Differential privacy-based protection of data in communication network environment

Abstract

Techniques for differential privacy-based protection of data in a communication network environment are disclosed. In one example, from the perspective of a user device, a method includes receiving, by the user device, a data privacy configuration specifying a differential privacy mechanism applied by the user device prior to sending data to a communication network for use by a third party application. The method further includes, at the user device, applying the differential privacy mechanism to the data, and in response to the application of the differential privacy mechanism, sending the data from the user device to the communication network. The method may also include applying a portion of the differential privacy at the network (e.g., the gNB) to achieve total differential privacy for the configured list of information elements.

Description

Technical Field

[0001] This field generally relates to communication networks, and more specifically, but not exclusively, to security management within such communication networks. Background Technology

[0002] The descriptions in this section may help to better understand the various aspects of this invention. Therefore, the statements in this section should be read in this context and should not be construed as an admission of what is prior art or what is not prior art.

[0003] Fourth-generation (4G) wireless mobile communication technology, also known as Long Term Evolution (LTE) technology, is designed to provide high-capacity mobile multimedia, especially high data rates for human interaction. Next-generation or fifth-generation (5G) technology is not only designed for human interaction but also for machine-type communication in so-called Internet of Things (IoT) networks.

[0004] While 5G networks are designed to enable large-scale IoT services (e.g., a very large number of limited-capacity devices) and mission-critical IoT services (e.g., requiring high reliability), they also support improvements to traditional mobile communication services in the form of enhanced mobile broadband (eMBB) services, providing improved wireless internet access for mobile devices.

[0005] In an example communication system, a user equipment (a 5G UE in a 5G network, or more broadly, a UE) such as a mobile terminal (subscriber) communicates via an air interface with a base station or access point of the access network (referred to as a 5G AN in a 5G network). This access point (e.g., a gNB) is schematically part of the access network of the communication system.

[0006] For example, in 5G networks, the access network, referred to as 5G AN, is described in 5G Technical Specification (TS) 23.501 (titled "Technical Specification Group Services and System Aspects; System Architecture for 5G Systems") and TS 23.502 (titled "Technical Specification Group Services and System Aspects; Processes for 5G Systems (5GS)"), the disclosure of which is incorporated herein by reference in its entirety. Typically, an access point (e.g., gNB) provides the UE with access to the core network (CN or 5GC), and the core network then provides the UE with access to other UEs and / or data networks (e.g., packet data networks such as the Internet).

[0007] TS 23.501 further defines a 5G Service-Based Architecture (SBA) that models services as network functions (NFs) that communicate with each other using a RESTful API that represents state transitions.

[0008] In addition, TS 33.501 (titled “Technical Specification Group Services and Systems Aspects; Security Architecture and Procedures for 5G Systems”) describes security management details associated with 5G networks, the contents of which are incorporated herein by reference in their entirety.

[0009] Security management is a critical consideration in any communication network environment. However, as efforts continue to improve the architecture and protocols associated with 5G networks to enhance network efficiency and / or subscriber convenience, the security management of data exchanged within these environments can pose significant challenges. For example, implementing data privacy algorithms in a communication network environment presents a technical challenge. Summary of the Invention

[0010] The illustrative embodiments provide techniques for protecting data based on differential privacy in a communication network environment.

[0011] In one illustrative embodiment, from the perspective of a user equipment, a method includes: receiving a data privacy configuration by the user equipment, wherein the data privacy configuration specifies a differential privacy mechanism to be applied by the user equipment before transmitting data to a communication network for use by a third-party application. The method further includes: applying the differential privacy mechanism to the data at the user equipment; and transmitting the data from the user equipment to the communication network in response to the application of the differential privacy mechanism.

[0012] In another illustrative embodiment, from the perspective of a network operator entity, a method includes: generating a data privacy configuration by the network operator entity, the data privacy configuration being applied by multiple user devices and a communication network, the multiple user devices sending data to a third-party application via the communication network, wherein the data privacy configuration specifies that a differential privacy mechanism will be applied to the data before it is sent to the third-party application. The method further includes: sending the data privacy configuration from the network operator entity to the multiple user devices and the communication network.

[0013] In yet another illustrative embodiment, from the perspective of an access point in a communication network, a method includes: receiving, at the access point, a data privacy configuration, wherein the data privacy configuration specifies a differential privacy mechanism with total data randomization to be applied before data is sent to a third-party application, the data being received from a plurality of user devices. The method further includes: applying the differential privacy mechanism to the data by the access point, wherein a portion of the total data randomization applied by the access point to the data depends on a portion of the total data randomization applied to the data by each of the plurality of user devices. The method further includes: sending the data from the access point to the third-party application in response to the application of the differential privacy mechanism.

[0014] In yet another illustrative embodiment, from the perspective of a third-party application entity, a method includes: receiving data from a communication system at the third-party application entity, wherein the data includes one or more of first data and / or second data, the first data originating from a plurality of user devices, the second data originating from an access point associated with a communication network, and wherein the data includes total data randomization applied according to a differential privacy mechanism. The method further includes: the third-party application entity comparing the data with a data quality metric, and the third-party application entity utilizing at least a portion of the data based on the data quality metric comparison.

[0015] In another illustrative embodiment, from the perspective of a communication network entity, a method includes: receiving, at the communication network entity, a data privacy configuration that will be applied by a plurality of user devices and an access point through which the user devices send data to a third-party application, wherein the data privacy configuration specifies a differential privacy mechanism to be applied to the data before it is sent to the third-party application. The method further includes: generating a specific data privacy configuration by the communication network entity based on the received data privacy configuration, wherein the specific data privacy configuration specifies a set of privacy parameters and the data elements to which the privacy parameters will be applied.

[0016] The method also includes sending specific data privacy configurations from a communication network entity to multiple user devices and access points.

[0017] Further illustrative embodiments are provided in the form of a non-transitory computer-readable storage medium containing executable program code that, when executed by a processor, causes the processor to perform the steps described above. Still other illustrative embodiments include an apparatus having a processor and memory, configured to perform the steps described above.

[0018] Advantageously, the illustrative embodiments provide techniques for implementing data privacy configurations in a communication network environment that dynamically (e.g., adjustable) and collaboratively (e.g., between entities in the communication network environment) consider the trade-off between user equipment data size, total local noise of the required / desired application, and aggregated global noise of the required / desired application.

[0019] These, along with other features and advantages described herein, will become more apparent from the accompanying drawings and the following detailed description. Attached Figure Description

[0020] Figure 1 The illustration depicts a communication network environment in which one or more illustrative embodiments can be implemented.

[0021] Figure 2 The illustrations depict user equipment and entities that can implement one or more illustrative embodiments.

[0022] Figure 3A , 3B The diagrams in Figure 3C illustrate a process for protecting data based on differential privacy in a communication network environment according to an illustrative embodiment. Detailed Implementation

[0023] The embodiments will be described in conjunction with example communication systems and related technologies for security management in communication systems. However, it should be understood that the scope of the claims is not limited to the specific type of communication system and / or process disclosed. The embodiments can be implemented in a variety of other types of communication systems using alternative processes and operations. For example, although described in the context of a wireless cellular system utilizing 3GPP system elements (e.g., 3GPP Next Generation System (5G)), the disclosed embodiments are directly applicable to a variety of other types of communication systems, such as 6G communication systems.

[0024] According to illustrative embodiments implemented in a 5G communication system environment, one or more 3GPP Technical Specifications (TS) and Technical Reports (TRs) may also explain network elements / functions and / or operations that may interact with a portion of the present invention's solution, such as the aforementioned 3GPP TS 23.501, TS 23.502, and TS 33.501. Other 3GPP TS / TR documents may provide additional details that will be apparent to those skilled in the art. Note that 3GPP TS / TR documents are non-limiting examples of communication network standards (e.g., specifications, procedures, reports, requirements, recommendations, etc.). However, while highly suitable for 5G-related 3GPP standards, the embodiments are not necessarily intended to be limited to any particular standard.

[0025] It should be understood that the terms 5G network, etc. (e.g., 5G system, 5G communication system, 5G environment, 5G communication environment, etc.) may be understood, in some illustrative embodiments, to include all or part of the access network and all or part of the core network. However, the terms 5G network, etc. may occasionally be used interchangeably with the terms 5GC network, etc., without loss of generality, as any distinction will be understood by those skilled in the art.

[0026] Before describing the illustrative embodiments, the following will be... Figure 1 and Figure 2 A general description of some of the key components of a 5G network within the context of [the context].

[0027] Figure 1 A communication system 100 implementing an illustrative embodiment is shown. It should be understood that the elements shown in the communication system 100 are intended to represent some of the main functions provided within the system, such as control plane functions, user plane functions, etc. Therefore, Figure 1The blocks in the text refer to specific elements in a 5G network that provide some of the main functions. However, other network elements can be used to implement some or all of the represented main functions. Furthermore, it should be understood that... Figure 1 Not all the functions of the 5G network are depicted. Instead, at least some functions are shown to help explain the illustrative embodiments. Subsequent figures may depict some additional elements / functions (i.e., network entities).

[0028] Therefore, as shown in the figure, the communication system 100 includes a user equipment (UE) 102 that communicates with an access point 104 via an air interface 103. It should be understood that, in addition to a gNB, the UE 102 can use one or more other types of access points (e.g., access functions, networks, etc.) to communicate with the 5GC network. By way of example only, the access point 104 can be any 5G access network (gNB), an untrusted non-3GPP access network using non-3GPP interoperability functions (N3IWF), a trusted non-3GPP network using trusted non-3GPP gateway functions (TNGF), a wired access network using wired access gateway functions (W-AGF), or it can correspond to a traditional access point (e.g., an eNB). Furthermore, the access point 104 can be a wireless local area network (WLAN) access point, which will be further explained in the exemplary embodiments described herein.

[0029] UE 102 can be a mobile station, which may include, for example, a mobile phone, a computer, an IoT device, or any other type of communication device. Therefore, the term "user equipment" as used herein is intended to be interpreted broadly to encompass various types of mobile stations, user stations, or more generally, communication devices, including examples such as combinations of other devices such as a data card inserted into a laptop and a smartphone. Such communication devices are also intended to encompass devices commonly referred to as access terminals.

[0030] In one exemplary embodiment, UE 102 comprises a Universal Integrated Circuit Card (UICC) portion and a Mobile Equipment (ME) portion. The UICC is the user-dependent portion of the UE and includes at least one Universal Subscriber Identity Module (USIM) and suitable application software. The USIM securely stores a persistent subscription identifier and its associated key, which is used to uniquely identify and authenticate the user for network access. The ME is the user-independent portion of the UE and includes Terminal Equipment (TE) functionality and various Mobile Terminal (MT) functions. Alternative exemplary embodiments may not use UICC-based authentication, for example, a Non-Public (NPN) network.

[0031] Please note that in one example, the permanent subscription identifier is the UE-specific International Mobile Subscriber Identity (IMSI). In one embodiment, the IMSI is a fixed 15-digit number and consists of a 3-digit Mobile Country Code (MCC), a 3-digit Mobile Network Code (MNC), and a 9-digit Mobile Station Identifier (MSIN). In 5G communication systems, the IMSI is called the Subscription Permanent Identifier (SUPI). For the IMSI as a SUPI, the MSIN provides the subscriber identity. Therefore, typically only the MSIN portion of the IMSI needs to be encrypted. The MCC and MNC portions of the IMSI provide routing information used by the serving network to route to the correct home network. When the MSIN of the SUPI is encrypted, it is called the Subscription Hidden Identifier (SUCI). Another example of the SUPI uses the Network Access Identifier (NAI). NAIs are commonly used for IoT communications.

[0032] Access point 104 is exemplarily part of the radio access network or RAN of communication system 100. Such radio access network may include, for example, a 5G system with multiple base stations. More generally, components of the radio access network can be considered as “radio access entities”.

[0033] Furthermore, in this exemplary embodiment, access point 104 is operatively coupled to access and mobility management function (AMF / SEAF) 106. In a 5G network, AMF / SEAF supports mobility management (MM) and security anchor (SEAF) functions among other things.

[0034] In this exemplary embodiment, AMF / SEAF 106 is operatively coupled to other network functions 108 (e.g., services using other network functions 108). As shown, some of these other network functions 108 include, but are not limited to, Authentication Server Function (AUSF), Unified Data Management (UDM) function, Application Function (AF), and Network Data Analytics Function (NWDAF). These listed examples of network functions are typically implemented in the UE subscriber's home network, as explained further below. Note that in a 5GC network, the HSS (Home Subscriber Server) 4G function is split into AMF, UDM, and Unified Data Repository (UDR, not explicitly shown) functions. Typically, AMF authenticates the UE and provides any necessary encryption keys, while UDR stores user data and UDM manages user data. AF exposes the application layer to interact with 5G NFs and network resources. NWDAF is a 5G network function that collects data from various 5GC network functions, application functions, and Operations, Administration and Maintenance (OAM) and Operations Support systems. NWDAF aims to facilitate the generation and consumption of 5GC data and generate analytical insights and take action based on analytical insights. It also recognizes that third-party applications can be enabled to run with one or more network functions in 5GC.

[0035] 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. Furthermore, when the service provided includes data, the NFp providing the data is called a data producer, and the NFc requesting the data is called a data consumer. A data producer can also be an NF that generates data by modifying or otherwise processing data produced by another NF. Note that more generally, an NF can be considered a "network entity".

[0036] Please note that UEs such as UE 102 typically subscribe to a so-called Home Public Land Mobile Network (HPLMN), where some or all of functions 106 and 108 reside. Alternatively, UEs such as UE 102 may receive services from the NPN where these functions may reside. The HPLMN is also known as the Home Environment (HE). If the UE is roaming (not in the HPLMMN), it typically connects to a Visited Public Land Mobile Network (VPLMN), also known as the Visited Network, and the network currently serving the UE is also called the Serving Network. In roaming situations, some network functions 106 and 108 may reside in the VPLMN, in which case the functions in the VPLMN communicate with the functions in the HPLMMN as needed. However, in non-roaming scenarios, access and mobility management function 106, along with other network functions 108, reside in the same communication network, i.e., the HPLMMN. Unless otherwise stated, the embodiments described herein are not necessarily limited to which functions reside in which PLMN (i.e., HPLMMN or VPLMN).

[0037] Access point 104 is also operatively coupled (via one or more of functions 106 and / or 108) to Session Management Function (SMF) 110, which in turn is operatively coupled to User Plane Function (UPF) 112. UPF 112 is operatively coupled to a packet data network, such as the Internet 114. Note that the thicker solid lines in the diagram represent the user plane (UP) of the communication network, while the thinner solid lines represent the control plane (CP) of the communication network. It is important to understand that... Figure 1 The network (e.g., the Internet) 114 in the diagram may additionally or alternatively represent other network infrastructures, including but not limited to cloud computing infrastructure and / or edge computing infrastructure. Furthermore, typical operation and functionality of these network elements are not described herein, as they are not the focus of this exemplary embodiment and can be found in the appropriate 3GPP 5G documentation. Note that the functions shown in 106, 108, 110, and 112 are examples of network functions (NFs).

[0038] It should be understood that this particular arrangement of system elements is merely an example, and in other embodiments, additional or alternative elements of other types and arrangements may be used to implement the communication system. For example, in other embodiments, communication system 100 may include other elements / functions not explicitly shown herein.

[0039] therefore, Figure 1 The proposed configuration is just one example of a wireless cellular system configuration; various alternative configurations of system elements can be used. For example, although... Figure 1 The examples illustrate only a single element / function, but this is for the sake of brevity and clarity. Of course, the given alternative embodiments may include a greater number of such system elements, as well as additional or alternative elements common in conventional system implementations.

[0040] It should also be noted that, although Figure 1 While system elements are illustrated as single functional blocks, the various subnetworks that make up a 5G network are divided into so-called network slices. Network slices (network partitions) are logical networks that provide specific network capabilities and characteristics, supporting corresponding service types, and optionally using Network Function Virtualization (NFV) on common physical infrastructure. With NFV, network slices can be instantiated according to the needs of a given service (e.g., eMBB service, large-scale IoT service, mission-critical IoT service). Therefore, a network slice or function is 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 in the underlying physical infrastructure. UE 102 is configured to access one or more of these services via access point 104.

[0041] Figure 2 This is a block diagram illustrating the computational architecture for various participants in a method according to an illustrative embodiment. More specifically, the illustrated system 200 includes a user equipment (UE) 202 and multiple entities 204-1, ..., 204-N. For example, in an exemplary embodiment and with reference to... Figure 1 UE 202 can represent UE 102, while entities 204-1, ..., 204-N can represent functions 106 and 108 (i.e., network entities) and access point 104 (i.e., radio access entity). It should be understood that UE 202 and entities 204-1, ..., 204-N are configured to interact to provide security management and other technologies as described herein.

[0042] User equipment 202 includes a processor 212 coupled to memory 216 and interface circuitry 210. The processor 212 of user equipment 202 includes a security management processing module 214, which may be implemented at least partially in the form of software executed by the processor. The security management processing module 214 performs security management as described in the following figures and elsewhere herein. The memory 216 of user equipment 202 includes a security management storage module 218 that stores data generated or used during security management operations.

[0043] Each entity (referred to herein individually or collectively as 204) includes a processor 222 (222-1, ..., 222-N) coupled to a memory 226 (226-1, ..., 226-N) and an interface circuitry system 220 (220-1, ..., 220-N). Each processor 222 of each entity 204 includes a security management processing module 224 (224-1, ..., 224-N), which may be implemented at least partially in the form of software executed by the processor 222. The security management processing module 224 performs security management operations as described in conjunction with the following figures and elsewhere herein. Each memory 226 of each entity 204 includes a security management storage module 228 (228-1, ..., 228-N) containing data generated or otherwise used during security management operations.

[0044] Processors 212 and 222 may include, 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 these elements.

[0045] Memory 216 and 226 may be used to store one or more software programs that are executed by corresponding processor 212 and 222 to implement at least a portion of the functions described herein. For example, security management operations and other functions, which can be implemented in a direct manner by software code that can be executed by processor 212 and 222, in conjunction with the following figures and other descriptions herein.

[0046] Therefore, either memory 216 or 226 can be considered as an example of what is more generally referred to herein as a computer program product or, more generally, a processor-readable storage medium having executable program code. Other examples of processor-readable storage media may include magnetic disks or other types of magnetic or optical media, or any combination thereof. Exemplary embodiments may include an article of manufacture that includes such computer program products or other processor-readable storage media.

[0047] Furthermore, memories 216 and 226 may more specifically include, 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 memory such as flash memory, magnetoresistive RAM (MRAM), phase-change RAM (PC-RAM), or ferroelectric RAM (FRAM). The term “memory” as used herein is intended to be interpreted broadly and may additionally or alternatively encompass, for example, read-only memory (ROM), disk-based memory, or other types of storage devices, and portions or combinations thereof.

[0048] Interface circuit systems 210 and 220 exemplarily include transceivers or other communication hardware or firmware that allow associated system components to communicate with each other in the manner described herein.

[0049] from Figure 2 It is evident that user equipment 202 and multiple entities 204 are configured to communicate with each other as security management participants via their respective interface circuit systems 210 and 220. This communication involves each participant sending data to one or more other participants and / or receiving data from one or more other participants. The term "data" as used herein is intended to be interpreted broadly 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 should be understood that Figure 2 The specific arrangement of components shown is merely an example, and many alternative configurations can be used in other embodiments. For example, any given network element / function and / or access point can be configured to include additional or extra components and support other communication protocols.

[0051] Other system elements (such as access point 104, SMF 110, and UPF 112) can each be configured to include components such as processors, memory, and network interfaces. Furthermore, third-party applications and network operators (will...) Figure 3A , 3B Entities (further explained in the context of 3C) can participate in the methodologies described herein via computing devices configured to include components such as processors, memory, and network interfaces. These elements and devices need not be implemented on separate, independent processing platforms, but can represent different functional parts of a single, common processing platform.

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

[0053] As stated above, 3GPP TS 23.501 defines the 5GC network architecture as service-based, for example, service-based architecture (SBA). This document recognizes that when deploying different NFs, there may be many situations where one NF may need to interact with entities outside the SBA-based 5GC network (e.g., including corresponding PLMNs, such as HPLMN and VPLMN). Therefore, as used herein, the term "internal" exemplarily refers to operations and / or communications within the SBA-based 5GC network (e.g., SBA-based interfaces), while the term "external" exemplarily refers to operations and / or communications outside the SBA-based 5GC network (non-SBA interfaces).

[0054] In light of the above general description of some characteristics of 5GC networks, the problems of existing security methods in protecting data based on differential privacy in communication network environments, and the solutions proposed according to exemplary embodiments, will now be described below.

[0055] Differential privacy guarantees data privacy at the cost of reducing the accuracy of the utility function required for the data. Randomizing the function of the data (function perturbation), purely primitive randomization of the data (input perturbation), or randomization of the function's output (output perturbation) are among the widely used differential privacy mechanisms. This paper recognizes that differential privacy mechanisms can be improved if we know which function to compute. In other words, this paper recognizes that having knowledge of the function can simultaneously improve both utility and privacy.

[0056] Differential privacy was initially proposed to provide formal privacy guarantees when commonly used anonymization techniques, such as k-anonymity and 1-divergence, were insufficient. Differential privacy is a statistically significant technique where, once an algorithm or data is protected by differential privacy, the privacy of individuals within the dataset remains protected regardless of how the output of the algorithm or data is further processed. The original motivation for proposing this strong privacy definition stemmed from the observation that combining data from different sources could compromise privacy. For example, the U.S. Census Bureau now employs differential privacy for privacy protection, and several major customer data-driven companies also use it. Differential privacy also helps in complying with data privacy regulations such as the General Data Protection Regulation (GDPR) and the California Consumer Privacy Act (CCPA).

[0057] Differential privacy guarantees a mathematical upper bound on the risk of leaking individual information and allows users to quantify the acceptable level of individual risk. Differential privacy is described using probability theory, and its mathematical definition is as follows. Assume two datasets... X and Y They are neighbors, if replacing a single data entry in one dataset yields the other. Let... ε >0 and ( ε and δ (This is a differential privacy parameter). A random function. M(X) : X N → yes( ε, δ -DP (where DP stands for Differential Privacy), if for each pair of neighbor datasets X and Y And for each result We have: .

[0058] parameter δ This can be interpreted as the probability of a complete data breach, while ε The smaller the value, the harder it is to determine whether the output originates from the dataset. X still Y That is, the less likely it is that any individual will be aware of the existence of the mechanism, the stronger its privacy protection.

[0059] In some differential privacy mechanisms, the number of dataset entries is related to the level of privacy provided ( ε, δ There is a fundamental relationship between the required noise variance and the number of data points. This relationship gives rise to the concept known as the small dataset size problem, where the more data there is, the less noise is needed to protect individual privacy. This problem arises in a federated learning setting where N sites each have D data points and publish a function of their datasets. If each user adds noise proportional to the size of their own dataset to utilize (ε, δ)-DP to protect their privacy, then the final aggregated dataset on the network side will have a smaller noise variance than all the data points combined. Data points are concentrated at a central point and used ( ε, δ The noise variance is N times larger in the case of DP. Several techniques have been developed to collect this data without compromising the security and privacy of the dataset; however, most of them either assume the existence of a trusted third party or use homomorphic encryption, the latter of which is computationally very resource-intensive.

[0060] Furthermore, this paper recognizes the benefits of using differential privacy in wireless networks. For example, the communication network can act as a regulator to protect the signaling required for communication between user devices participating in NWDAF and third-party AI / machine learning engines or neural network model trainers. Less noise may be required when many UEs are involved in AI / ML-based solutions. Existing solutions, among other technical shortcomings, do not offer the flexibility to allow such configuration and decisions from the network side.

[0061] Furthermore, this paper recognizes that gNBs may independently send some data to AI / ML applications. This data may be privacy-sensitive for operators. For example, cell names or cell identifiers (IDs) could reveal the location of a Base Transceiver Station (BTS). If third-party AI / ML applications are implemented, country-specific privacy aspects may also be important. This could include data that operators deem privacy-sensitive. For example, operators in some countries may not want consumers abroad to know the cell names. If the public cloud hosting the AI / ML application is located abroad, network operators may also want to ensure the privacy of network data (e.g., cell names). In this case, the BTS or other NFs may also need to add noise—using differential privacy.

[0062] This paper further recognizes that networks can act as trusted third parties, enhancing the trade-off between privacy and utility by addressing the aforementioned small dataset size issue. If federated learning with a Gaussian mechanism is used, with 100 UEs, the gain in utility (training accuracy) is 100-fold. In other words, noise with a variance 100-fold smaller needs to be added.

[0063] Ensuring the privacy of data shared by UEs for AI / ML applications on wireless networks (in OAM, wireless edge, centralized cloud, and third-party applications) is an unresolved issue within 3GPP and the Open RAN (ORAN) standardization bodies.

[0064] Exemplary embodiments address the aforementioned technical challenges and / or shortcomings of existing data privacy methods in communication network environments by providing differential privacy-based protection for dynamic and collaborative data (e.g., sensitive data). More specifically, in one or more exemplary embodiments, the UE and gNB are configured to collaboratively generate differential privacy-enabled data.

[0065] The reason for involving the gNB lies in the aforementioned small dataset problem. If a UE attempts to provide a given (ε, δ)-DP individually, the variance of additive noise in a practical and widely accepted Gaussian approach will be inversely proportional to the number of data points a particular UE has. However, this paper recognizes that if data from multiple UEs is aggregated to the gNB before noise is added, the required amount of noise is significantly reduced (proportional to the number of UEs). Therefore, the exemplary embodiment enables the UE to add some relatively small local noise to obtain minimal privacy guarantees, while the remaining required noise is added by the gNB after all other UEs have uploaded their data.

[0066] Advantageously, the exemplary embodiments have the following advantages in particular: (i) allowing orders of magnitude gains in data learning and utility by adding less noise; (ii) strategically positioning the network as a trusted third party to handle privacy aspects of data between the UE and the AI / ML application; (iii) allowing the AI / ML application to know which datasets have less noise (higher data quality), thereby enabling better and faster neural network training (i.e., more energy-efficient neural network training based on this knowledge); and (iv) concise signaling, allowing coordination and cooperation between the data holder (UE) and the data processor (AI / ML application).

[0067] Therefore, in one or more exemplary embodiments, a global (ε, δ) is configured for the user equipment from the network side. When scrambled data is used, the UE ultimately decides how much (ε, δ) to use and transmit to the network. Note that the term (ε, δ)-DP exemplarily refers to the amount of noise defined by (ε, δ) introduced using a specific differential privacy mechanism.

[0068] The network shares the available (ε, δ)-DP based on the number of active UEs. Each UE can have an independent privacy objective (ε). i , δ j )-DP.

[0069] If (ε i , δ j If (ε, δ) ≥ (ε), meaning the gNB's privacy requirements are stricter than the user equipment's, then the UE only needs to send its data to the gNB; otherwise, if (ε) ≥ (ε, δ), then the UE only needs to send its data to the gNB. i , δ j If (ε, δ) < (ε, δ), then the UE provides (ε'). j , δ' j (ε)-DP, such that after gNB provides additional (ε, δ)-DP, the published dataset becomes (ε)-DP. i , δ j ).

[0070] gNB collects data from all UEs. Based on the number of UEs, N ue Given the target (ε, δ)-DP, gNB performs differential privacy mechanisms to achieve the target; for example, gNB can access privacy from (0, σ) 2 Sampling is performed in a Gaussian noise distribution, where σ ~ 1 / N ue And add it to the data.

[0071] On the network side, based on the privacy objective (ε) for each UE i , δ j To understand this, a data quality score is introduced. This data quality metric (score) is related to (ε). i , δ j This is inversely proportional. The network operator or the network itself informs the AI / ML functions of this score and takes it into account when training the neural network.

[0072] Figure 3A , 3B Reference 3C illustrates a process 300 for differential privacy-based protection of data in a communication network environment, according to an exemplary embodiment. More specifically, process 300 illustrates a detailed end-to-end message sequence between entities including UE 302, gNB 304, serving network 5GC 306, AI / ML application (or any other consumer application) 308, and (network) operator 310. Reference numerals in process 300 correspond to steps 1-15 (including stages, sub-procedures, operations, etc.) performed at or between entities. Prior to steps 1-15, it is assumed that for each UE 302, primary authentication has been successful relative to gNB 304 and serving network 5GC 306, and non-access stratum (NAS) and access stratum (AS) security contexts have been established. It should also be noted that it is assumed that AI / ML application 308 can be trained on a third party, OAM, wireless edge platform, or centralized cloud platform.

[0073] In step 1, operator 310 defines a privacy policy that specifies the types of data considered privacy-sensitive. As an example only, operator 310 may specify that any data including public Internet Protocol (IP) addresses, cell names, subscriber information, and personal data such as email addresses is considered privacy-sensitive.

[0074] In step 2, the serving network 5GC 306 derives the global and target epsilon and delta (ε) from the data shared by UE 302 and the data with added noise from UE 302 and gNB 304. i , δ j )parameter.

[0075] In step 3, the serving network 5GC 306 also translates the privacy policy into specific information elements (IEs) and data points, which are collected by the UE 302 and gNB 304 for use by various AI / ML applications, including AI / ML application 308. Note that some reports, such as Minimized Drive Test Reports (MDTs) and Radio Link Failure (RLF) reports, are shared from the UE side, and the privacy-sensitive elements in these reports are also deduced by the serving network 5GC 306.

[0076] In step 4, the privacy configuration derived from the above two steps is then sent from service network 5GC 306 to gNB 304.

[0077] In step 5, gNB 304 extracts its privacy configuration as a data source and stores the configuration locally for use in subsequent step 11.

[0078] Note that steps 6 through 10 are performed for each UE 302 involved in collecting data for AI / ML application 308.

[0079] In step 6, the privacy configuration related to the UE is sent to each UE 302. As mentioned above, certain reports (e.g., MDT and RLF) sent by each UE 302 may contain specific privacy-sensitive elements (e.g., UE location). Each UE 302 should only scramble these elements in these reports.

[0080] In step 7, each UE 302 determines the exact epsilon, delta (ε) for each IE that needs scrambling (adding noise). i , δ j )parameter.

[0081] In step 8, before sending data, each UE 302 applies differential privacy, using the epsilon and delta (ε) values ​​determined in the above steps. i , δ j )parameter.

[0082] In step 9, each UE 302 notifies gNB 304 of the privacy parameters for each IE (i.e., (ε) i , δ j (Parameters). In this step, a merged list with associated privacy parameters can be sent.

[0083] In step 10, each UE 302 sends relevant data to gNB 304, including privacy-sensitive data that has been scrambled / noised.

[0084] In step 11, gNB 304 performs differential privacy on its own data and the data received from UE 302. For the data received from UE 302, gNB 304 considers the target epsilon, delta (ε) i , δ j The parameters are then used to apply differential privacy.

[0085] Note that steps 12-15 are performed for each dataset shared with AI / ML application 308.

[0086] In step 12, gNB 304 provides AI / ML application 308 with details about which IEs utilize how much noise and which differential privacy mechanism / scheme is protected by privacy.

[0087] For steps 13-15, AI / ML application 308 decides to use only data that has enough noise to protect privacy but not so much that it would corrupt the data.

[0088] In step 13, the AI / ML application 308 will set the privacy parameter (i.e., (ε)). i , δ j The parameter is mapped to a data quality metric. (ε) i , δ j The parameters are inversely proportional, and the data quality is determined by gNB 304.

[0089] In step 14, AI / ML application 308 labels the data with the corresponding data quality metric (e.g., a value within a data quality value range).

[0090] In step 15, the labeled data is used for training and / or inference of the neural network.

[0091] It's important to understand that, in order to determine the distinction between noise added by the local UE and noise added by the gNB, the network can share the possible (ε, δ)-DP (i.e., total data randomization) available based on the number of active UEs. Recall that, as mentioned above, each UE can have an independent privacy objective (ε). i , δ j )-DP. If (ε i , δ j If (ε > (ε, δ), meaning the gNB has stricter privacy requirements than the UE, then the UE only needs to send its data to the gNB. Otherwise, if (ε i , δ j If (ε, δ) < (ε, δ), then the UE provides (ε'). j ,δ' jThe (ε, δ)-DP (i.e., the UE portion of total data randomization) makes the published dataset (ε) become (ε) after gNB provides additional (ε, δ)-DP (i.e., the gNB portion of total data randomization). i , δ j ).

[0092] While there are multiple methods to achieve (ε, δ)-DP, there is only one way to add noise. If the method is fixed, (ε') can be calculated. j , δ' j (ε, δ)-DP. As an example, let's assume both the gNB and the UE use a Gaussian mechanism. Then, the steps above determine how much noise each UE must add to its data, based on the amount of noise provided by the gNB. For a fixed (ε, δ)-DP value, the variance of the required noise can be calculated as:

[0093] If (ε i , δ j If (ε, δ) < (ε), then (ε) is realized. i , δ j The variance of the noise required for DP can be calculated using the above formula:

[0094] Then, σ' = σ i — σ indicates the amount of noise that needs to be added locally.

[0095] Note that the example above uses the property of an independent Gaussian random variable. However, other mechanisms can also be used, such as the Laplace mechanism, random response, etc. Note that gNB does not need to disclose the mechanism it uses.

[0096] Note that the number of active UEs in each round may include the issue of lagging UEs. That is, if a subset of UEs fails to participate and does not share their results, gNB needs to add more noise to achieve the same (ε, δ)-DP. Again assuming a Gaussian mechanism, Δ~ = 1 / N is the sensitivity parameter, which depends particularly on the number of participants.

[0097] In other words, at the start of each round of data collection, the gNB either waits for the lagging UEs to provide their data or readjusts its noise level based on the new Δ value.

[0098] As used herein, it should be understood that the term "communication network" in some embodiments may include two or more independent communication networks. Furthermore, the specific processing operations and other system functions described herein in conjunction with the accompanying drawings are presented as illustrative examples only and should not be construed as limiting the scope of this disclosure in any way. Alternative embodiments may use other types of processing operations and messaging protocols. For example, in other embodiments, the order of steps may vary, or certain steps may be executed at least partially concurrently with each other rather than serially. Furthermore, one or more steps may be performed periodically, or multiple method instances may be executed in parallel.

[0099] It should be emphasized again that the various embodiments described herein are presented as illustrative examples only and should not be construed as limiting the scope of the claims. For example, alternative embodiments may utilize different communication system configurations, user equipment configurations, base station configurations, provisioning and usage procedures, messaging protocols, and message formats than those described in the exemplary embodiments above. These, and numerous other, alternative embodiments within the scope of the claims will be apparent to those skilled in the art.

Claims

1. An apparatus comprising: Components for receiving data privacy configurations, wherein the data privacy configurations specify differential privacy mechanisms applied by the device before data is sent to a communication network for use by a third-party application; Components for applying the differential privacy mechanism to the data; as well as A component for transmitting the data to the communication network in response to the application of the differential privacy mechanism.

2. The apparatus of claim 1, wherein the differential privacy mechanism applied by the apparatus represents at least a portion of the total data randomization to be applied to the data before the data is received by the third-party application.

3. The apparatus of claim 2, wherein the component for applying the differential privacy mechanism further comprises: A component for selecting a set of privacy parameters associated with the differential privacy mechanism, the set of privacy parameters being applied to the data, wherein the selected set of privacy parameters will affect a portion of the total data randomization of the data.

4. The apparatus of claim 3, further comprising a component for transmitting the selected set of privacy parameters to the communication network.

5. The apparatus of claim 3, wherein the data privacy configuration further specifies one or more types of data elements to which the selected set of privacy parameters will be applied.

6. A method comprising: The user equipment receives a data privacy configuration, wherein the data privacy configuration specifies a differential privacy mechanism applied by the user equipment before the data is sent to the communication network for use by a third-party application; At the user equipment, the differential privacy mechanism is applied to the data; as well as In response to the application of the differential privacy mechanism, the data is transmitted from the user equipment to the communication network.

7. An apparatus comprising: Components for generating data privacy configurations, which will be applied by multiple user devices and communication networks, which send data to third-party applications through the communication networks, wherein the data privacy configuration specifies that a differential privacy mechanism will be applied to the data before the data is sent to the third-party applications; as well as A component for sending the data privacy configuration to the plurality of user devices and the communication network.

8. The apparatus of claim 7, wherein the data privacy configuration further specifies one or more types of data elements to which the differential privacy mechanism will be applied.

9. The apparatus of claim 7, wherein the data privacy configuration further specifies that the total data to be applied to the data is randomized before the data is sent to the third-party application.

10. The apparatus of claim 7, wherein the differential privacy mechanism to be applied to the data before sending the data to the third-party application is selected by the apparatus based on the number of user devices among the plurality of user devices.

11. A method comprising: A data privacy configuration is generated by a network operator entity. This data privacy configuration will be applied by multiple user devices and a communication network. These multiple user devices send data to a third-party application through the communication network. The data privacy configuration specifies that a differential privacy mechanism will be applied to the data before it is sent to the third-party application. as well as The data privacy configuration is sent from the network operator entity to the plurality of user devices and the communication network.

12. An apparatus comprising: A component for receiving data privacy configurations, wherein the data privacy configurations specify a differential privacy mechanism with total data randomization to be applied before sending data to a third-party application, the data being received from multiple user devices; Components for applying the differential privacy mechanism to the data, wherein a portion of the total data randomization applied by the means to the data depends on a portion of the total data randomization applied to the data by each of the plurality of user devices; as well as A component for sending the data to the third-party application in response to the application of the differential privacy mechanism.

13. The apparatus of claim 12, further comprising: Components for applying the differential privacy mechanism to data generated by the device; as well as A component for sending the data generated by the device to the third-party application in response to the application of the differential privacy mechanism.

14. The apparatus of claim 13, further comprising: A component for sending the set of privacy parameters applied according to the differential privacy mechanism to the third-party application.

15. The apparatus of claim 12, further comprising: A component for adjusting a portion of the total data randomization applied by the device to the data based on the number of the plurality of user devices providing data to the device.

16. A method comprising: At the access point, a data privacy configuration is received, wherein the data privacy configuration specifies a differential privacy mechanism with total data randomization to be applied before the data is sent to a third-party application, the data being received from multiple user devices; The differential privacy mechanism is applied to the data by the access point, wherein a portion of the total data randomization applied to the data by the access point depends on a portion of the total data randomization applied to the data by each of the plurality of user devices; as well as In response to the application of the differential privacy mechanism, the data is sent from the access point to the third-party application.

17. An apparatus comprising: A component for receiving data from a communication system, wherein the data includes one or more of first data and second data, the first data originating from multiple user devices, the second data originating from an access point associated with the communication network, and wherein the data includes total data randomization applied according to a differential privacy mechanism. Components for comparing the data with data quality metrics; as well as A component for utilizing at least a portion of the data based on comparisons using the data quality metric.

18. The apparatus of claim 17, further comprising: A component for receiving a set of privacy parameters applied according to the differential privacy mechanism.

19. The apparatus of claim 17, wherein the component for utilizing at least a portion of the data further comprises: The aforementioned portion of the data is utilized in artificial intelligence / machine learning applications.

20. A method comprising: At a third-party application entity, data is received from a communication system, wherein the data includes one or more of first data and second data, the first data originating from multiple user devices, the second data originating from an access point associated with the communication network, and wherein the data includes total data randomized according to a differential privacy mechanism. as well as The third-party application entity compares the data with data quality metrics; as well as The third-party application entity uses at least a portion of the data based on the data quality metric comparison.

21. An apparatus comprising: A component for receiving data privacy configurations, which will be applied by multiple user devices and access points, which send data to a third-party application through the access points, wherein the data privacy configuration specifies that a differential privacy mechanism will be applied to the data before it is sent to the third-party application; A component for generating a specific data privacy configuration based on the received data privacy configuration, wherein the specific data privacy configuration specifies a set of privacy parameters and the data elements to which the set of privacy parameters will be applied; as well as A component for sending the specific data privacy configuration to the plurality of user devices and the access point.

22. The apparatus of claim 21, wherein the set of privacy parameters will influence a corresponding portion of the total data randomization of the data elements by the plurality of user devices and the access point.

23. The apparatus of claim 21, wherein the specific data privacy configuration further specifies which data elements correspond to the plurality of user equipments and which data elements correspond to the access point.

24. A method comprising: At the communication network entity, a data privacy configuration is received, which will be applied by multiple user devices and access points, which send data to a third-party application through the access points. The data privacy configuration specifies that a differential privacy mechanism will be applied to the data before it is sent to the third-party application. The communication network entity generates a specific data privacy configuration based on the received data privacy configuration, wherein the specific data privacy configuration specifies a set of privacy parameters and the data elements to which the set of privacy parameters will be applied; as well as The specific data privacy configuration is sent from the communication network entity to the plurality of user devices and the access point.

Images