using alternative processes and operations. For example, although illustrated in the context of wireless cellular systems utilizing the 3rd Generation Partnership Project (3GPP) system elements such as a 3GPP next generation system (5G), the disclosed embodiments can be adapted in a straightforward manner to a variety of other types of communication systems such as 6G communication systems.

[0027] In accordance with illustrative embodiments implemented in a 5G communication system environment, one or more 3GPP technical specifications (TS) and technical reports (TR) may provide further explanation of network elements / functions and / or operations that may interact with parts of the inventive solutions, e.g., the above-referenced 3GPP TS 23.501, TS 23.502, and TS 33.501. Other 3GPP TS / TR documents may provide other details that one of ordinary skill in the art will realize, for example, 3GPP TS 23.288, entitled “Technical Specification Group Services and System Aspects; Architecture Enhancements for 5G System (5GS) to Support Network Data Analytics Services,” the disclosure of which is incorporated by reference herein in its entirety. Note that 3GPP TS / TR documents are non-limiting examples of communication network standards (e.g., specifications, procedures, reports, requirements, recommendations, and the like). However, while well-suited for 5G-related 3GPP standards, embodiments are not necessarily intended to be limited to any particular standards.

[0028] It is to be understood that the term 5G network, and the like (e.g., 5G system, 5G communication system, 5G environment, 5G communication environment etc.), in some illustrative embodiments, may be understood to comprise all or part of an access network and all or part of a core network. However, the term 5G network, and the like, may also occasionally be used interchangeably herein with the term 5GC network, and the like, without any loss of generality, since one of ordinary skill in the art understands any distinctions.

[0029] Prior to describing illustrative embodiments, a general description of certain main components of a 5G network will be described below in the context of FIGS. 1 and 2.

[0030] FIG. 1 shows a communication system 100 within which illustrative embodiments are implemented. It is to be understood that the elements shown in communication system 100 are intended to represent some main functions provided within the system, e.g., control plane functions, user plane functions, etc. As such, the blocks shown in FIG. 1 reference specific elements in 5G networks that provide some of these main functions. However, other network elements may be used to implement some or all of the main functions represented. Also, it is to be understood that not all functions of a 5G network are depicted in FIG. 1. Rather, at least some functions that facilitate an explanation of illustrative embodiments are represented. Subsequent figures may depict some additional elements / functions (i.e., network entities).

[0031] Accordingly, as shown, communication system 100 comprises user equipment (UE) 102 that communicates via an air interface 103 with an access point 104. It is to be understood that UE 102 may use one or more other types of access points (e.g., access functions, networks, etc.) to communicate with the 5GC network other than a gNB. By way of example only, the access point 104 may be any 5G access network (gNB), an untrusted non-3GPP access network that uses an Non-3GPP Interworking Function (N3IWF), a trusted non-3GPP network that uses a Trusted Non-3GPP Gateway Function (TNGF) or wireline access that uses a Wireline Access Gateway Function (W-AGF) or may correspond to a legacy access point (e.g., eNB). Furthermore, access point 104 may be a wireless local area network (WLAN) access point as will be further explained in illustrative embodiments described herein.

[0032] The UE 102 may be a mobile station, and such a mobile station may comprise, by way of example, a mobile telephone, a computer, an loT device, or any other type of communication device. The term “user equipment” as used herein is therefore intended to be construed broadly, so as to encompass a variety of different types of mobile stations, subscriber stations or, more generally, communication devices, including examples such as a combination of a data card inserted in a laptop or other equipment such as a smart phone. Such communication devices are also intended to encompass devices commonly referred to as access terminals. In one illustrative embodiment, UE 102 is comprised of a Universal Integrated Circuit Card (UICC) part and a Mobile Equipment (ME) part. The UICC is the user-dependent part of the UE and contains at least one Universal Subscriber Identity Module (USIM) and appropriate application software. The USIM securely stores a permanent subscription identifier and its related key, which are used to uniquely identify and authenticate subscribers to access networks. The ME is the user-independent part of the UE and contains terminal equipment (TE) functions and various mobile termination (MT) functions. Alternative illustrative embodiments may not use UICC-based authentication, e.g., a Non-Public (Private) Network (NPN).

[0033] Note that, in one example, the permanent subscription identifier is an International Mobile Subscriber Identity (IMSI) unique to the UE. In one embodiment, the IMSI is a fixed 15 -digit length and consists of a 3 -digit Mobile Country Code (MCC), a 3 -digit Mobile Network Code (MNC), and a 9-digit Mobile Station Identification Number (MSIN). In a 5G communication system, an IMSI is referred to as a Subscription Permanent Identifier (SUPI). In the case of an IMSI as a SUPI, the MSIN provides the subscriber identity. Thus, only the MSIN portion of the IMSI typically needs to be encrypted. The MNC and MCC portions of the IMSI provide routing information, used by the serving network to route to the correct home network. When the MSIN of a SUPI is encrypted, it is referred to as Subscription Concealed Identifier (SUCI). Another example of a SUPI uses a Network Access Identifier (NAI). NAI is typically used for loT communication.

[0034] The access point 104 is illustratively part of a radio access network or RAN of the communication system 100. Such a radio access network may comprise, for example, a 5G System having a plurality of base stations. Components of a radio access network may, more generally, be considered “radio access entities.”

[0035] Further, the access point 104 in this illustrative embodiment is operatively coupled to an Access and Mobility Management Function (AMF / SEAF) 106. In a 5G network, the AMF / SEAF supports, inter alia, mobility management (MM) and security anchor (SEAF) functions.

[0036] AMF / SEAF 106 in this illustrative embodiment is operatively coupled to (e.g., uses the services of) other network functions 108. As shown, some of these other network functions 108 include, but are not limited to, an Authentication Server Function (AUSF), a Unified Data Management (UDM) function, an Application Function (AF), and a Network Data Analytics Function (NWDAF). These listed network function examples are typically implemented in the home network of the UE subscriber, further explained below. Note that, in a 5GC network, the 4G function of the HSS (home subscriber server) is split into the AUSF, UDM, and a Unified Data Repository (UDR, not expressly shown) functions. Typically, AUSF authenticates UEs and provides any needed cryptographic keys, while UDR stores the user data and UDM manages the user data. The AF exposes the application layer for interaction with 5G NFs and network resources. The NWDAF is a 5G network function that collects data from various 5GC network functions, application functions, as well as operations, administration, and management (GAM) systems, and operational support systems. The NWDAF is configured to facilitate the way 5GC data is produced and consumed, as well as to generate analytical insights and take actions based on the analytical insights. It is also realized that third-party applications can also be enabled to operate with one or more network functions in the 5GC.

[0037] Other network functions 108 may include network functions that can act as service producers (NFp) and / or service consumers (NFc). Note that any network function can be a service producer for one service and a service consumer for another service. Further, when the service being provided includes data, the data-providing NFp is referred to as a data producer, while the data-requesting NFc is referred to as a data consumer. A data producer may also be an NF that generates data by modifying or otherwise processing data produced by another NF. Note that NFs may, more generally, be considered “network entities.”

[0038] Note that a UE, such as UE 102, is typically subscribed to what is referred to as a Home Public Land Mobile Network (HPLMN) in which some or all of the functions 106 and 108 reside. Alternatively the UE, such as UE 102, may receive services from an NPN where these functions may reside. The HPLMN is also referred to as the Home Environment (HE). If the UE is roaming (not in the HPLMN), it is typically connected with a Visited Public Land Mobile Network (VPLMN) also referred to as a visited network, while the network that is currently serving the UE is also referred to as a serving network. In the roaming case, some of the functions 106 and 108 can reside in the VPLMN, in which case, functions in the VPLMN communicate with functions in the HPLMN as needed. However, in a non-roaming scenario, access and mobility management functions 106 and the other network functions 108 reside in the same communication network, i.e., HPLMN. Embodiments described herein, unless otherwise specified, are not necessarily limited by which functions reside in which PLMN (i.e., HPLMN or VPLMN). The access point 104 is also operatively coupled (via one or more of functions 106 and / or 108) to a Session Management Function (SMF) 110, which is operatively coupled to a User Plane Function (UPF) 112. UPF 112 is operatively coupled to a Packet Data Network, e.g., Internet 114. Note that the thicker solid lines in this figure denote a user plane (UP) of the communication network, as compared to the thinner solid lines that denote a control plane (CP) of the communication network. It is to be appreciated that network (e.g., Internet) 114 in FIG. 1 may additionally or alternatively represent other network infrastructures including, but not limited to, cloud computing infrastructure and / or edge computing infrastructure. Further typical operations and functions of such network elements are not described here since they are not the focus of the illustrative embodiments and may be found in appropriate 3GPP 5G documentation. Note that functions shown in 106, 108, 110 and 112 are examples of network functions (NFs).

[0039] It is to be appreciated that this particular arrangement of system elements is an example only, and other types and arrangements of additional or alternative elements can be used to implement a communication system in other embodiments. For example, in other embodiments, the communication system 100 may comprise other elements / functions not expressly shown herein.

[0040] Accordingly, the FIG. 1 arrangement is just one example configuration of a wireless cellular system, and numerous alternative configurations of system elements may be used. For example, although only single elements / functions are shown in the FIG. 1 embodiment, this is for simplicity and clarity of description only. A given alternative embodiment may of course include larger numbers of such system elements, as well as additional or alternative elements of a type commonly associated with conventional system implementations.

[0041] It is also to be noted that while FIG. 1 illustrates system elements as singular functional blocks, the various subnetworks that make up the 5G network are partitioned into so-called network slices. Network slices (network partitions) are logical networks that provide specific network capabilities and network characteristics that can support a corresponding service type, optionally using network function virtualization (NFV) on a common physical infrastructure. With NFV, network slices are instantiated as needed for a given service, e.g., eMBB service, massive loT service, and mission-critical loT service. A network slice or function is thus instantiated when an instance of that network slice or function is created. In some embodiments, this involves installing or otherwise running the network slice or function on one or more host devices of the underlying physical infrastructure. UE 102 is configured to access one or more of these services via access point 104.

[0042] FIG. 2 is a block diagram illustrating computing architectures for various participants in methodologies according to illustrative embodiments. More particularly, system 200 is shown comprising user equipment (UE) 202 and a plurality of entities 204-1, . . . . , 204-N. For example, in illustrative embodiments and with reference back to FIG. 1, UE 202 can represent UE 102, while entities 204-1, . . . , 204-N can represent functions 106 and 108 (i.e., network entities such as, but not limited to, AMF, NWDAF, UDM), as well as access point 104 (i.e., radio access entity such as, but not limited to, a RAN node or gNB). It is to be appreciated that the UE 202 and entities 204-1, . . . . , 204-N are configured to interact to provide security management and other techniques described herein.

[0043] The user equipment 202 comprises a processor 212 coupled to a memory 216 and interface circuitry 210. The processor 212 of the user equipment 202 includes a security management processing module 214 that may be implemented at least in part in the form of software executed by the processor. The security management processing module 214 performs security management described in conjunction with subsequent figures and otherwise herein. The memory 216 of the user equipment 202 includes a security management storage module 218 that stores data generated or otherwise used during security management operations.

[0044] Each of the entities (individually or collectively referred to herein as 204) comprises a processor 222 (222-1, . . . , 222-N) coupled to a memory 226 (226-1, . . . , 226-N) and interface circuitry 220 (220-1, . . . , 220-N). Each processor 222 of each entity 204 includes a security management processing module 224 (224-1, . . . , 224-N) that may be implemented at least in part in the form of software executed by the processor 222. The security management processing module 224 performs security management operations described in conjunction with subsequent figures and otherwise herein. Each memory 226 of each entity 204 includes a security management storage module 228 (228-1, . . . , 228-N) that stores data generated or otherwise used during security management operations.

[0045] The processors 212 and 222 may comprise, for example, microprocessors such as central processing units (CPUs), application-specific integrated circuits (ASICs), digital signal processors (DSPs) or other types of processing devices, as well as portions or combinations of such elements. The memories 216 and 226 may be used to store one or more software programs that are executed by the respective processors 212 and 222 to implement at least a portion of the functionality described herein. For example, security management operations and other functionality as described in conjunction with subsequent figures and otherwise herein may be implemented in a straightforward manner using software code executed by processors 212 and 222.

[0046] A given one of the memories 216 and 226 may therefore be viewed as an example of what is more generally referred to herein as a computer program product or still more generally as a computer or processor readable (non-transitory or storage) medium that has executable program code embodied therein. Other examples of computer or processor readable media may include disks or other types of magnetic or optical media, in any combination. Illustrative embodiments can include articles of manufacture comprising such computer program products or other computer or processor readable media.

[0047] Further, the memories 216 and 226 may more particularly comprise, for example, electronic random- access memory (RAM) such as static RAM (SRAM), dynamic RAM (DRAM) or other types of volatile or non-volatile electronic memory. The latter may include, for example, non-volatile memories such as flash memory, magnetic RAM (MRAM), phasechange RAM (PC-RAM) or ferroelectric RAM (FRAM). The term “memory” as used herein is intended to be broadly construed, and may additionally or alternatively encompass, for example, a read-only memory (ROM), a disk-based memory, or other type of storage device, as well as portions or combinations of such devices.

[0048] The interface circuitries 210 and 220 illustratively comprise transceivers or other communication hardware or firmware that allows the associated system elements to communicate with one another in the manner described herein.

[0049] It is apparent from FIG. 2 that user equipment 202 and plurality of entities 204 are configured for communication with each other as security management participants via their respective interface circuitries 210 and 220. This communication involves each participant sending data to and / or receiving data from one or more of the other participants. The term “data” as used herein is intended to be construed broadly, so as to encompass any type of information that may be sent between participants including, but not limited to, identity data, key pairs, key indicators, tokens, secrets, security management messages, registration request / response messages and data, request / response messages, authentication request / response messages and data, metadata, control data, audio, video, multimedia, consent data, other messages, etc.

[0050] It is to be appreciated that the particular arrangement of components shown in FIG. 2 is an example only, and numerous alternative configurations may be used in other embodiments. For example, any given network element / function and / or access point can be configured to incorporate additional or alternative components and to support other communication protocols.

[0051] Other system elements such as access point 104, SMF 110, and UPF 112 may each be configured to include components such as a processor, memory and network interface. Also, entities such as third-party applications and network operators can participate in methodologies described herein via computing devices configured to include components such as a processor, memory and network interface. These elements and devices need not be implemented on separate stand-alone processing platforms, but could instead, for example, represent different functional portions of a single common processing platform.

[0052] More generally, FIG. 2 can be considered to represent processing devices configured to provide respective security management functionalities and operatively coupled to one another in a communication system. By way of example only, all or parts of each of UE 202 and the plurality of entities 204 (e.g., processor and memory) can be considered examples of means for performing one or more operations, one or more steps, one or more functions, one or more processes, etc. as described herein.

[0053] As mentioned above, the 3GPP TS 23.501 defines the 5GC network architecture as service-based, e.g., Service-Based Architecture (SBA). It is realized herein that in deploying different NFs, there can be many situations where an NF may need to interact with an entity external to the SBA-based 5GC network (e.g., including the corresponding PLMN(s), e.g., HPLMN and VPLMN). Thus, the term “internal” as used herein illustratively refers to operations and / or communications within the SBA-based 5GC network (e.g., SBA-based interfaces) and the term “external” illustratively refers to operations and / or communications outside the SBA-based 5GC network (non-SBA interfaces).

[0054] Given the above general description of some features of a 5GC network, problems with existing security approaches in preserving the privacy of UE data during training of a machine learning (ML) model in a communication network environment, and solutions proposed in accordance with illustrative embodiments, will now be described herein below. While not limited thereto, communication network environments can employ data privacy mechanisms such as differential privacy. Differential privacy provides guarantees on the privacy of the data in exchange for reducing the accuracy of a utility function that needs the data. Randomizing a function of the data (functional perturbation), the pure raw randomization of the data (input perturbation) or randomizing the output of a function (output perturbation), are among the widely used differential privacy mechanisms. It is realized herein that the differential privacy mechanisms could be improved if it was known which function is to be computed. In other words, it is realized herein that having the knowledge of the function can improve both utility and privacy.

[0055] Differential privacy was originally proposed to give a formal guarantee of privacy in case commonly used anonymization techniques (such as k-anonymity and 1-divergence) are not sufficient. Differential privacy is a technique of such a statistical nature that once an algorithm or data is protected with differential privacy, no matter how the output of the algorithm or the data is further processed, the privacy of individuals in the data set stays protected. The original motivation for proposing this strong definition of privacy came from the observation that combining data from different sources can break privacy. Nowadays, for example, the United States Census Bureau has adopted differential privacy in privacy protection and several major customer data-driven companies use it. Differential privacy can also help to comply with data privacy regulations such as the General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA).

[0056] A differential privacy guarantee provides a mathematical upper bound for the risk of leaking information about an individual and allows a user to quantify a precise level of acceptable individual level risk. Differential privacy is described using probability theory and mathematically the definition is given as follows. Assume that two data sets X and Y are neighbors if you get one by replacing a single data entry in the other. Let e > 0 and 8 I [0,1] (e and 8 are the differential privacy parameters). A randomized function M(X) : XN-> Jk is (e, 5) -DP (DP standing for differential privacy here) or if for every pair of neighboring data sets X and Y and for every outcome S c R we have: p(M(x) e s) < e£P(M(y) e s) + 6.

[0057] The parameter 8 can be interpreted as the probability of a total data breach, and the smaller the s, the less probable it is to tell whether the output originated from data sets X or Y, i.e., the less probable it is to notice the presence of any individual and the more privacypreserving is the mechanism.

[0058] There is a fundamental relationship in some mechanisms of differential privacy with the number of data set entries and the noise variance needed to provide a certain level of (E, 6). The relationship raises a concept referred to as the small data size problem, wherein the more data, the less noise is needed to protect an individual’s privacy. This problem appears in the federated learning setting, where N sites each with D number of data points are publishing functions of their data set. If each user adds noise proportional to its own data set size to protect its privacy with (e, 5), the aggregated final data set at the network side will end up with a noise variance N time larger than the case where all of the YnDn- ND was at one central point with (e, 5) -DP. There are some techniques developed to collect this data without compromising privacy and security of the data set; however, most of them either assume a trusted third party or they use homomorphic encryption, which is very computationally resource hungry.

[0059] Moreover, it is further realized herein that using differential privacy in wireless networks is beneficial. For example, differential privacy can be used to protect the signaling that is required for communication by the UEs that are participating in a NWDAF and a third- party artificial intelligence / machine learning ( A I / M L) engine or neural network model trainer. More particularly, model training via an NWDAF model training logical function (MTLF) has been described in the above-referenced TS 23.288. Furthermore, there have been proposals for cross domain machine learning, wherein the UE and RAN and / or the UE and 5G Core collaboratively train an ML model. It is thus realized that such cross-domain ML proposals have a direct impact to the UE data privacy both with respect to training data sharing and model training. More particularly, there is well established problem of privacy leakage during an ML model training process.

[0060] Illustrative embodiments address the above and other technical challenges and / or technical deficiencies in existing UE data privacy approaches in a communication network environment by providing UE privacy preserving techniques during various ML model training scenarios wherein the leakage of privacy sensitive data might occur.

[0061] Examples of some scenarios wherein illustrative embodiments can be implemented will now be described. However, it is to be understood that UE data privacy preserving techniques described herein are not limited to these example scenarios and may be applied in other scenarios as appropriate. For example, in a first scenario, assume a UE trains an ML model using its privacy sensitive data and provides the trained ML model to the RAN node (e.g., gNB) or to 5G Core, and a malicious or curious entity may try to reconstruct the training data that the UE has used while model training.

[0062] In a second scenario, assume the UE trains the ML model partially (in the case of joint ML training between the UE and the 5G Core) and sends only model parameters to the communication network. Even then, the 5G Core network can with certain probability discover the training data that has been used by the UE using the model parameters received. This scenario is also applicable in the case of federated learning.

[0063] In a third scenario, the UE sends the raw training data to the RAN node (e.g., gNB) or the 5G Core network to be used for model training.

[0064] Illustrative embodiments address the above-described scenarios and other privacy issues with various solutions: (i) a network-side solution wherein the privacy mechanism is applied at the communication network; and (ii) a UE-side solution wherein the privacy mechanism is applied at the UE itself.

[0065] As will be further illustrated in the context of FIG. 3, in the network-side solution, the UE first defines the level of privacy it needs. The privacy level can alternatively or additionally be set by the network operator based upon regulatory requirements or based upon a category of the ML model training. The network then translates this privacy requirement to the actual privacy configuration parameters needed by the implemented privacy mechanism. For instance, if the system uses differential privacy as the mechanism, then the level of privacy translates into corresponding values of s and 8, as explained above. Once the privacy values are defined, then it is either pushed to the 5G Core ML training management NF and / or to the RAN node performing model training, such that when the privacy sensitive data of the UE is received, the RAN node / 5G Core applies the privacy configuration before initiating the model training process.

[0066] As will be further illustrated in the context of FIG. 4, in the UE-side solution, these privacy configurations are directly pushed to the UE by the 5G Core network, and the privacy application occurs at the UE side itself before sending the data or the ML model toward the 5G Core network.

[0067] FIG. 3 illustrates a procedure 300 for preserving privacy of user equipment data during training of a machine learning model in a communication network environment according to an illustrative embodiment. More particularly, as shown, procedure 300 (e.g., network-side solution) involves a UE 302, a RAN node model training management function 304, an AMF 306, a 5G Core model training management function (e.g., NWDAF) 308, and a 5G Core NF (e.g., UDM or another NF) 310.

[0068] In step 1, UE 302 sets privacy level requirements (e.g., to low, medium, or high) when performing initial registration with the 5G Core network. In some embodiments, UE 302 can share its privacy level requirements with the 5G Core network via one or more methods, e.g., a short messaging service (SMS), a subscriber portal, a subscriber care calling, etc.

[0069] In step 2, the 5G Core network (e.g., 5G Core NF 310) sends a confirmation message back to UE 302 that the privacy level requirements have been received.

[0070] In step 3, the 5G Core network internally maps the privacy level requirements for UE 302 to technical privacy parameters. More particularly, the 5G Core network (5G Core NF 310 such as UDM or another NF) maps the nominal privacy level (e.g., low, medium, high) to the technical privacy level (e.g., to (c,5)-DP) and stores the same. For instance, given the privacy level (e.g., low, medium, high) for a certain policy, and if differential privacy is the privacy enhancing technology (PET), a specific use case of services could be as follows ((E, 5)- DPi, (s, 6 -DPI, E, 5)-DPa), respectively.

[0071] In step 4, based on the 5G Core network operator’s policy / regulations, the network operator updates or stores a different privacy level per subscriber and per data use case. For instance, if the use case is advertisement services, the network mapping of the UE nominal privacy level is tightly secured, i.e., high privacy level. However, if the use case is network performance improvement, the network aims for the smallest possible privacy level.

[0072] In step 5a / 5b, RAN node (e.g., gNB) model training management function 304 receives the privacy level of UE 302 stored in 5G Core NF 310 (e.g., UDM) via AMF 306 (5b), or 5G Core model training management function 308 (e.g., NWDAF) receives the privacy level of UE 302 directly from 5G Core NF 310 (5a).

[0073] For this, RAN node model training management function 304 can request the privacy data from 5G Core NF 310 and the 5G Core NF 310 provides the same. Alternatively, as a part of UE registration, 5G Core NF 310 can provide the privacy data to AMF 306 and AMF 306 provides the privacy data to RAN node model training management function 304. Similarly, the NWDAF or other 5G Core NF can also get this from the UDM or the UDM can push the same to NWDAF / 5G Core NF. In step 6a / 6b, UE 302 sends privacy sensitive data (either the complete training data / raw data or the model parameters after training) to 5G Core model training management function 308 (6a) or RAN node model training management function 304 (6b) over a secure interface. In some embodiments, the 5G Core network may request the privacy sensitive data from UE 302 and UE 302 provides the same in response to the request.

[0074] In step 7, the model training management function of the RAN node (304) or of the 5G Core network (308), after receiving the UE sensitive data, applies a privacy preserving mechanism using information received from steps 5a / 5b. This privacy is enhanced either for training data itself (before ML model training is initiated) or at ML model parameters sent by UE 302. The gNB / network enhances the privacy of the data according to the strictest requirement from all UEs (e.g., lowest c). For instance, the model training management function of the RAN node (304) or of the 5G Core network (308) trains the complete model after applying the privacy configuration received to UE training data or trains the remaining part of the model (in case of a two-sided model) or aggregates the ML model parameters (in the case of federated learning), after applying privacy mechanism on the model parameters received.

[0075] In step 8, 5G Core model training management function 308 sends the privacy- preserved, trained ML model to UE 302.

[0076] Turning now to FIG. 4, a procedure 400 is illustrated for preserving privacy of user equipment data during training of a machine learning model in a communication network environment according to another illustrative embodiment. More particularly, as shown, procedure 400 (e.g., UE-side solution) involves a UE 402, a RAN node model training management function 404, an AMF 406, a 5G Core model training management function (e.g., NWDAF) 408, and a 5G Core NF (e.g., UDM) 410.

[0077] Step 1, 2, 3 and 4 in procedure 400 are respectively performed in the same or a similar manner as steps 1, 2, 3 and 4 in procedure 300.

[0078] In step 5, once subscription data for UE 402 is updated with the privacy level, 5G Core NF 410 sends the policy (e.g., one or more privacy parameters) to UE 402 via a UE parameter update (UPU) or a policy control function (PCF) instructing what privacy to apply when the ML model runs at UE 402 or when data from UE 402 is exposed to the 5G Core network. In some embodiments, one or more of RAN node model training management function 404 and 5G Core model training management function 408 can also receive the policy (e.g., one or more privacy parameters).

[0079] In step 6, once UE 402 receives a privacy policy, UE 402 applies the privacy level to training data before using the training data to train the ML model or before sending the raw training data to the 5G Core network, or UE 402 applies the privacy policy to the model parameters before sending them to the 5G Core network for collaborative training or federated learning.

[0080] In step 7a / 7b, UE 402 sends the privacy preserved, trained data / model parameters / complete trained ML model to RAN node model training management function 404 or 5G Core model training management function 408.

[0081] As used herein, it is to be understood that the term “communication network” in some embodiments can comprise two or more separate communication networks. Further, the particular processing operations and other system functionality described in conjunction with the diagrams described herein are presented by way of illustrative example only and should not be construed as limiting the scope of the disclosure in any way. Alternative embodiments can use other types of processing operations and messaging protocols. For example, the ordering of the steps may be varied in other embodiments, or certain steps may be performed at least in part concurrently with one another rather than serially. Also, one or more of the steps may be repeated periodically, or multiple instances of the methods can be performed in parallel with one another.

[0082] It should again be emphasized that the various embodiments described herein are presented by way of illustrative example only and should not be construed as limiting the scope of the claims. For example, alternative embodiments can utilize different communication system configurations, user equipment configurations, base station configurations, provisioning and usage processes, messaging protocols and message formats than those described above in the context of the illustrative embodiments. These and numerous other alternative embodiments within the scope of the appended claims will be readily apparent to those skilled in the art.

[0083] As used herein, “at least one of the following: ”, “at least one of ”, “one or more of ” and similar wording, where the list of two or more elements are joined by “and” or “or”, mean at least any one of the elements, or at least any two or more of the elements, or at least all the elements.

Claims

We Claim:

1. An apparatus comprising: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to: send a selected privacy level to a communication network with which the apparatus is participating in a machine learning model training process; receive one or more privacy parameters corresponding to the selected privacy level; and apply the one or more privacy parameters to data to be sent to the communication network.

2. The apparatus of claim 1, wherein the data to which the one or more privacy parameters are applied comprises at least one of: machine learning model training data; one or more machine learning model parameters; and a trained machine learning model.

3. The apparatus of claims 1 or 2, wherein the one or more privacy parameters are associated with a privacy mechanism.

4. The apparatus of any one of claims 1-3, wherein the privacy level sent by the apparatus is one of a plurality of privacy levels that are selectable based on a communication network service with which the machine learning model training process is being performed.

5. The apparatus of any one of claims 1-4, wherein the apparatus is further caused to send the data, after application of the one or more privacy parameters, to a machine learning model training management function of the communication network.

6. The apparatus of claim 5, wherein the machine learning model training function of the communication network is a core network entity.

7. The apparatus of claim 6, wherein the core network entity is a network function.

8. The apparatus of claim 7, wherein the network function is a network data analytics function.

9. The apparatus of claim 8, wherein the network data analytics function comprises a model training logical function.

10. The apparatus of any one of claims 1-4, wherein the apparatus is further caused to send the data, after application of the one or more privacy parameters, to a machine learning model training management function of a radio access network associated with the communication network.

11. The apparatus of any one of claims 1-10, wherein the apparatus comprises or is comprised in a user equipment.

12. A method comprising: sending, from a user equipment, a selected privacy level to a communication network with which the user equipment is participating in a machine learning model training process; receiving, at the user equipment, one or more privacy parameters corresponding to the selected privacy level; and applying, by the user equipment, the one or more privacy parameters to data to be sent to the communication network.

13. The method of claim 12, wherein the data to which the one or more privacy parameters are applied comprises at least one of: machine learning model training data; one or more machine learning model parameters; and a trained machine learning model.

14. The method of claims 12 or 13, wherein the one or more privacy parameters are associated with a privacy mechanism.

15. The method of any of claims 12-14, wherein the privacy level sent by the user equipment is one of a plurality of privacy levels that are selectable based on a communication network service with which the machine learning model training process is being performed.

16. An apparatus comprising: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to: receive a selected privacy level from a user equipment with which the apparatus, in accordance with a communication network, is participating in a machine learning model training process; and determine one or more privacy parameters corresponding to the selected privacy level.

17. The apparatus of claim 16, wherein the apparatus is further caused to send the one or more privacy parameters to the user equipment.

18. The apparatus of claims 16 or 17, wherein the apparatus is further caused to send the one or more privacy parameters to one or more of: a machine learning model training management function of the communication network; and a machine learning model training management function of a radio access network associated with the communication network.

19. The apparatus of any one of claims 16-18, wherein the apparatus is further caused to determine the one or more privacy parameters corresponding to the selected privacy level by accessing a mapping of privacy levels to privacy parameters based on a privacy policy.

20. The apparatus of any one of claims 16-19, wherein the apparatus comprises or is comprised in a core network entity.

21. The apparatus of claim 20, wherein the core network entity is a network function.

22. The apparatus of claim 21, wherein the network function is a unified data management.

23. A method comprising:receiving, at a communication network entity, a selected privacy level from user equipment with which the communication network entity is participating in a machine learning model training process; and determining, by the communication network entity, one or more privacy parameters corresponding to the selected privacy level.

24. The method of claim 23, further comprising sending, from the communication network entity, the one or more privacy parameters to the user equipment.

25. The method of claims 23 or 24, further comprising sending, from the communication network entity, the one or more privacy parameters to one or more of: a machine learning model training management function of a communication network; and a machine learning model training management function of a radio access network associated with the communication network.

26. The method of any one of claims 23-25, wherein determining the one or more privacy parameters corresponding to the selected privacy level further comprises accessing a mapping of privacy levels to privacy parameters based on a privacy policy.

27. An apparatus comprising: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to: receive one or more privacy parameters, from a communication network entity, corresponding to a selected privacy level associated with a user equipment participating in a machine learning model training process; receive data from the user equipment; and apply the one or more privacy parameters to the data received from the user equipment.

28. The apparatus of claim 27, wherein the apparatus comprises or is comprised in a machine learning model training management function.

29. The apparatus of claim 28, wherein the machine learning model training management function is a core network entity.

30. The apparatus of claim 29, wherein the core network entity is a network function.

31. The apparatus of claim 30, wherein the network function is a network data analytics function.

32. The apparatus of claim 31 , wherein the network data analytics function comprises a model training logical function.

33. The apparatus of claim 28, wherein the machine learning model training management function is a radio access network entity.

34. A method comprising, receiving, at a machine learning model training management function, one or more privacy parameters, from a communication network entity, corresponding to a selected privacy level associated with a user equipment participating in a machine learning model training process; receiving, at the machine learning model training management function, data from the user equipment; and applying, by the machine learning model training management function, the one or more privacy parameters to the data received from the user equipment.

35. An apparatus comprising: means for sending a selected privacy level to a communication network with which the apparatus is participating in a machine learning model training process; means for receiving one or more privacy parameters corresponding to the selected privacy level; and means for applying the one or more privacy parameters to data to be sent to the communication network.

36. An apparatus comprising: means for receiving a selected privacy level from a user equipment with which the apparatus, in accordance with a communication network, is participating in a machine learning model training process; and means for determining one or more privacy parameters corresponding to the selected privacy level.

37. An apparatus comprising: means for receiving one or more privacy parameters, from a communication network entity, corresponding to a selected privacy level associated with a user equipment participating in a machine learning model training process; means for receiving data from the user equipment; and means for applying the one or more privacy parameters to the data received from the user equipment.

38. A computer readable non-transitory medium comprising program instructions stored thereon for performing at least the method of any one of claims 12-15, 23-26, or 34.

39. A computer readable medium comprising program instructions stored thereon for performing at least the method of any one of claims 12-15, 23-26, or 34.

40. A computer program comprising instructions for performing at least the method of any one of claims 12-15, 23-26, or 34.

41. A system comprising: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the system at least to: send a selected privacy level to a communication network participating in a machine learning model training process with a user equipment; receive one or more privacy parameters corresponding to the selected privacy level;apply the one or more privacy parameters to data to be sent to the communication network; receive a selected privacy level in accordance with the communication network participating in the machine learning model training process with the user equipment; determine one or more privacy parameters corresponding to the selected privacy level; receive one or more privacy parameters corresponding to the selected privacy level associated with the user equipment participating in the machine learning model training process; receive data from the user equipment; and apply the one or more privacy parameters to the data received from the user equipment.

42. A system comprising: means for sending a selected privacy level to a communication network participating in a machine learning model training process with a user equipment; means for receiving one or more privacy parameters corresponding to the selected privacy level; means for applying the one or more privacy parameters to data to be sent to the communication network; means for receiving a selected privacy level in accordance with the communication network participating in the machine learning model training process with the user equipment; means for determining one or more privacy parameters corresponding to the selected privacy level; means for receiving one or more privacy parameters corresponding to the selected privacy level associated with the user equipment participating in the machine learning model training process; means for receiving data from the user equipment; and means for applying the one or more privacy parameters to the data received from the user equipment.

Images