Techniques for enabling artificial intelligence (AI) model sharing with privacy-preservation

Abstract

Methods of enabling Artificial Intelligence (Al) models to be trained in a privacy preserving manner are disclosed. A method comprises the formation of virtual cluster per privacy level and selection of a cluster head that performs the training and coordination per privacy level. A method comprises training an Al primary model on a private user data set. The Al primary model may be used to label a public data set. A secondary model may be trained using the labeled public data set. The secondary model may be shared with a network or other UEs.

Description

Client Ref. No. P62335WO1 TECHNIQUES FOR ENABLING ARTIFICIAL INTELLIGENCE (AI) MODEL SHARING WITH PRIVACY-PRESERVATION FIELD

[0001] Embodiments of the invention relate to wireless communications, including apparatuses, systems, and methods which may be used for artificial intelligence (AI) model sharing. DESCRIPTION OF THE RELATED ART

[0002] Wireless communication systems are rapidly growing in usage. In recent years, wireless devices such as smart phones and tablet computers have become increasingly sophisticated. In addition to supporting telephone calls, many mobile devices now provide access to the internet, email, text messaging, and navigation using the global positioning system (GPS) and are capable of operating sophisticated applications that utilize these functionalities.

[0003] Long Term Evolution (LTE) has been the technology of choice for the majority of wireless network operators worldwide, providing mobile broadband data and high-speed Internet access to their subscriber base. LTE was first proposed in 2004 and was first standardized in 2008. Since then, as usage of wireless communication systems has expanded exponentially, demand has risen for wireless network operators to support a higher capacity for a higher density of mobile broadband users. In 2015, a study of a new radio access technology began and, in 2017, a first release of Fifth Generation New Radio (5G NR) was standardized.

[0004] 5G-NR, also simply referred to as NR, provides, as compared to LTE, a higher capacity for a higher density of mobile broadband users, while also supporting device-to-device, ultra-reliable, and massive machine type communications with lower latency and / or lower battery consumption. Further, NR may allow for more flexible scheduling as compared to current LTE. Consequently, efforts are being made in ongoing developments of NR to take advantage of higherClient Ref. No. P62335WO1 throughputs possible at higher frequencies. BRIEF DESCRIPTION OF THE DRAWINGS

[0005] A better understanding of the present subject matter can be obtained when the following detailed description of various embodiments is considered in conjunction with the following drawings, in which:

[0006] FIG. 1A illustrates an example wireless communication system according to some embodiments.

[0007] FIG.1B illustrates an example of a base station and an access point in communication with a user equipment (UE) device, according to some embodiments.

[0008] FIG. 2 illustrates an example block diagram of a base station, according to some embodiments.

[0009] FIG.3 illustrates an example block diagram of a server according to some embodiments.

[0010] FIG.4 illustrates an example block diagram of a UE according to some embodiments.

[0011] FIG.5 illustrates an example block diagram of cellular communication circuitry, according to some embodiments.

[0012] FIG.6 illustrates an example of a baseband processor architecture for a UE, according to some embodiments.

[0013] FIG.7 illustrates an example block diagram of an interface of baseband circuitry according to some embodiments.

[0014] FIG. 8 illustrates an example of a control plane protocol stack in accordance with some embodiments.

[0015] FIG. 9 illustrates an example of a user plane protocol stack in accordance with some embodiments.

[0016] FIG.10 illustrates example components of a core network in accordanceClient Ref. No. P62335WO1 with some embodiments.

[0017] FIG.11 illustrates an example of using synthetic data for training an AI model according to some embodiments.

[0018] FIG.12 illustrates an example of distributed collaboration for training an AI model according to some embodiments.

[0019] FIG.13 illustrates an example of decentralized collaboration for training an AI model according to some embodiments.

[0020] FIG.14 illustrates an example of collaboration for training an AI model according to some embodiments.

[0021] FIG.15 illustrates an example of distributed collaboration for training an AI model with privacy preservation according to some embodiments.

[0022] FIG.16 illustrates an example of decentralized collaboration for training an AI model with privacy preservation according to some embodiments.

[0023] FIG.17 illustrates a flow chart of an example method of AI training at a user equipment (UE), according to some embodiments.

[0024] FIGS. 18A-18B illustrates illustrate examples where physical cluster of UEs form virtual clusters for supporting privacy levels, according to some embodiments.

[0025] FIGS.19A-19B illustrate an example of training an AI model with privacy preservation according to some embodiments.

[0026] FIGS.20A-20B illustrate an example of training an AI model with privacy preservation according to some embodiments.

[0027] FIG.21 illustrates a flow chart of an example method of AI training at a user equipment (UE), according to some embodiments.

[0028] FIG.22 illustrates a flow chart of an example method of AI training in a D2D architecture, according to some embodiments.

[0029] While the features described herein may be susceptible to various modifications and alternative forms, specific embodiments thereof are shown by way of example in the drawings and are herein described in detail. It should beClient Ref. No. P62335WO1 understood, however, that the drawings and detailed description thereto are not intended to be limiting to the particular form disclosed, but on the contrary, the intention is to cover all modifications, equivalents and alternatives falling within the spirit and scope of the subject matter as defined by the appended claims. DETAILED DESCRIPTION Terms

[0030] The following is a glossary of terms used in this disclosure:

[0031] Memory Medium or Memory – Any of various types of non-transitory memory devices or storage devices. The term “memory medium” is intended to include an installation medium, e.g., a CD-ROM, floppy disks, or tape device; a computer system memory or random-access memory such as DRAM, DDR RAM, SRAM, EDO RAM, Rambus RAM, etc.; a non-volatile memory such as a Flash, magnetic media, e.g., a hard drive, or optical storage; registers, or other similar types of memory elements, etc. The memory medium may include other types of non-transitory memory as well or combinations thereof. In addition, the memory medium may be located in a first computer system in which the programs are executed, or may be located in a second different computer system which connects to the first computer system over a network, such as the Internet. In the latter instance, the second computer system may provide program instructions to the first computer for execution. The term “memory medium” may include two or more memory mediums which may reside in different locations, e.g., in different computer systems that are connected over a network. The memory medium may store program instructions (e.g., embodied as computer programs) that may be executed by one or more processors.

[0032] Carrier Medium – a memory medium as described above, as well as a physical transmission medium, such as a bus, network, and / or other physical transmission medium that conveys signals such as electrical, electromagnetic, or digital signals.

[0033] Programmable Hardware Element includes various hardware devicesClient Ref. No. P62335WO1 comprising multiple programmable function blocks connected via a programmable interconnect. Examples include FPGAs (Field Programmable Gate Arrays), PLDs (Programmable Logic Devices), FPOAs (Field Programmable Object Arrays), and CPLDs (Complex PLDs). The programmable function blocks may range from fine grained (combinatorial logic or look up tables) to coarse grained (arithmetic logic units or processor cores). A programmable hardware element may also be referred to as "reconfigurable logic”.

[0034] Computer System (or Computer) – any of various types of computing or processing systems, including a personal computer system (PC), mainframe computer system, workstation, network appliance, Internet appliance, personal digital assistant (PDA), television system, grid computing system, or other device or combinations of devices. In general, the term "computer system" can be broadly defined to encompass any device (or combination of devices) having at least one processor that executes instructions from a memory medium.

[0035] User Equipment (UE) (or “UE Device”) – any of various types of computer systems devices which are mobile or portable and which performs wireless communications. Examples of UE devices include mobile telephones or smart phones (e.g., iPhone™, Android™-based phones), portable gaming devices (e.g., Nintendo DS™, PlayStation Portable™, Gameboy Advance™, iPhone™), laptops, wearable devices (e.g., smart watch, smart glasses), PDAs, portable Internet devices, Internet of Things, music players, data storage devices, other handheld devices, unmanned aerial vehicles (UAVs) (e.g., drones), UAV controllers (UACs), and so forth. In general, the term “UE” or “UE device” can be broadly defined to encompass any electronic, computing, and / or telecommunications device (or combination of devices) which is easily transported by a user and capable of wireless communication.

[0036] Base Station – The term “Base Station” has the full breadth of its ordinary meaning, and at least includes a wireless communication station installed at a fixed location and used to communicate with UEs as part of a wireless telephone system or radio system, including but not limited Next Generation Node-Bs (gNB or gNodeB) in NR and NG-RAN nodes.Client Ref. No. P62335WO1

[0037] Processing Element (or Processor) – refers to various elements or combinations of elements that are capable of performing a function in a device, such as a user equipment or a cellular network device. Processing elements may include, for example: processors and associated memory, portions or circuits of individual processor cores, entire processor cores, processor arrays, circuits such as an ASIC (Application Specific Integrated Circuit), programmable hardware elements such as a field programmable gate array (FPGA), as well any of various combinations of the above.

[0038] Channel - a medium used to convey information from a sender (transmitter) to a receiver. It should be noted that since characteristics of the term “channel” may differ according to different wireless protocols, the term “channel” as used herein may be considered as being used in a manner that is consistent with the standard of the type of device with reference to which the term is used. In some standards, channel widths may be variable (e.g., depending on device capability, band conditions, etc.). For example, LTE may support scalable channel bandwidths from 1.4 MHz to 20MHz. 5G NR can support scalable channel bandwidths from 5 MHz to 100 MHz in Frequency Range 1 (FR1) and up to 400 MHz in FR2. In other radio access technologies, WLAN channels may be 22 MHz wide while Bluetooth channels may be 1 MHz wide. Other protocols and standards may include different definitions of channels. Furthermore, some standards may define and use multiple types of channels, e.g., different channels for uplink or downlink and / or different channels for different uses such as data, control information, etc.

[0039] Band - The term "band" has the full breadth of its ordinary meaning, and at least includes a section of spectrum (e.g., radio frequency spectrum) in which channels are used or set aside for the same purpose.

[0040] Automatically – refers to an action or operation performed by a computer system (e.g., software executed by the computer system) or device (e.g., circuitry, programmable hardware elements, ASICs, etc.), without user input directly specifying or performing the action or operation. Thus, the term "automatically" is in contrast to an operation being manually performed or specifiedClient Ref. No. P62335WO1 by the user, where the user provides input to directly perform the operation. An automatic procedure may be initiated by input provided by the user, but the subsequent actions that are performed “automatically” are not specified by the user, i.e., are not performed “manually”, where the user specifies each action to perform. For example, a user filling out an electronic form by selecting each field and providing input specifying information (e.g., by typing information, selecting check boxes, radio selections, etc.) is filling out the form manually, even though the computer system will update the form in response to the user actions. The form may be automatically filled out by the computer system where the computer system (e.g., software executing on the computer system) analyzes the fields of the form and fills in the form without any user input specifying the answers to the fields. As indicated above, the user may invoke the automatic filling of the form, but is not involved in the actual filling of the form (e.g., the user is not manually specifying answers to fields but rather they are being automatically completed). The present specification provides various examples of operations being automatically performed in response to actions the user has taken.

[0041] Approximately - refers to a value that is almost correct or exact. For example, approximately may refer to a value that is within 1 to 10 percent of the exact (or desired) value. It should be noted, however, that the actual threshold value (or tolerance) may be application dependent. For example, in some embodiments, “approximately” may mean within 0.1% of some specified or desired value, while in various other embodiments, the threshold may be, for example, 2%, 3%, 5%, and so forth, as desired or as set by the particular application.

[0042] Concurrent – refers to parallel execution or performance, where tasks, processes, or programs are performed in an at least partially overlapping manner. For example, concurrency may be implemented using “strong” or strict parallelism, where tasks are performed (at least partially) in parallel on respective computational elements, or using “weak parallelism”, where the tasks are performed in an interleaved manner, e.g., by time multiplexing of execution threads.

[0043] Legacy - The 3rd Generation Partnership Project (3GPP) producesClient Ref. No. P62335WO1 specifications that define 3GPP technologies. 3GPP specifications cover cellular telecommunications technologies, including radio access, core network and service capabilities, which provide a complete system description for mobile telecommunications. 3GPP uses a system of parallel “Releases” that provide developers with a stable platform for the implementation of features at a given point and then allow for the addition of new functionality in subsequent releases. Release 17 was released in 2022. Release 18 (Rel-18), at the time of this disclosure, is nearing release on June 22, 2024, as its specifications have been largely defined. Accordingly, implementations and concepts compatible with Rel-18, or previous Releases, are sometimes referred to herein as “Legacy Releases.” One or more embodiments of the present disclosure may be adopted in future Releases, e.g., Release 19.

[0044] Various components may be described as “configured to” perform a task or tasks. In such contexts, “configured to” is a broad recitation generally meaning “having structure that” performs the task or tasks during operation. As such, the component can be configured to perform the task even when the component is not currently performing that task (e.g., a set of electrical conductors may be configured to electrically connect a module to another module, even when the two modules are not connected). In some contexts, “configured to” may be a broad recitation of structure generally meaning “having circuitry that” performs the task or tasks during operation. As such, the component can be configured to perform the task even when the component is not currently on. In general, the circuitry that forms the structure corresponding to “configured to” may include hardware circuits.

[0045] Various components may be described as performing a task or tasks, for convenience in the description. Such descriptions should be interpreted as including the phrase “configured to.” Reciting a component that is configured to perform one or more tasks is expressly intended not to invoke 35 U.S.C. § 112(f) interpretation for that component.

[0046] The example embodiments may be further understood with reference to the following description and the related appended drawings, wherein like elements are provided with the same reference numerals. The example embodiments relateClient Ref. No. P62335WO1 to apparatuses, systems and method for reducing energy usage by network components, e.g., base stations in wireless communication systems.

[0047] The example embodiments are described with regard to communication between a Next Generation Node B (gNB) and a user equipment (UE). However, reference to a gNB or a UE is merely provided for illustrative purposes. The example embodiments may be utilized with any electronic component that may establish a connection to a network and is configured with the hardware, software, and / or firmware to support for reducing energy usage by network components in wireless communication systems. Therefore, the gNB or UE as described herein is used to represent any appropriate type of electronic component.

[0048] The example embodiments are also described with regard to a fifth generation (5G) New Radio (NR). However, reference to a 5G NR network is merely provided for illustrative purposes. The example embodiments may be utilized with any appropriate type of network.

[0049] Throughout this description various information elements (IEs) are referred to by specific names. It should be understood that these names are only examples and the IEs carrying the information referred to throughout this description may be referred to by other names by various entities. Figures 1A and 1B: Communication Systems

[0050] FIG.1A illustrates a simplified example wireless communication system, according to some embodiments. It is noted that the system of FIG.1A is merely one example of a possible system, and that features of this disclosure may be implemented in any of various systems, as desired.

[0051] As shown, the example wireless communication system includes a base station 102A which communicates over a transmission medium with one or more user devices 106A, 106B, etc., through 106N. Each of the user devices may be referred to herein as a “user equipment” (UE). Thus, the user devices 106 are referred to as UEs or UE devices.Client Ref. No. P62335WO1

[0052] The base station (BS) 102A may be a base transceiver station (BTS) or cell site (a “cellular base station”) and may include hardware that enables wireless communication with the UEs 106A through 106N.

[0053] The communication area (or coverage area) of the base station may be referred to as a “cell.” The base station 102A and the UEs 106 may be configured to communicate over the transmission medium using any of various radio access technologies (RATs), also referred to as wireless communication technologies, or telecommunication standards, such as GSM, UMTS (associated with, for example, WCDMA or TD-SCDMA air interfaces), LTE, LTE-Advanced (LTE-A), 5G new radio (5G NR), HSPA, 3GPP2 CDMA2000 (e.g., 1xRTT, 1xEV-DO, HRPD, eHRPD), etc. Note that if the base station 102A is implemented in the context of LTE, also referred to as the Evolved Universal Terrestrial Radio Access Network (E-UTRAN, it may alternately be referred to as an 'eNodeB' or ‘eNB’. Note that if the base station 102A is implemented in the context of 5G NR, it may alternately be referred to as ‘gNodeB’ or ‘gNB’.

[0054] As shown, the base station 102A may also be equipped to communicate with a network 100 (e.g., a core network of a cellular service provider, a telecommunication network such as a public switched telephone network (PSTN), and / or the Internet, among various possibilities). Thus, the base station 102A may facilitate communication between the user devices and / or between the user devices and the network 100. In particular, the cellular base station 102A may provide UEs 106 with various telecommunication capabilities, such as voice, SMS and / or data services.

[0055] Base station 102A and other similar base stations (such as base stations 102B…102N) operating according to the same or a different cellular communication standard may thus be provided as a network of cells, which may provide continuous or nearly continuous overlapping service to UEs 106A-N and similar devices over a geographic area via one or more cellular communication standards.

[0056] Thus, while base station 102A may act as a “serving cell” for UEs 106A-Client Ref. No. P62335WO1 N as illustrated in FIG.1A, each UE 106 may also be capable of receiving signals from (and possibly within communication range of) one or more other cells (which might be provided by base stations 102B-N and / or any other base stations), which may be referred to as “neighboring cells”. Such cells may also be capable of facilitating communication between user devices and / or between user devices and the network 100. Such cells may include “macro” cells, “micro” cells, “pico” cells, and / or cells which provide any of various other granularities of service area size. For example, base stations 102A-B illustrated in FIG. 1A might be macro cells, while base station 102N might be a micro cell. Other configurations are also possible.

[0057] In some embodiments, base station 102A may be a next generation base station, e.g., a 5G New Radio (5G NR) base station, or “gNB”. In some embodiments, a gNB may be connected to a legacy evolved packet core (EPC) network and / or to a NR core (NRC) network. In addition, a gNB cell may include one or more transition and reception points (TRPs). In addition, a UE capable of operating according to 5G NR may be connected to one or more TRPs within one or more gNBs.

[0058] Note that a UE 106 may be capable of communicating using multiple wireless communication standards. For example, the UE 106 may be configured to communicate using a wireless networking (e.g., Wi-Fi) and / or peer-to-peer wireless communication protocol (e.g., Bluetooth, Wi-Fi peer-to-peer, etc.) in addition to at least one cellular communication protocol (e.g., GSM, UMTS (associated with, for example, WCDMA or TD-SCDMA air interfaces), LTE, LTE-A, 5G NR, HSPA, 3GPP2 CDMA2000 (e.g., 1xRTT, 1xEV-DO, HRPD, eHRPD), etc.). The UE 106 may also or alternatively be configured to communicate using one or more global navigational satellite systems (GNSS, e.g., GPS or GLONASS), one or more mobile television broadcasting standards (e.g., ATSC-M / H or DVB-H), and / or any other wireless communication protocol, if desired. Other combinations of wireless communication standards (including more than two wireless communication standards) are also possible.

[0059] In some embodiments, the base station 102A may select a pagingClient Ref. No. P62335WO1 configuration and a PEI configuration for UEs 106. The base station 102A may encode and transmit the paging configuration and the PEI configuration to UEs 106 as part of a registration process. Using the paging configuration, UEs 106 can determine which PO and PF to monitor in a paging cycle. Using the PEI configuration, UEs 106 can determine the radio frame that carries relevant PEI.

[0060] FIG. 1B illustrates user equipment 106 (e.g., one of the devices 106A through 106N) in communication with a base station 102 and an access point 112, according to some embodiments. The UE 106 may be a device with both cellular communication capability and non-cellular communication capability (e.g., Bluetooth, Wi-Fi, and so forth) such as a mobile phone, a hand-held device, a computer or a tablet, or virtually any type of wireless device.

[0061] The UE 106 may include a processor that is configured to execute program instructions stored in memory. The UE 106 may perform any of the method embodiments described herein by executing such stored instructions. Alternatively, or in addition, the UE 106 may include a programmable hardware element such as an FPGA (field-programmable gate array) that is configured to perform any of the method embodiments described herein, or any portion of any of the method embodiments described herein.

[0062] The UE 106 may include one or more antennas for communicating using one or more wireless communication protocols or technologies. In some embodiments, the UE 106 may be configured to communicate using, for example, CDMA2000 (1xRTT / 1xEV-DO / HRPD / eHRPD), LTE / LTE-Advanced, or 5G NR using a single shared radio and / or GSM, LTE, LTE-Advanced, or 5G NR using the single shared radio. The shared radio may couple to a single antenna, or may couple to multiple antennas (e.g., for MIMO) for performing wireless communications. In general, a radio may include any combination of a baseband processor, analog RF signal processing circuitry (e.g., including filters, mixers, oscillators, amplifiers, etc.), or digital processing circuitry (e.g., for digital modulation as well as other digital processing). Similarly, the radio may implement one or more receive and transmit chains using the aforementioned hardware. For example, the UE 106 may share one or more parts of a receive and / or transmit chain betweenClient Ref. No. P62335WO1 multiple wireless communication technologies, such as those discussed above.

[0063] In some embodiments, the UE 106 may include separate transmit and / or receive chains (e.g., including separate antennas and other radio components) for each wireless communication protocol with which it is configured to communicate. As a further possibility, the UE 106 may include one or more radios which are shared between multiple wireless communication protocols, and one or more radios which are used exclusively by a single wireless communication protocol. For example, the UE 106 might include a shared radio for communicating using either of LTE or 5G NR (or LTE or 1xRTTor LTE or GSM), and separate radios for communicating using each of Wi-Fi and Bluetooth. Other configurations are also possible. FIG.2: Block Diagram of a Base Station (gNB)

[0064] FIG. 2 illustrates an example block diagram of a base station 102, according to some embodiments. It is noted that the base station of FIG.2 is merely one example of a possible base station. As shown, the base station 102 may include processor(s) 204 which may execute program instructions for the base station 102. The processor(s) 204 may also be coupled to memory management unit (MMU) 240, which may be configured to receive addresses from the processor(s) 204 and translate those addresses to locations in memory (e.g., memory 260 and read only memory (ROM) 250) or to other circuits or devices.

[0065] The base station 102 may include at least one network port 270. The network port 270 may be configured to couple to a telephone network and provide a plurality of devices, such as UE devices 106, access to the telephone network as described above in Figures 1 and 2.

[0066] The network port 270 (or an additional network port) may also or alternatively be configured to couple to a cellular network, e.g., a core network of a cellular service provider. The core network may provide mobility related services and / or other services to a plurality of devices, such as UE devices 106. In some cases, the network port 270 may couple to a telephone network via the coreClient Ref. No. P62335WO1 network, and / or the core network may provide a telephone network (e.g., among other UE devices serviced by the cellular service provider).

[0067] In some embodiments, base station 102 may be a next generation base station, e.g., a 5G New Radio (5G NR) base station, or “gNB”. In such embodiments, base station 102 may be connected to a legacy evolved packet core (EPC) network and / or to a NR core (NRC) network. In addition, base station 102 may be considered a 5G NR cell and may include one or more transition and reception points (TRPs). In addition, a UE capable of operating according to 5G NR may be connected to one or more TRPs within one or more gNBs.

[0068] The base station 102 may include at least one antenna 234, and possibly multiple antennas. The at least one antenna 234 may be configured to operate as a wireless transceiver and may be further configured to communicate with UE devices 106 via radio 230. The antenna 234 communicates with the radio 230 via communication chain 232. Communication chain 232 may be a receive chain, a transmit chain or both. The radio 230 may be configured to communicate via various wireless communication standards, including, but not limited to, 5G NR, LTE, LTE-A, GSM, UMTS, CDMA2000, Wi-Fi, etc.

[0069] The base station 102 may be configured to communicate wirelessly using multiple wireless communication standards. In some instances, the base station 102 may include multiple radios, which may enable the base station 102 to communicate according to multiple wireless communication technologies. For example, as one possibility, the base station 102 may include an LTE radio for performing communication according to LTE as well as a 5G NR radio for performing communication according to 5G NR. In such a case, the base station 102 may be capable of operating as both an LTE base station and a 5G NR base station. As another possibility, the base station 102 may include a multi-mode radio which is capable of performing communications according to any of multiple wireless communication technologies (e.g., 5G NR and Wi-Fi, LTE and Wi-Fi, LTE and UMTS, LTE and CDMA2000, UMTS and GSM, etc.).

[0070] As described further subsequently herein, the base station 102 mayClient Ref. No. P62335WO1 include hardware and software components for implementing or supporting implementation of features described herein. The processor 204 of the base station 102 may be configured to implement or support implementation of part or all of the methods described herein, e.g., by executing program instructions stored on a memory medium (e.g., a non-transitory computer-readable memory medium). Alternatively, the processor 204 may be configured as a programmable hardware element, such as an FPGA (Field Programmable Gate Array), or as an ASIC (Application Specific Integrated Circuit), or a combination thereof. Alternatively (or in addition) the processor 204 of the base station 102, in conjunction with one or more of the other components 230, 232, 234, 240, 250, 260, 270 may be configured to implement or support implementation of part or all of the features described herein.

[0071] In addition, as described herein, processor(s) 204 may be comprised of one or more processing elements. In other words, one or more processing elements may be included in processor(s) 204. Thus, processor(s) 204 may include one or more integrated circuits (ICs) that are configured to perform the functions of processor(s) 204. In addition, each integrated circuit may include circuitry (e.g., first circuitry, second circuitry, etc.) configured to perform the functions of processor(s) 204.

[0072] Further, as described herein, radio 230 may be comprised of one or more processing elements. In other words, one or more processing elements may be included in radio 230. Thus, radio 230 may include one or more integrated circuits (ICs) that are configured to perform the functions of radio 230. In addition, each integrated circuit may include circuitry (e.g., first circuitry, second circuitry, etc.) configured to perform the functions of radio 230. FIG.3: Block Diagram of a Server

[0073] FIG. 3 illustrates an example block diagram of a server 104, according to some embodiments. It is noted that the server of FIG.3 is merely one example of a possible server. As shown, the server 104 may include processor(s)Client Ref. No. P62335WO1 344 which may execute program instructions for the server 104. The processor(s) 344 may also be coupled to memory management unit (MMU) 374, which may be configured to receive addresses from the processor(s) 344 and translate those addresses to locations in memory (e.g., memory 364 and read only memory (ROM) 354) or to other circuits or devices.

[0074] The server 104 may be configured to provide a plurality of devices, such as base station 102, and UE devices 106 access to network functions, e.g., as further described herein.

[0075] In some embodiments, the server 104 may be part of a radio access network, such as a 5G New Radio (5G NR) radio access network. In some embodiments, the server 104 may be connected to a legacy evolved packet core (EPC) network and / or to a NR core (NRC) network.

[0076] As described herein, the server 104 may include hardware and software components for implementing or supporting implementation of features described herein. The processor 344 of the server 104 may be configured to implement or support implementation of part or all of the methods described herein, e.g., by executing program instructions stored on a memory medium (e.g., a non-transitory computer-readable memory medium). Alternatively, the processor 344 may be configured as a programmable hardware element, such as an FPGA (Field Programmable Gate Array), or as an ASIC (Application Specific Integrated Circuit), or a combination thereof. Alternatively (or in addition) the processor 344 of the server 104, in conjunction with one or more of the other components 354, 364, and / or 374 may be configured to implement or support implementation of part or all of the features described herein.

[0077] In addition, as described herein, processor(s) 344 may be comprised of one or more processing elements. In other words, one or more processing elements may be included in processor(s) 344. Thus, processor(s) 344 may include one or more integrated circuits (ICs) that are configured to perform the functions of processor(s) 344. In addition, each integrated circuit may include circuitry (e.g., first circuitry, second circuitry, etc.) configured to perform the functions of processor(s)Client Ref. No. P62335WO1 344. FIG.4: Block Diagram of a User Equipment (UE)

[0078] FIG. 4 illustrates an example simplified block diagram of a communication device 106, according to some embodiments. It is noted that the block diagram of the communication device of FIG. 4 is only one example of a possible communication device. According to embodiments, communication device 106 may be a user equipment (UE) device, a mobile device or mobile station, a wireless device or wireless station, a desktop computer or computing device, a mobile computing device (e.g., a laptop, notebook, or portable computing device), a tablet, an unmanned aerial vehicle (UAV), a UAV controller (UAC) and / or a combination of devices, among other devices. As shown, the communication device 106 may include a set of components 400 configured to perform core functions. For example, this set of components may be implemented as a system on chip (SOC), which may include portions for various purposes. Alternatively, this set of components 400 may be implemented as separate components or groups of components for the various purposes. The set of components 400 may be coupled (e.g., communicatively; directly or indirectly) to various other circuits of the communication device 106.

[0079] For example, the communication device 106 may include various types of memory (e.g., including NAND flash 410), an input / output interface such as connector I / F 420 (e.g., for connecting to a computer system; dock; charging station; input devices, such as a microphone, camera, keyboard; output devices, such as speakers; etc.), the display 460, which may be integrated with or external to the communication device 106, and cellular communication circuitry 430 such as for 5G NR, LTE, GSM, etc., and short to medium range wireless communication circuitry 429 (e.g., Bluetooth™ and WLAN circuitry). In some embodiments, communication device 106 may include wired communication circuitry (not shown), such as a network interface card, e.g., for Ethernet.

[0080] The cellular communication circuitry 430 may couple (e.g.,Client Ref. No. P62335WO1 communicatively; directly or indirectly) to one or more antennas, such as antennas 435 and 436 as shown. The short to medium range wireless communication circuitry 429 may also couple (e.g., communicatively; directly or indirectly) to one or more antennas, such as antennas 437 and 438 as shown. Alternatively, the short to medium range wireless communication circuitry 429 may couple (e.g., communicatively; directly or indirectly) to the antennas 435 and 436 in addition to, or instead of, coupling (e.g., communicatively; directly or indirectly) to the antennas 437 and 438. The short to medium range wireless communication circuitry 429 and / or cellular communication circuitry 430 may include multiple receive chains and / or multiple transmit chains for receiving and / or transmitting multiple spatial streams, such as in a multiple-input multiple output (MIMO) configuration.

[0081] In some embodiments, as further described below, cellular communication circuitry 430 may include dedicated receive chains (including and / or coupled to, e.g., communicatively; directly or indirectly. dedicated processors and / or radios) for multiple RATs (e.g., a first receive chain for LTE and a second receive chain for 5G NR). In addition, in some embodiments, cellular communication circuitry 430 may include a single transmit chain that may be switched between radios dedicated to specific RATs. For example, a first radio may be dedicated to a first RAT, e.g., LTE, and may be in communication with a dedicated receive chain and a transmit chain shared with an additional radio, e.g., a second radio that may be dedicated to a second RAT, e.g., 5G NR, and may be in communication with a dedicated receive chain and the shared transmit chain.

[0082] The communication device 106 may also include and / or be configured for use with one or more user interface elements. The user interface elements may include any of various elements, such as display 460 (which may be a touchscreen display), a keyboard (which may be a discrete keyboard or may be implemented as part of a touchscreen display), a mouse, a microphone and / or speakers, one or more cameras, one or more buttons, and / or any of various other elements capable of providing information to a user and / or receiving or interpreting user input.

[0083] The communication device 106 may further include one or more smart cards 445 that include SIM (Subscriber Identity Module) functionality, such as oneClient Ref. No. P62335WO1 or more UICC(s) (Universal Integrated Circuit Card(s)) cards 445. Note that the term “SIM” or “SIM entity” is intended to include any of various types of SIM implementations or SIM functionality, such as the one or more UICC(s) cards 445, one or more eUICCs, one or more eSIMs, either removable or embedded, etc. In some embodiments, the UE 106 may include at least two SIMs. Each SIM may execute one or more SIM applications and / or otherwise implement SIM functionality. Thus, each SIM may be a single smart card that may be embedded, e.g., may be soldered onto a circuit board in the UE 106, or each SIM 410 may be implemented as a removable smart card. Thus, the SIM(s) may be one or more removable smart cards (such as UICC cards, which are sometimes referred to as “SIM cards”), and / or the SIMs 410 may be one or more embedded cards (such as embedded UICCs (eUICCs), which are sometimes referred to as “eSIMs” or “eSIM cards”). In some embodiments (such as when the SIM(s) include an eUICC), one or more of the SIM(s) may implement embedded SIM (eSIM) functionality; in such an embodiment, a single one of the SIM(s) may execute multiple SIM applications. Each of the SIMs may include components such as a processor and / or a memory; instructions for performing SIM / eSIM functionality may be stored in the memory and executed by the processor. In some embodiments, the UE 106 may include a combination of removable smart cards and fixed / non-removable smart cards (such as one or more eUICC cards that implement eSIM functionality), as desired. For example, the UE 106 may comprise two embedded SIMs, two removable SIMs, or a combination of one embedded SIMs and one removable SIMs. Various other SIM configurations are also contemplated.

[0084] As noted above, in some embodiments, the UE 106 may include two or more SIMs. The inclusion of two or more SIMs in the UE 106 may allow the UE 106 to support two different telephone numbers and may allow the UE 106 to communicate on corresponding two or more respective networks. For example, a first SIM may support a first RAT such as LTE, and a second SIM 410 support a second RAT such as 5G NR. Other implementations and RATs are of course possible. In some embodiments, when the UE 106 comprises two SIMs, the UE 106 may support Dual SIM Dual Active (DSDA) functionality. The DSDAClient Ref. No. P62335WO1 functionality may allow the UE 106 to be simultaneously connected to two networks (and use two different RATs) at the same time, or to simultaneously maintain two connections supported by two different SIMs using the same or different RATs on the same or different networks. The DSDA functionality may also allow the UE 106 to simultaneously receive voice calls or data traffic on either phone number. In certain embodiments the voice call may be a packet switched communication. In other words, the voice call may be received using voice over LTE (VoLTE) technology and / or voice over NR (VoNR) technology. In some embodiments, the UE 106 may support Dual SIM Dual Standby (DSDS) functionality. The DSDS functionality may allow either of the two SIMs in the UE 106 to be on standby waiting for a voice call and / or data connection. In DSDS, when a call / data is established on one SIM, the other SIM is no longer active. In some embodiments, DSDx functionality (either DSDA or DSDS functionality) may be implemented with a single SIM (e.g., a eUICC) that executes multiple SIM applications for different carriers and / or RATs.

[0085] As shown, the SOC 400 may include processor(s) 402, which may execute program instructions for the communication device 106 and display circuitry 404, which may perform graphics processing and provide display signals to the display 460. The processor(s) 402 may also be coupled to memory management unit (MMU) 440, which may be configured to receive addresses from the processor(s) 402 and translate those addresses to locations in memory (e.g., memory 406, read only memory (ROM) 450, NAND flash memory 410) and / or to other circuits or devices, such as the display circuitry 404, short to medium range wireless communication circuitry 429, cellular communication circuitry 430, connector I / F 420, and / or display 460. The MMU 440 may be configured to perform memory protection and page table translation or set up. In some embodiments, the MMU 440 may be included as a portion of the processor(s) 402.

[0086] As described herein, the communication device 106 may include hardware and software components for implementing the above features for a communication device 106 to communicate a scheduling profile for power savings to a network. The processor 402 of the communication device 106 may beClient Ref. No. P62335WO1 configured to implement part or all of the features described herein, e.g., by executing program instructions stored on a memory medium (e.g., a non-transitory computer-readable memory medium). Alternatively (or in addition), processor 402 may be configured as a programmable hardware element, such as an FPGA (Field Programmable Gate Array), or as an ASIC (Application Specific Integrated Circuit). Alternatively (or in addition) the processor 402 of the communication device 106, in conjunction with one or more of the other components 400, 404, 406, 410, 420, 429, 430, 440, 445, 450, 460 may be configured to implement part or all of the features described herein.

[0087] In addition, as described herein, processor 402 may include one or more processing elements. Thus, processor 402 may include one or more integrated circuits (ICs) that are configured to perform the functions of processor 402. In addition, each integrated circuit may include circuitry (e.g., first circuitry, second circuitry, etc.) configured to perform the functions of processor(s) 402.

[0088] Further, as described herein, cellular communication circuitry 430 and short to medium range wireless communication circuitry 429 may each include one or more processing elements. In other words, one or more processing elements may be included in cellular communication circuitry 430 and, similarly, one or more processing elements may be included in short to medium range wireless communication circuitry 429. Thus, cellular communication circuitry 430 may include one or more integrated circuits (ICs) that are configured to perform the functions of cellular communication circuitry 430. In addition, each integrated circuit may include circuitry (e.g., first circuitry, second circuitry, etc.) configured to perform the functions of cellular communication circuitry 430. Similarly, the short to medium range wireless communication circuitry 429 may include one or more ICs that are configured to perform the functions of short to medium range wireless communication circuitry 429. In addition, each integrated circuit may include circuitry (e.g., first circuitry, second circuitry, etc.) configured to perform the functions of short to medium range wireless communication circuitry 429.Client Ref. No. P62335WO1 FIG.5: Block Diagram of Cellular Communication Circuitry

[0089] FIG. 5 illustrates an example simplified block diagram of cellular communication circuitry, according to some embodiments. It is noted that the block diagram of the cellular communication circuitry of FIG.5 is only one example of a possible cellular communication circuit. According to embodiments, cellular communication circuitry 530, which may be cellular communication circuitry 430, may be included in a communication device, such as communication device 106 described above. As noted above, communication device 106 may be a user equipment (UE) device, a mobile device or mobile station, a wireless device or wireless station, a desktop computer or computing device, a mobile computing device (e.g., a laptop, notebook, or portable computing device), a tablet and / or a combination of devices, among other devices.

[0090] The cellular communication circuitry 530 may couple (e.g., communicatively; directly or indirectly) to one or more antennas, such as antennas 435a-b and 436 as shown (in FIG. 4). In some embodiments, cellular communication circuitry 530 may include dedicated receive chains (including and / or coupled to, e.g., communicatively; directly or indirectly. dedicated processors and / or radios) for multiple RATs (e.g., a first receive chain for LTE and a second receive chain for 5G NR). For example, as shown in FIG. 5, cellular communication circuitry 530 may include a modem 510 and a modem 520. Modem 510 may be configured for communications according to a first RAT, e.g., such as LTE or LTE-A, and modem 520 may be configured for communications according to a second RAT, e.g., such as 5G NR.

[0091] As shown, modem 510 may include one or more processors 512 and a memory 516 in communication with processors 512. Modem 510 may be in communication with a radio frequency (RF) front end 535. RF front end 535 may include circuitry for transmitting and receiving radio signals. For example, RF front end 535 may include receive circuitry (RX) 532 and transmit circuitry (TX) 534. In some embodiments, receive circuitry 532 may be in communication with downlink (DL) front end 550, which may include circuitry for receiving radio signals via antenna 335a.Client Ref. No. P62335WO1

[0092] Similarly, modem 520 may include one or more processors 522 and a memory 526 in communication with processors 522. Modem 520 may be in communication with an RF front end 540. RF front end 540 may include circuitry for transmitting and receiving radio signals. For example, RF front end 540 may include receive circuitry 542 and transmit circuitry 544. In some embodiments, receive circuitry 542 may be in communication with DL front end 560, which may include circuitry for receiving radio signals via antenna 335b.

[0093] In some embodiments, a switch 570 may couple transmit circuitry 534 to uplink (UL) front end 572. In addition, switch 570 may couple transmit circuitry 544 to UL front end 572. UL front end 572 may include circuitry for transmitting radio signals via antenna 336. Thus, when cellular communication circuitry 530 receives instructions to transmit according to the first RAT (e.g., as supported via modem 510), switch 570 may be switched to a first state that allows modem 510 to transmit signals according to the first RAT (e.g., via a transmit chain that includes transmit circuitry 534 and UL front end 572). Similarly, when cellular communication circuitry 530 receives instructions to transmit according to the second RAT (e.g., as supported via modem 520), switch 570 may be switched to a second state that allows modem 520 to transmit signals according to the second RAT (e.g., via a transmit chain that includes transmit circuitry 544 and UL front end 572).

[0094] As described herein, the modem 510 may include hardware and software components for implementing the above features or for time division multiplexing UL data for NSA NR operations, as well as the various other techniques described herein. The processors 512 may be configured to implement part or all of the features described herein, e.g., by executing program instructions stored on a memory medium (e.g., a non-transitory computer-readable memory medium). Alternatively (or in addition), processor 512 may be configured as a programmable hardware element, such as an FPGA (Field Programmable Gate Array), or as an ASIC (Application Specific Integrated Circuit). Alternatively (or in addition) the processor 512, in conjunction with one or more of the other components 530, 532, 534, 535, 550, 570, 572, 335a, 335b, and 336 may be configured to implement part or all of the features described herein.Client Ref. No. P62335WO1

[0095] In addition, as described herein, processors 512 may include one or more processing elements. Thus, processors 512 may include one or more integrated circuits (ICs) that are configured to perform the functions of processors 512. In addition, each integrated circuit may include circuitry (e.g., first circuitry, second circuitry, etc.) configured to perform the functions of processors 512.

[0096] The processors 522 may be configured to implement part or all of the features described herein, e.g., by executing program instructions stored on a memory medium (e.g., a non-transitory computer-readable memory medium). Alternatively (or in addition), processor 522 may be configured as a programmable hardware element, such as an FPGA (Field Programmable Gate Array), or as an ASIC (Application Specific Integrated Circuit). Alternatively (or in addition) the processor 522, in conjunction with one or more of the other components 540, 542, 544, 550, 570, 572, 335a, 335b, and 336 may be configured to implement part or all of the features described herein.

[0097] In addition, as described herein, processors 522 may include one or more processing elements. Thus, processors 522 may include one or more integrated circuits (ICs) that are configured to perform the functions of processors 522. In addition, each integrated circuit may include circuitry (e.g., first circuitry, second circuitry, etc.) configured to perform the functions of processors 522. FIG.6: Block Diagram of a Baseband Processor Architecture for a UE

[0098] FIG. 6 illustrates example components of a device 600 in accordance with some embodiments. It is noted that the device of FIG.6 is merely one example of a possible system, and that features of this disclosure may be implemented in any of various UEs, as desired.

[0099] In some embodiments, the device 600 may include application circuitry 602, baseband circuitry 604, Radio Frequency (RF) circuitry 606, front-end module (FEM) circuitry 608, one or more antennas 610, and power management circuitry (PMC) 612 coupled together at least as shown. The components of the illustrated device 600 may be included in a UE 106 or a RAN node 102A. In someClient Ref. No. P62335WO1 embodiments, the device 600 may include less elements (e.g., a RAN node may not utilize application circuitry 602, and instead include a processor / controller to process IP data received from an EPC). In some embodiments, the device 600 may include additional elements such as, for example, memory / storage, display, camera, sensor, or input / output (I / O) interface. In other embodiments, the components described below may be included in more than one device (e.g., said circuitries may be separately included in more than one device for Cloud-RAN (C- RAN) implementations).

[0100] The application circuitry 602 may include one or more application processors. For example, the application circuitry 602 may include circuitry such as, but not limited to, one or more single-core or multi-core processors. The processor(s) may include any combination of general-purpose processors and dedicated processors (e.g., graphics processors, application processors, etc.). The processors may be coupled with or may include memory / storage and may be configured to execute instructions stored in the memory / storage to enable various applications or operating systems to run on the device 600. In some embodiments, processors of application circuitry 602 may process IP data packets received from an EPC.

[0101] The baseband circuitry 604 may include circuitry such as, but not limited to, one or more single-core or multi-core processors. The baseband circuitry 604 may include one or more baseband processors or control logic to process baseband signals received from a receive signal path of the RF circuitry 606 and to generate baseband signals for a transmit signal path of the RF circuitry 606. Baseband processing circuity 604 may interface with the application circuitry 602 for generation and processing of the baseband signals and for controlling operations of the RF circuitry 606. For example, in some embodiments, the baseband circuitry 604 may include a third generation (3G) baseband processor 604A, a fourth generation (4G) baseband processor 604B, a fifth generation (5G) baseband processor 604C, or other baseband processor(s) 604D for other existing generations, generations in development or to be developed in the future (e.g., second generation (2G), sixth generation (6G), etc.). The baseband circuitry 604Client Ref. No. P62335WO1 (e.g., one or more of baseband processors 604A-D) may handle various radio control functions that enable communication with one or more radio networks via the RF circuitry 606. In other embodiments, some or all of the functionality of baseband processors 604A-D may be included in modules stored in the memory 604G and executed via a Central Processing Unit (CPU) 604E. The radio control functions may include, but are not limited to, signal modulation / demodulation, encoding / decoding, radio frequency shifting, etc. In some embodiments, modulation / demodulation circuitry of the baseband circuitry 604 may include Fast- Fourier Transform (FFT), precoding, or constellation mapping / demapping functionality. In some embodiments, encoding / decoding circuitry of the baseband circuitry 604 may include convolution, tail-biting convolution, turbo, Viterbi, or Low Density Parity Check (LDPC) encoder / decoder functionality. Embodiments of modulation / demodulation and encoder / decoder functionality are not limited to these examples and may include other suitable functionality in other embodiments.

[0102] In some embodiments, the baseband circuitry 604 may include one or more audio digital signal processor(s) (DSP) 604F. The audio DSP(s) 604F may be include elements for compression / decompression and echo cancellation and may include other suitable processing elements in other embodiments. Components of the baseband circuitry may be suitably combined in a single chip, a single chipset, or disposed on a same circuit board in some embodiments. In some embodiments, some or all of the constituent components of the baseband circuitry 604 and the application circuitry 602 may be implemented together such as, for example, on a system on a chip (SOC).

[0103] In some embodiments, the baseband circuitry 604 may provide for communication compatible with one or more radio technologies. For example, in some embodiments, the baseband circuitry 604 may support communication with an evolved universal terrestrial radio access network (EUTRAN) or other wireless metropolitan area networks (WMAN), a wireless local area network (WLAN), a wireless personal area network (WPAN). Embodiments in which the baseband circuitry 604 is configured to support radio communications of more than one wireless protocol may be referred to as multi-mode baseband circuitry.Client Ref. No. P62335WO1

[0104] RF circuitry 606 may enable communication with wireless networks using modulated electromagnetic radiation through a non-solid medium. In various embodiments, the RF circuitry 606 may include switches, filters, amplifiers, etc. to facilitate the communication with the wireless network. RF circuitry 606 may include a receive signal path which may include circuitry to down-convert RF signals received from the FEM circuitry 608 and provide baseband signals to the baseband circuitry 604. RF circuitry 606 may also include a transmit signal path which may include circuitry to up-convert baseband signals provided by the baseband circuitry 604 and provide RF output signals to the FEM circuitry 608 for transmission.

[0105] In some embodiments, the receive signal path of the RF circuitry 606 may include mixer circuitry 606a, amplifier circuitry 606b and filter circuitry 606c. In some embodiments, the transmit signal path of the RF circuitry 606 may include filter circuitry 606c and mixer circuitry 606a. RF circuitry 606 may also include synthesizer circuitry 606d for synthesizing a frequency for use by the mixer circuitry 606a of the receive signal path and the transmit signal path. In some embodiments, the mixer circuitry 606a of the receive signal path may be configured to down-convert RF signals received from the FEM circuitry 608 based on the synthesized frequency provided by synthesizer circuitry 606d. The amplifier circuitry 606b may be configured to amplify the down-converted signals and the filter circuitry 606c may be a low-pass filter (LPF) or band-pass filter (BPF) configured to remove unwanted signals from the down-converted signals to generate output baseband signals. Output baseband signals may be provided to the baseband circuitry 604 for further processing. In some embodiments, the output baseband signals may be zero-frequency baseband signals, although this is not a necessity. In some embodiments, mixer circuitry 606a of the receive signal path may comprise passive mixers, although the scope of the embodiments is not limited in this respect.

[0106] In some embodiments, the mixer circuitry 606a of the transmit signal path may be configured to up-convert input baseband signals based on the synthesized frequency provided by the synthesizer circuitry 606d to generate RFClient Ref. No. P62335WO1 output signals for the FEM circuitry 608. The baseband signals may be provided by the baseband circuitry 604 and may be filtered by filter circuitry 606c.

[0107] In some embodiments, the mixer circuitry 606a of the receive signal path and the mixer circuitry 606a of the transmit signal path may include two or more mixers and may be arranged for quadrature downconversion and upconversion, respectively. In some embodiments, the mixer circuitry 606a of the receive signal path and the mixer circuitry 606a of the transmit signal path may include two or more mixers and may be arranged for image rejection (e.g., Hartley image rejection). In some embodiments, the mixer circuitry 606a of the receive signal path and the mixer circuitry 606a may be arranged for direct downconversion and direct upconversion, respectively. In some embodiments, the mixer circuitry 606a of the receive signal path and the mixer circuitry 606a of the transmit signal path may be configured for super-heterodyne operation.

[0108] In some embodiments, the output baseband signals and the input baseband signals may be analog baseband signals, although the scope of the embodiments is not limited in this respect. In some alternate embodiments, the output baseband signals and the input baseband signals may be digital baseband signals. In these alternate embodiments, the RF circuitry 606 may include analog- to-digital converter (ADC) and digital-to-analog converter (DAC) circuitry and the baseband circuitry 604 may include a digital baseband interface to communicate with the RF circuitry 606.

[0109] In some dual-mode embodiments, a separate radio IC circuitry may be provided for processing signals for each spectrum, although the scope of the embodiments is not limited in this respect.

[0110] In some embodiments, the synthesizer circuitry 606d may be a fractional-N synthesizer or a fractional N / N+1 synthesizer, although the scope of the embodiments is not limited in this respect as other types of frequency synthesizers may be suitable. For example, synthesizer circuitry 606d may be a delta-sigma synthesizer, a frequency multiplier, or a synthesizer comprising a phase-locked loop with a frequency divider.Client Ref. No. P62335WO1

[0111] The synthesizer circuitry 606d may be configured to synthesize an output frequency for use by the mixer circuitry 606a of the RF circuitry 606 based on a frequency input and a divider control input. In some embodiments, the synthesizer circuitry 606d may be a fractional N / N+1 synthesizer.

[0112] In some embodiments, frequency input may be provided by a voltage controlled oscillator (VCO), although that is not a necessity. Divider control input may be provided by either the baseband circuitry 604 or the applications processor 602 depending on the desired output frequency. In some embodiments, a divider control input (e.g., N) may be determined from a look-up table based on a channel indicated by the applications processor 602.

[0113] Synthesizer circuitry 606d of the RF circuitry 606 may include a divider, a delay-locked loop (DLL), a multiplexer and a phase accumulator. In some embodiments, the divider may be a dual modulus divider (DMD) and the phase accumulator may be a digital phase accumulator (DPA). In some embodiments, the DMD may be configured to divide the input signal by either N or N+1 (e.g., based on a carry out) to provide a fractional division ratio. In some example embodiments, the DLL may include a set of cascaded, tunable, delay elements, a phase detector, a charge pump and a D-type flip-flop. In these embodiments, the delay elements may be configured to break a VCO period up into Nd equal packets of phase, where Nd is the number of delay elements in the delay line. In this way, the DLL provides negative feedback to help ensure that the total delay through the delay line is one VCO cycle.

[0114] In some embodiments, synthesizer circuitry 606d may be configured to generate a carrier frequency as the output frequency, while in other embodiments, the output frequency may be a multiple of the carrier frequency (e.g., twice the carrier frequency, four times the carrier frequency) and used in conjunction with quadrature generator and divider circuitry to generate multiple signals at the carrier frequency with multiple different phases with respect to each other. In some embodiments, the output frequency may be a LO frequency (fLO). In some embodiments, the RF circuitry 606 may include an IQ / polar converter.Client Ref. No. P62335WO1

[0115] FEM circuitry 608 may include a receive signal path which may include circuitry configured to operate on RF signals received from one or more antennas 610, amplify the received signals and provide the amplified versions of the received signals to the RF circuitry 606 for further processing. FEM circuitry 608 may also include a transmit signal path which may include circuitry configured to amplify signals for transmission provided by the RF circuitry 606 for transmission by one or more of the one or more antennas 610. In various embodiments, the amplification through the transmit or receive signal paths may be done solely in the RF circuitry 606, solely in the FEM 608, or in both the RF circuitry 606 and the FEM 608.

[0116] In some embodiments, the FEM circuitry 608 may include a TX / RX switch to switch between transmit mode and receive mode operation. The FEM circuitry may include a receive signal path and a transmit signal path. The receive signal path of the FEM circuitry may include an LNA to amplify received RF signals and provide the amplified received RF signals as an output (e.g., to the RF circuitry 606). The transmit signal path of the FEM circuitry 608 may include a power amplifier (PA) to amplify input RF signals (e.g., provided by RF circuitry 606), and one or more filters to generate RF signals for subsequent transmission (e.g., by one or more of the one or more antennas 610).

[0117] In some embodiments, the PMC 612 may manage power provided to the baseband circuitry 604. In particular, the PMC 612 may control power-source selection, voltage scaling, battery charging, or DC-to-DC conversion. The PMC 612 may often be included when the device 600 is capable of being powered by a battery, for example, when the device is included in a UE. The PMC 612 may increase the power conversion efficiency while providing desirable implementation size and heat dissipation characteristics.

[0118] While FIG. 6 shows the PMC 612 coupled only with the baseband circuitry 604, in other embodiments the PMC 612 may be additionally or alternatively coupled with, and perform similar power management operations for, other components such as, but not limited to, application circuitry 602, RF circuitry 606, or FEM 608.Client Ref. No. P62335WO1

[0119] In some embodiments, the PMC 612 may control, or otherwise be part of, various power saving mechanisms of the device 600. For example, if the device 600 is in a radio resource control_Connected (RRC_Connected) state, where it is still connected to the RAN node as it expects to receive traffic shortly, then it may enter a state known as Discontinuous Reception Mode (DRX) after a period of inactivity. During this state, the device 600 may power down for brief intervals of time and thus save power.

[0120] If there is no data traffic activity for an extended period of time, then the device 600 may transition off to an RRC_Idle state, where it disconnects from the network and does not perform operations such as channel quality feedback, handover, etc. The device 600 goes into a very low power state and it performs paging where, again, it periodically wakes up to listen to the network and then powers down at least portions of the device again. The device 600 may not receive data in this state. In order to receive data, it will transition back to an RRC_Connected state.

[0121] An additional power saving mode may allow a device to be unavailable to the network for periods longer than a paging interval (ranging from seconds to a few hours). During this time, the device is totally unreachable to the network and may power down completely. Any data sent during this time incurs a large delay and it is assumed the delay is acceptable. FIG.7: Block Diagram of an Interface of Baseband Circuitry

[0122] FIG.7 illustrates example interfaces of baseband circuitry in accordance with some embodiments. It is noted that the baseband circuitry of FIG.7 is merely one example of a possible circuitry, and that features of this disclosure may be implemented in any of various systems, as desired.

[0123] As discussed above, the baseband circuitry 604 of FIG.6 may comprise processors 604A-604E and a memory 604G utilized by said processors. Each of the processors 604A-604E may include a memory interface, 704A-704E, respectively, to send / receive data to / from the memory 604G.Client Ref. No. P62335WO1

[0124] The baseband circuitry 604 may further include one or more interfaces to communicatively couple to other circuitries / devices, such as a memory interface 712 (e.g., an interface to send / receive data to / from memory external to the baseband circuitry 604), an application circuitry interface 714 (e.g., an interface to send / receive data to / from the application circuitry 602 of FIG. 6), an RF circuitry interface 716 (e.g., an interface to send / receive data to / from RF circuitry 606 of FIG. 6), a wireless hardware connectivity interface 718 (e.g., an interface to send / receive data to / from Near Field Communication (NFC) components, Bluetooth® components (e.g., Bluetooth® Low Energy), Wi-Fi® components, and other communication components), and a power management interface 720 (e.g., an interface to send / receive power or control signals to / from the PMC 612. FIG.8: Control Plane Protocol Stack

[0125] FIG.8 is an illustration of a control plane protocol stack in accordance with some embodiments. In this embodiment, a control plane 800 is shown as a communications protocol stack between the UE 106a (or alternatively, the UE 106b), the RAN node 102A (or alternatively, the RAN node 102B), and the mobility management entity (MME) 621.

[0126] The PHY layer 801 may transmit or receive information used by the MAC layer 802 over one or more air interfaces. The PHY layer 801 may further perform link adaptation or adaptive modulation and coding (AMC), power control, cell search (e.g., for initial synchronization and handover purposes), and other measurements used by higher layers, such as the RRC layer 805. The PHY layer 801 may still further perform error detection on the transport channels, forward error correction (FEC) coding / decoding of the transport channels, modulation / demodulation of physical channels, interleaving, rate matching, mapping onto physical channels, and Multiple Input Multiple Output (MIMO) antenna processing.

[0127] The MAC layer 802 may perform mapping between logical channels and transport channels, multiplexing of MAC service data units (SDUs) from one orClient Ref. No. P62335WO1 more logical channels onto transport blocks (TB) to be delivered to PHY via transport channels, de-multiplexing MAC SDUs to one or more logical channels from transport blocks (TB) delivered from the PHY via transport channels, multiplexing MAC SDUs onto TBs, scheduling information reporting, error correction through hybrid automatic repeat request (HARQ), and logical channel prioritization.

[0128] The RLC layer 803 may operate in a plurality of modes of operation, including: Transparent Mode (TM), Unacknowledged Mode (UM), and Acknowledged Mode (AM). The RLC layer 803 may execute transfer of upper layer protocol data units (PDUs), error correction through automatic repeat request (ARQ) for AM data transfers, and concatenation, segmentation and reassembly of RLC SDUs for UM and AM data transfers. The RLC layer 803 may also execute re-segmentation of RLC data PDUs for AM data transfers, reorder RLC data PDUs for UM and AM data transfers, detect duplicate data for UM and AM data transfers, discard RLC SDUs for UM and AM data transfers, detect protocol errors for AM data transfers, and perform RLC re-establishment.

[0129] The PDCP layer 804 may execute header compression and decompression of IP data, maintain PDCP Sequence Numbers (SNs), perform in- sequence delivery of upper layer PDUs at re-establishment of lower layers, eliminate duplicates of lower layer SDUs at re-establishment of lower layers for radio bearers mapped on RLC AM, cipher and decipher control plane data, perform integrity protection and integrity verification of control plane data, control timer- based discard of data, and perform security operations (e.g., ciphering, deciphering, integrity protection, integrity verification, etc.).

[0130] The main services and functions of the RRC layer 805 may include broadcast of system information (e.g., included in Master Information Blocks (MIBs) or System Information Blocks (SIBs) related to the non-access stratum (NAS)), broadcast of system information related to the access stratum (AS), paging, establishment, maintenance and release of an RRC connection between the UE and E-UTRAN (e.g., RRC connection paging, RRC connection establishment, RRC connection modification, and RRC connection release),Client Ref. No. P62335WO1 establishment, configuration, maintenance and release of point to point Radio Bearers, security functions including key management, inter radio access technology (RAT) mobility, and measurement configuration for UE measurement reporting. Said MIBs and SIBs may comprise one or more information elements (IEs), which may each comprise individual data fields or data structures.

[0131] The UE 601 and the RAN node 102A may utilize a Uu interface (e.g., an LTE-Uu interface) to exchange control plane data via a protocol stack comprising the PHY layer 801, the MAC layer 802, the RLC layer 803, the PDCP layer 804, and the RRC layer 805.

[0132] The non-access stratum (NAS) protocols 806 form the highest stratum of the control plane between the UE 601 and the MME 621. The NAS protocols 806 support the mobility of the UE 601 and the session management procedures to establish and maintain IP connectivity between the UE 601 and the P-GW 623.

[0133] The S1 Application Protocol (S1-AP) layer 815 may support the functions of the S1 interface and comprise Elementary Procedures (EPs). An EP is a unit of interaction between the RAN node 102A and the network 100. The S1-AP layer services may comprise two groups: UE-associated services and non UE- associated services. These services perform functions including, but not limited to: E-UTRAN Radio Access Bearer (E-RAB) management, UE capability indication, mobility, NAS signaling transport, RAN Information Management (RIM), and configuration transfer.

[0134] The Stream Control Transmission Protocol (SCTP) layer (alternatively referred to as the SCTP / IP layer) 814 may ensure reliable delivery of signaling messages between the RAN node 102A and the MME 621 based, in part, on the IP protocol, supported by the IP layer 813. The L2 layer 812 and the L1 layer 811 may refer to communication links (e.g., wired or wireless) used by the RAN node and the MME to exchange information.

[0135] The RAN node 102A and the MME 621 may utilize an S1-MME interface to exchange control plane data via a protocol stack comprising the L1 layer 811, the L2 layer 812, the IP layer 813, the SCTP layer 814, and the S1-AP layer 815.Client Ref. No. P62335WO1 FIG.9: User Plane Protocol Stack

[0136] FIG.9 is an illustration of an example of a user plane protocol stack in accordance with some embodiments. In this embodiment, a user plane 900 is shown as a communications protocol stack between the UE 106A (or alternatively, the UE 106B or 106N), the RAN node 102A (or alternatively, the RAN node 102B), the S-GW 622, and the P-GW 623. The user plane 900 may utilize at least some of the same protocol layers as the control plane 800. For example, the UE 601 and the RAN node 102A may utilize a Uu interface (e.g., an LTE-Uu interface) to exchange user plane data via a protocol stack comprising the PHY layer 801, the MAC layer 802, the RLC layer 803, the PDCP layer 804.

[0137] The General Packet Radio Service (GPRS) Tunneling Protocol for the user plane (GTP-U) layer 904 may be used for carrying user data within the GPRS core network and between the radio access network and the core network. The user data transported can be packets in any of IPv4, IPv6, or PPP formats, for example. The UDP and IP security (UDP / IP) layer 903 may provide checksums for data integrity, port numbers for addressing different functions at the source and destination, and encryption and authentication on the selected data flows. The RAN node 102A and the S-GW 622 may utilize an S1-U interface to exchange user plane data via a protocol stack comprising the L1 layer 811, the L2 layer 812, the UDP / IP layer 903, and the GTP-U layer 904. The S-GW 622 and the P-GW 623 may utilize an S5 / S8a interface to exchange user plane data via a protocol stack comprising the L1 layer 811, the L2 layer 812, the UDP / IP layer 903, and the GTP- U layer 904. As discussed above with respect to FIG.8, NAS protocols support the mobility of the UE 106 and the session management procedures to establish and maintain IP 813 connectivity between the UE 106 and the P-GW 623.

[0138] For the remainder of this disclosure, references to base station (gNB) and user equipment (UE) are assumed to refer to base station (gNB) 102 and user equipment (UE) 106, even though specific reference numerals may be omitted.Client Ref. No. P62335WO1 FIG.10: Core Network

[0139] FIG.10 illustrates an example architecture of a system 1000 including a core network (CN) 1020 in accordance with various embodiments. The CN 1020 may be a core network for a 5G System (which may be referred to as a 5GC). The system 1000 is shown to include a UE 1001, which may be the same or similar to the UEs 106A, 106B, or 106N discussed previously; a (R)AN 1010, which may be the same or similar to the BSs 102A or 102N discussed previously; and a data network (DN) 1003, which may be, for example, operator services, Internet access, or 3rd party services; and a CN 1020. The CN 1020 may include a number of network functions including an Authentication Server Function (AUSF) 1022; an Access and Mobility Management Function (AMF) 1021; a Session Management Function (SMF) 1024; a Network Exposure Function (NEF) 1023; a Policy Control Function (PCF) 1026; a Network Repository Function (NRF) 1025; a Unified Data Management (UDM) 1027; an Application Function (AF) 1028; a User Plane Function (UPF) 1002; and a Network Slice Selection Function (NSSF) 1029. These network functions may be implemented, in some cases, as virtualized software based functions / services.

[0140] The UPF 1002 may act as an anchor point for intra-RAT and inter-RAT mobility, an external packet data unit (PDU) session point of interconnect to DN 1003, and a branching point to support mufti-homed PDU session. A PDU session is a logical connection between the UE and the DN. The UPF 1002 may also perform packet routing and forwarding, perform packet inspection, enforce the user plane part of policy rules, lawfully intercept packets (user plane (UP) collection), perform traffic usage reporting, perform quality of service (QoS) handling for a user plane (e.g., packet filtering, gating, UL / DL rate enforcement), perform Uplink Traffic verification (e.g., Service Data Flows (SDF) to QoS flow mapping), transport level packet marking in the uplink and downlink, and perform downlink packet buffering and downlink data notification triggering. UPF 1002 may include an uplink classifier to support routing traffic flows to a data network, The DN 1003 may represent various network operator services, Internet access, or third party services. DN 1003 may include, or be similar to, application server 104 discussed previously.Client Ref. No. P62335WO1 The UPF 1002 may interact with the SMF 1024 via an N4 reference point between the SMF 1024 and the UPF 1002.

[0141] A location management function 1030 can be used to receive location management information from the UE 106 or base station 102 and use the information to determine a location of the UE. Alternatively, the UE may provide the location of the UE to the LMF 1030 using information such as GPS information and triangulation information based on signals received from 3 or more base stations 102N.

[0142] The AUSF 1022 may store data for authentication of UE 1001 and handle authentication-related functionality, The AUSF 1022 may facilitate a common authentication frame work for various access types. The AUSF 1022 may communicate with the AMF 1021 via an N12 reference point between the AMF 1021 and the AUSF 1022; and may communicate with the UDM 1027 via an N13 reference point between the UDM 1027 and the AUSF 1022. Additionally, the AUSF 1022 may exhibit an Nausf service-based interface.

[0143] The AMF 1021 may be responsible for registration management (e.g., for registering UE 1001, etc.), connection management, reachability management, mobility management, and lawful interception of AMF-related events, and access authentication and authorization. The AMF 1021 may be a termination point for the an N11 reference point between the AMF 1021 and the SMF 1024. The AMF 1021 may provide transport for SM messages between the UE 1001 and the SMF 1024, and act as a transparent proxy for routing SM messages. AMF 1021 may also provide transport for Short Message Service (SMS) messages between UE 1001 and an SMSF (not shown by FIG. 10). AMF 1021 may act as a security anchor function (SEAF), which may include interaction with the AUSF 1022 and the UE 1001, receipt of an intermediate key that was established as a result of the UE 1001 authentication process. Where Universal Subscriber Identity Module (USIM) based authentication is used, the AMF 1021 may retrieve the security material from the AUSF 1022. AMF 1021 may also include a Security Context Management (SCM) function, which receives a key from the SEAF that it uses to derive access- network specific keys. Furthermore, AMF 1021 may be a termination point of aClient Ref. No. P62335WO1 RAN control plane (CP) interface, which may include or be an N2 reference point between the (R)AN 1010 and the AMF 1021; and the AMF 1021 may be a termination point of NAS (Nl) signaling, and perform NAS ciphering and integrity protection.

[0144] AMF 1021 may also support NAS signaling with a UE 1001 over a non- 3GPP Inter-Working Function (N3IWF) interface. The N3IWF may be used to provide access to untrusted entities. N3IWF may be a termination point for the N2 interface between the (R)AN 1010 and the AMF 1021 for the control plane, and may be a termination point for the N3 reference point between the (R)AN 1010 and the UPF 1002 for the user plane. As such, the AMF 1021 may handle N2 signaling from the SMF 1024 and the AMF 1021 for PDU sessions and encapsulate / de- encapsulate packets for IPSec and N3 tunneling, mark N3 user-plane packets in the uplink, and enforce QoS corresponding to N3 packet marking while considering QoS requirements associated with such marking received over N2. N3IWF may also relay uplink and downlink control plane non-access stratum (NAS) signaling between the UE 1001 and AMF 1021 via an N1 reference point between the UE 1001 and the AMF 1021, and relay uplink and downlink user-plane packets between the UE 1001 and UPF 1002. The N3IWF also provides mechanisms for internet protocol security (IPsec) tunnel establishment with the UE 1001. The AMF 1021 may exhibit an Namf service based interface, and may be a termination point for an N14 reference point between two AMFs 1021 and an N17 reference point between the AMF 1021 and a 5G Equipment Identity Register (5G-EIR) (not shown by FIG.10).

[0145] The UE 1001 may need to register with the AMF 1021 in order to receive network services. Registration Management (RM) is used to register or deregister the UE 1001 with the network (e.g., AMF 1021), and establish a UE context in the network (e.g., AMF 1021). The UE 1001 may operate in an RM-REGISTERED state or an RM-DEREGISTERED state. In the RM-DEREGISTERED state, the UE 1001 is not registered with the network, and the UE context in AMF 1021 holds no valid location or routing information for the UE 1001 so the UE 1001 is not reachable by the AMF 1021. In the RM REGISTERED state, the UE 1001 isClient Ref. No. P62335WO1 registered with the network, and the UE context in AMF 1021 may hold a valid location or routing information for the UE 1001 so the UE 1001 is reachable by the AMF 1021. In the RM-REGISTERED state, the UE 1001 may perform mobility registration update procedures, perform periodic registration update procedures triggered by expiration of the periodic update timer (e.g., to notify the network that the UE 1001 is still active), and perform a Registration Update procedure to update UE capability information or to re-negotiate protocol parameters with the network, among others.

[0146] The AMF 1021 may store one or more RM contexts for the UE 1001, where each RM context is associated with a specific access to the network. The RM context may be a data structure, database object, etc. that indicates or stores, inter glia, a registration state per access type and the periodic update timer. The AMF 1021 may also store a 5GC mobility management (MM) context that may be the same or similar to the evolved packet services (EPS) Mobility Management (E)MM context discussed previously. In various embodiments, the AMF 1021 may store a CE mode B Restriction parameter of the UE 1001 in an associated MM context or registration management (RM) context. The AMF 1021 may also derive the value, when needed, from the UE's usage setting parameter already stored in the UE context (and / or MM / RM context).

[0147] Connection Management (CM) may be used to establish and release a signaling connection between the UE 1001 and the AMF 1021 over the N1 interface. The signaling connection is used to enable NAS signaling exchange between the UE 1001 and the CN 1020, and comprises both the signaling connection between the UE and the AN (e.g., RRC connection or UE-N3IWF connection for non-3GPP access) and the N2 connection for the UE 1001 between the AN (e.g., AN 1010) and the AMF 1021. The UE 1001 may operate in one of two CM states, CM-IDLE mode or CM-CONNECTED mode. When the UE 1001 is operating in the CM-IDLE state / mode, the UE 1001 may have no NAS signaling connection established with the AMF 1021 over the N1 interface, and there may be (R)AN 1010 signaling connection (e.g., N2 and / or N3 connections) for the UE 1001. When the UE 1001 is operating in the CM-CONNECTED state / mode, theClient Ref. No. P62335WO1 UE 1001 may have an established NAS signaling connection with the AMF 1021 over the Nl interface, and there may be a (R)AN 1010 signaling connection (e.g., N2 and / or N3 connections) for the UE 1001. Establishment of an N2 connection between the (R)AN 1010 and the AMF 1021 may cause the UE 1001 to transition from CM-IDLE mode to CM-CONNECTED mode, and the UE 1001 may transition from the CM-CONNECTED mode to the CM-IDLE mode when N2 signaling between the (R)AN 1010 and the AMF 1021 is released.

[0148] The SMF 1024 may be responsible for session management (SM) session establishment, modify and release, including tunnel maintain between UPF and AN node); UE IP address allocation and management (including optional authorization); selection and control of UP function; configuring traffic steering at UPF to route traffic to proper destination; termination of interfaces toward policy control functions; controlling part of policy enforcement and QoS; lawful intercept (for SM events and interface to LI system); termination of SM parts of NAS messages; downlink data notification; initiating AN specific SM information, sent via AMF over N2 to AN; and determining SSC mode of a session. SM may refer to management of a PDU session, and a PDU session or "session" may refer to a PDU connectivity service that provides or enables the exchange of PDUs between a UE 1001 and a data network (DN) 1003 identified by a Data Network Name (DNN). PDU sessions may be established upon UE 1001 request, modified upon UE 1001 and CN 1020 request, and released upon UE 1001 and CN 1020 request using NAS SM signaling exchanged over the N1 reference point between the UE 1001 and the SMF 1024. Upon request from an application server, the CN 1020 may trigger a specific application in the UE 1001. In response to receipt of the trigger message, the UE 1001 may pass the trigger message (or relevant parts / information of the trigger message) to one or more identified applications in the UE 1001. The identified application(s) in the UE 1001 may establish a PDU session to a specific data network name (DNN). The SMF 1024 may check whether the UE 1001 requests are compliant with user subscription information associated with the UE 1001. In this regard, the SMF 1024 may retrieve and / or request to receive update notifications on SMF 1024 level subscription data from the UDMClient Ref. No. P62335WO1 1027.

[0149] The SMF 1024 may include the following roaming functionality: handling local enforcement to apply QoS SLAB virtual Public Land Mobile Network (VPLMN); charging data collection and charging interface (VPLMN); lawful intercept (in VPLMN for SM events and interface to LI system); and support for interaction with external DN for transport of signaling for PDU session authorization / authentication by external DN. An N16 reference point between two SMFs 1024 may be included in the system 1000, which may be between another SMF 1024 in a visited network and the SMF 1024 in the home network in roaming scenarios. Additionally, the SMF 1024 may exhibit the Nsmf service-based interface.

[0150] The NEF 1023 may provide means for securely exposing the services and capabilities provided by 3GPP network functions for third party, internal exposure / re-exposure, Application Functions (e.g., AF 1028), edge computing or fog computing systems, etc. In such embodiments, the NEF 1023 may authenticate, authorize, and / or throttle the AFS. NEF 1023 may also translate information exchanged with the AF 1028 and information exchanged with internal network functions. For example, the NEF 1023 may translate between an AF- Service-Identifier and an internal SCC information. NEF 1023 may also receive information from other network functions (NFs) based on exposed capabilities of other network functions. This information may be stored at the NEF 1023 as structured data, or at a data storage NF using standardized interfaces. The stored information can then be re-exposed by the NEF 1023 to other NFs and AFs, and / or used for other purposes such as analytics. Additionally, the NEF 1023 may exhibit an Nnef service-based interface.

[0151] The NRF 1025 may support service discovery functions, receive NF discovery requests from NF instances, and provide the information of the discovered NF instances to the NF instances. NRF 1025 also maintains information of available NF instances and their supported services. As used herein, the terms "instantiate," "instantiation," and the like may refer to the creation of an instance, and an "instance" may refer to a concrete occurrence of an object, which mayClient Ref. No. P62335WO1 occur, for example, during execution of program code. Additionally, the NRF 1025 may exhibit the Nnrf service based interface.

[0152] The PCF 1026 may provide policy rules to control plane function(s) to enforce them, and may also support unified policy framework to govern network behavior, The PCF 1026 may also implement a front end (FE) to access subscription information relevant for policy decisions in a UDR of the UDM 1027. The PCF 1026 may communicate with the AMF 1021 via an N15 reference point between the PCF 1026 and the AMF 1021, which may include a PCF 1026 in a visited network and the AMF 1021 in case of roaming scenarios. The PCF 1026 may communicate with the AF 1028 via an NS reference point between the PCF 1026 and the AF 1028; and with the SMF 1024 via an N7 reference point between the PCF 1026 and the SMF 1024, The system 1000 and / or CN 1020 may also include an N24 reference point between the PCF 1026 (in the home network) and a PCF 1026 in a visited network, Additionally, the PCF 1026 may exhibit an Npcf service-based interface.

[0153] The UDM 1027 may handle subscription-related information to support the network entities' handling of communication sessions, and may store subscription data of UE 1001. For example, subscription data may be communicated between the UDM 1027 and the AMF 1021 via an NS reference point between the UDM 1027 and the AMF. The UDM 1027 may include two parts, an application FE and a UDR (the FE and UDR are not shown by FIG. 10). The UDR may store subscription data and policy data for the UDM 1027 and the PCF 1026, and / or structured data for exposure and application data (including PFDs for application detection, application request information for multiple UEs 1001) for the NEF 1023. The Nadr service-based interface may be exhibited by the UDR 221 to allow the UDM 1027, PCF 1026, and NEF 1023 to access a particular set of the stored data, as well as to read, update (e.g., add, modify), delete, and subscribe to notification of relevant data changes in the UDR. The UDM may include a UDM- FE, which is in charge of processing credentials, location management, subscription management and so on. Several different front ends may serve the same user in different transactions. The UDM-FE accesses subscriptionClient Ref. No. P62335WO1 information stored in the UDR and performs authentication credential processing, user identification handling, access authorization, registration / mobility management, and subscription management. The UDR may interact with the SMF 1024 via an Nl0 reference point between the UDM 1027 and the SMF 1024. UDM 1027 may also support SMS management, wherein an SMS-FE implements the similar application logic as discussed previously. Additionally, the UDM 1027 may exhibit the Nudm service based interface.

[0154] The AF 1028 may provide application influence on traffic routing, provide access to the NCE, and interact with the policy framework for policy control. The NCE may be a mechanism that allows the CN 1020 and AF 1028 to provide information to each other via NEF 1023, which may be used for edge computing implementations. In such implementations, the network operator and third party services may be hosted close to the UE 1001 access point of attachment to achieve an efficient service delivery through the reduced end-to-end latency and load on the transport network. For edge computing implementations, the 5GC may select a UPF 1002 close to the UE 1001 and execute traffic steering from the UPF 502 to DN 1003 via the N6 interface. This may be based on the UE subscription data, UE location, and information provided by the AF 1028. In this way, the AF 1028 may influence UPF (re)selection and traffic routing. Based on operator deployment, when AF 1028 is considered to be a trusted entity, the network operator may permit AF 1028 to interact directly with relevant NFs. Additionally, the AF 1028 may exhibit an Naf service-based interface.

[0155] The NSSF 1029 may select a set of network slice instances serving the UE 501. The NSSF 1029 may also determine allowed Network Slice Selection Assistance Information (NSSAI) and the mapping to the subscribed single NSSAI (S-NSSAI) is, if needed. The NSSF 1029 may also determine the AMF set to be used to serve the UE 1001, or a list of candidate AMF(s) 1021 based on a suitable configuration and possibly by querying the NRF 1025. The selection of a set of network slice instances for the UE 1001 may be triggered by the AMF 1021 with which the UE 1001 is registered by interacting with the NSSF 1029, which may lead to a change of AMF 1021. The NSSF 1029 may interact with the AMF 1021Client Ref. No. P62335WO1 via an N22 reference point between AMF 1021 and NSSF 1029; and may communicate with another NSSF 1029 in a visited network via an N31 reference point (not shown by FIG. 10). Additionally, the NSSF 1029 may exhibit an Nnssf service-based interface.

[0156] As discussed previously, the CN 1020 may include a short message service function (SMSF), which may be responsible for SMS subscription checking and verification, and relaying SM messages to / from the UE 1001 to / from other entities, such as an SMS-GMSC / IWMSC / SMS-router. The SMS may also interact with AMF 1021 and UDM 1027 for a notification procedure that the UE 1001 is available for SMS transfer (e.g., set a UE not reachable flag, and notifying UDM 1027 when UE 1001 is available for SMS).

[0157] The CN 1020 may also include other elements that are not shown by FIG.10, such as a Data Storage system / architecture, a 5G-EIR, a Security Edge Protection Proxy (SEPP), and the like. The Data Storage system may include a Structured Data Storage Network Function (SDSF), air Unstructured Data Storage Function (UDSF), and / or the like. Any network function (NF) may store and retrieve unstructured data into / from the UDSF (e.g., UE contexts), via N18 reference point between any NF and the UDSF (not shown by FIG.10), Individual NFs may share a UDSF for storing their respective unstructured data or individual NFs may each have their own UDSF located at or near the individual NFs. Addition- ally, the UDSF may exhibit an Nudsf service-based interface (not shown by FIG.10). The 5G-EIR may be an NF that checks the status of permanent equipment identifier (PEI) for determining whether particular equipment / entities are blacklisted from the network; and the SEPP may be a non-transparent proxy that performs topology hiding, message filtering, and policing on inter-PLMN control plane interfaces.

[0158] Additionally, there may be many more reference points and / or service- based interfaces between the NF services in the NFs; however, these interfaces and reference points have been omitted from FIG.10 for clarity. In one example, the CN 1020 may include an Nx interface, which is an inter-CN interface between a mobility management entity (MME) and the AMF 1021 in order to enable interworking between CN 1020 and a CN in a 4G system. Other exampleClient Ref. No. P62335WO1 interfaces / reference points may include an N5G-EIR service-based interface exhibited by a 5G-EIR, an N27 reference point between the NRF in the visited network and the NRF in the home network; and an N31 reference point between the NSSF in the visited network and the NSSF in the home network. Distributed Artificial (AI) and Machine Learning (ML) in Cellular Networks Model Development

[0159] Native AI wireless networks may enhance wireless networks by integrating AI into elements of the wireless networks, including integrating AI into devices connected to the network. Next generation wireless technologies (i.e., B5G / 6G) are expected to support inherently distributed AI-enabled applications. That is, so-called AI-as-a-service (e.g., gesture tracking, distributed control, voice recognition, object detection, etc.) may be supported. Next generation wireless technologies are also expected to leverage AI to optimize network features and operations (e.g., Channel State Information (CSI) compression, scheduling, mobility management, etc.).

[0160] AI tasks may be computationally intensive and require aggregation of large amounts of user data, both contextual data and cellular data, for training efficient and accurate AI models. However, devices have computation, bandwidth, storage, power and privacy constraints. In order to balance these requirements and constraints, wireless networks may leverage advanced wireless communication and mobile computing technologies by flexibly distributing functionality across devices. In both AI-as-a-service and AI for communications applications, on-device intelligence plays a key role in the AI learning process by enabling data processing closest to the source. That is, devices may complement the cloud / network (NW) and edge processing capabilities which offers lower latency, enhanced privacy, and efficient use of network bandwidth.

[0161] Contextual data may include location and mobility trajectory of the UEs, voice and data activity, user preferences, data traffic patterns, network usage patterns, and the like. Such information is used to optimize both network and deviceClient Ref. No. P62335WO1 operations. Examples of context-aware network optimization include context- aware handover optimization, context-aware radio source management for device- to-device (D2D) and cellular links, context-aware multi-radio access technology (RAT) traffic steering, and context-aware channel access and spectrum sharing. AI / ML Model Development

[0162] The development of the AI / ML models can comprise four main phases: a training, emulation (validation), deployment, and inference phase. The main task involved in each phase are briefly described in the proceeding paragraphs.

[0163] Training Phase: In this phase, the AI model is trained on a dataset. This involves feeding the model with input data and corresponding correct output labels, allowing the model to learn patterns and relationships within the data. Training typically involves optimization algorithms to adjust the model's parameters to minimize errors. It should be noted that data may be processed prior to being input into the model. This may be referred to as a data processing phase.

[0164] Emulation Phase: In the emulation phase, the trained model is tested extensively to ensure it performs well on data it has not seen before. This phase involves evaluating the model's performance metrics such as accuracy, precision, recall, etc., using validation datasets. Emulation helps identify any issues with the model's generalization and performance before deployment.

[0165] Deployment Phase: Once the model has been trained and successfully emulated, it's ready for deployment. Deployment involves integrating the model into a production environment where it can make predictions or classifications on new, unseen data. This may involve creating application programming interfaces (APIs) or integrating the model into applications or systems where it will be used.

[0166] Inference Phase: In this phase, the deployed model is used to make predictions or classifications on real-world data referred to herein as a “scenario.” A scenario typically refers to a specific situation or problem domain in which an AI / ML model is applied or evaluated. Scenarios help frame the context in which AI / ML model is are deployed. The AI / ML model takes input data, processes it, andClient Ref. No. P62335WO1 produces an output referred to as an inference, e.g., a predicted RSRP for a beam (assisted), or a predicted beam ID. AI / ML Model Monitoring

[0167] Monitoring and Evaluation: A monitoring entity, such as a UE 106, base station 102, or network 1020, can continuously monitor various factors such as data characteristics, system performance metrics, or environmental conditions.

[0168] Decision Making: Based on the monitored factors, the monitoring entity can decide whether to switch to a different machine learning model that is better suited for the current conditions or task, finetune the current model using transfer learning to better match the environmental conditions or indicate the need to fall back to non-AI based positioning.

[0169] AI model switching refers to the process of dynamically selecting or switching between different machine learning models or algorithms based on certain conditions or criteria. This approach is often used in adaptive systems where the optimal model for a particular task may change over time or in different contexts. AI model switching may include the following.

[0170] Model Selection: The monitoring entity selects the most appropriate model from a set of pre-defined models or algorithms. This selection can be based on factors such as accuracy, efficiency, or robustness.

[0171] Model finetuning: Once a model is selected, the model can be further trained on a dataset that is specific to a task. This is known as finetuning. Finetuning a pre-trained model can reduce the amount of initial training for the model, while ensuring the model is trained for the specific task for which it will be used. This enables models to be trained more generally for multiple specific tasks. Finetuning a pre-trained model may be optional, depending on how different the initial training is from the end use of the model.

[0172] Adaptation: Once a new model is selected and optionally finetuned, the monitoring entity adapts its operation to use the newly chosen model for makingClient Ref. No. P62335WO1 predictions or decisions. AI / ML Model Life Cycle Management

[0173] The life cycle management (LCM) of the AI / ML models may comprise various aspects of development and monitoring described above. Signaling and protocols, including identification related signaling, may be utilized with LCM to enable functionality and model, if justified, selection, activation, deactivation, switching, fallback, and the like. That is, signaling, protocols, and / or mechanisms are necessary for LCM in order to facilitate model training, inference, performance monitoring, and data collection for both UE-sided and NW-sided models, including signaling mechanisms of applicable functionalities and models. It should be noted that data collection may be used for UE-sided model training and NW-sided model training. In some cases, data collection for LCM may include data collection for NW-sided model training which may be distinct from CN, operation and management (OAM), and over-the-top (OTT) collection of UE-sided model training data.

[0174] A set of general data collection principles may be determined for network-side model training, including, for example, whether a UE is to support data logging, whether a UE is to report the collected data periodically, using event- based reporting, and / or on-demand. Further, UE memory, processing power, energy consumption, and signaling overhead may be considered. For CSI (Channel State Information) and beam management use cases, the training of network-side models can consider both gNB and OAM-centric data collection mechanisms. The gNB-centric data collection implies that the gNB can configure the UE to initiate / terminate the data collection procedure. It should be noted that the potential impact of layer 3 (L3) signaling, such as radio resource control (RRC) signaling, for the reporting of collected data should be assessed when determining data collection principles.

[0175] Thus, for the NW-side data collection related to beam management use cases, BS-centric (e.g. gNB-centric) and OAM-centric approaches may beClient Ref. No. P62335WO1 considered. It may be beneficial if the same measurement framework is applied to both gNB-centric data collection and OAM-centric data collection for NW-side data collection.

[0176] Accessing and processing data may be one of the most challenging tasks in AI / ML. Data is either generated by infrastructure providers, or device / users which might be reluctant to share the data, as it may contain sensitive information. Synthetic data may be an accurate and scalable replacement for real-world records. Synthetic data provides the ability to generate large training datasets with no manual labeling of data. It is possible to create synthetic data to model different scenarios and match circumstances that real data does not permit. Synthetic data is an affordable option compared to real data. For instance, for evaluation and validation in complex NW deployments, data can be generated instead of the costly deployment of testbeds. Since synthetic data is not captured from real-world events, it is possible to generate, as well as construct a dataset much faster with suitable tools and hardware. Synthetic data can remove the restrictions associated with the use of regulated or sensitive data as it may not contain any traces about real data.

[0177] FIG.11 illustrates an example of using synthetic data for training an AI model according to some embodiments. In FIG. 11, generator 1102 generates synthetic data having particular characteristics. For example, generator 1102 may optimize random noise and generate synthetic data such that the synthetic data produces statistical distributions (e.g., Batch Normalization (BN) statistics) that closely match distributions of a pre-trained model. Discriminator 1104 receives the synthetic data and real data and determines whether the synthetic data produces statistical distributions that closely match distributions of a pre-trained model. If the synthetic data produces statistical distributions that closely match distributions of a pre-trained model, discriminator 1104 can provide updated weights to generator 1102 and discriminator 1104 based on synthetic data. FIGS.12, 13 and 14: Distributed and Decentralized AI / MLClient Ref. No. P62335WO1

[0178] Distributed and decentralized learning provides an efficient approach to exploit distributed data and computing resources to collaboratively train machine learning models. Different approaches can be considered depending on the use case and the related constraints in terms of device deployment, tasks, and resource limitations.

[0179] Distributed learning can include distributed collaboration which relies on the support of a centralized entity such as a server or a base station (e.g. gNB) for coordinating the UEs. The centralized entity may be a trusted server and be referred to as an aggregation unit (AU). In some cases, the centralized entity server organizes the training, but does not access the raw data. Examples of distributed collaboration include federated learning (FL) and centralized critic distributed actor approaches. FL is a distributed machine learning approach where multiple UEs collaboratively train a model, while keeping the raw data decentralized without being moved to a single server. FL only allows the intermediate AI models to be transferred among the distributed computing resources while avoiding the transfer of training data.

[0180] FIG.12 illustrates an example of distributed collaboration according to the techniques herein. In accordance with an embodiment, procedure 1200 is a distributed collaboration procedure for training artificial intelligence or machine learning models, referred to hereinafter as machine learning models. In the example of FIG. 12, aggregation unit 1202 transmits a model training request to each of UEs 106A-106C. Each of the UEs 106A-106C can perform on-device training using the data from the aggregation unit 1202, which may include sensitive user data. Each of the UEs 106A-106C can send model updates to the aggregation unit 1202. The aggregation unit 1202 can perform model aggregation and sends a model update to each of the UEs 106A-106C. In this manner, UEs 106A-106C can collaboratively train a machine learning model, while keeping their sensitive data decentralized. As described in further detail below, each of the UEs 106A-106C and aggregation unit 1202 may implement mechanisms to preserve privacy.

[0181] Decentralized learning can include decentralized collaboration where UEs can coordinate and exchange data with machine learning models withoutClient Ref. No. P62335WO1 relying on a centralized entity. Decentralized AI training techniques include: fully distributed FL, Multi-agent reinforcement learning (MARL), and Split learning, including cross-device split learning. MARL addresses the sequential decision- making problem of multiple UEs that operate in a common environment. In MARL, each UE aims to optimize its own long-term Key Performance Indicators (KPI) by interacting with the environment and other UEs. In cases where a large-sized deep neural network (NN) cannot fit into edge devices’ small memory, split learning resolves this problem by dividing a single NN into multiple segments and distributing the lower segments across multiple devices storing raw data. It should be noted that different variants of both FL and MARL have emerged and FL and MARL can be decentralized, distributed, cooperative, competitive, etc.

[0182] FIG.13 illustrates an example of decentralized collaboration according to the techniques herein. In accordance with an embodiment, procedure 1300 is a decentralized collaboration procedure for training machine learning models. In the example of FIG. 13, a peer-to-peer topology is provided for UEs 106A-106C. It should be noted that a peer-to-peer topology may be based on UEs that are collocated in the same area forming a group. For example, a group may include a group of friends, family, colleagues, etc. Further, a peer-to-peer topology may be based on UEs belonging to a single (or household) of users connected to local area network, UEs forming a personal area network, or the like. The peer-to-peer topology may be dependent on UEs in the group using a same wireless radio access technology (RAT), such as Bluetooth, Wi-Fi direct, or 3GPP Sidelink RATs. Alternatively, the peer-to-peer topology may be independent of the type of RAT used by each UE in the group.

[0183] As illustrated in FIG.13, each of UEs 106A-106C may perform on-device training using the UE’s data, and provide updates to another UE, e.g., upon a request. In this manner, UEs 106A-106C collaboratively train a model, while keeping their sensitive data decentralized. It should be noted that although FIG.13 illustrates a particular example of a communications flow, the peer-to-peer topology in FIG.13 allows various ways in which UEs may collaboratively train a model. As described in further detail below, each of UEs 106A-106C and may implementClient Ref. No. P62335WO1 mechanisms to preserve privacy.

[0184] It should be noted that although, FIG. 12, FIG. 13, and other figures illustrate a particular number of UEs, any number of UEs may be used to collaboratively train a model (e.g., one or two, dozens, hundreds, thousands, etc.). Further, it should be noted that a combination of distributed collaboration and decentralized collaboration may be utilized. That is, aggregation unit 1202 may transmit a model training request to a UE in a group (or cluster) of other UEs. This UE may be designated as a cluster head and may coordinate training and collect data from a group of UEs. FIG.14 illustrates an example of collaboration according to the techniques herein. In accordance with an embodiment, procedure 1400 is a procedure for training machine learning models. FIG. 14 illustrates an example where aggregation unit 1202 provides a training request to UE 106A and UE 106B. UE 106A acts in a manner as described above with respected to FIG.12. A peer- to-peer topology is provided between UE 106B and UE 106C and in the example of FIG.14, UE 106B acts as a cluster head for UE 106B and UE 106C. That is, UE 106B sends a model training request to UE 106C and receives a model update from 106C. UE 106B then performs model training based on the update from 106C and sends the model update to aggregation unit 1202. Is this manner, UE 106B coordinates training and collects data for a group of UEs.

[0185] It should be noted, as described in further detail below, how a group of UEs share data, (i.e., models, data, inference outcomes) may be based on trust levels. AI / ML Privacy-Preserving Mechanisms

[0186] As described above, distributed and decentralized learning provides an efficient approach to exploit distributed data and computing resources to collaboratively train machine learning models. At the network edge, distributed and decentralized AI techniques are applied to data collected by devices. In most of these settings, the training data is sensitive. At the same time, training data leakage is ubiquitous since models are known to implicitly memorize details about trainingClient Ref. No. P62335WO1 data. As such, several techniques may be used to preserve privacy. During the data processing phase, one or more of: Naive data anonymization, K- anonymization, Differential privacy, Information-theoretic privacy, and / or Encryption may be applied. Naive anonymization refers to the removal of identifiers from data, such as, the names and addresses of the participants, to protect privacy. However, an adversary with auxiliary knowledge (e.g., from the publicly available records about individual subscribers) may be able to easily identify the user and uncover potentially sensitive information. A dataset has K-anonymity property, if each participant’s information cannot be distinguished from at least ^−1 other participants whose information is in the dataset. It should be noted that K- anonymization has been shown to perform poorly on the anonymization of high- dimensional datasets.

[0187] During the training phase, one or more of: Differential privacy, Homomorphic Encryption, and / or Secure multi-party computation may be applied. During the inference phase, one or more of: Differential privacy (DP), Homomorphic Encryption, Secure multi-party computation, and / or Information- theoretic privacy may be applied. Semantic security (encryption and secure multiparty computation), is a standard privacy requirement of encryption schemes which states that the advantage of an adversary with background information should be cryptographically small. Information-theoretic privacy is a context-aware privacy solution. Context-aware solutions explicitly model the dataset statistics, unlike context-free solutions that assume worst-case dataset statistics and adversaries. Differential privacy perturbs a data set, an AI model or its outcome by adding noise without significantly affecting the model accuracy. Differential privacy may include Input perturbation, Gradients perturbation, Output perturbation, and Objective function perturbation.

[0188] Training an AI model at the UE and sharing the models between UEs and / or a NW can reveal details about the sensitive information used by the UE. One way to prevent model leakage is DP. When properly applied, DP guarantees that the amount of sensitive information that the trained AI models can potentially leak at inference time is bounded by the privacy budget. However, there exist trade-Client Ref. No. P62335WO1 offs between the degree of privacy introduced by DP and the model’s utility. That is, the higher the degree of privacy, the lower the model’s utility. According to the techniques described herein, privacy-preserving mechanisms for enabling AI model sharing among UEs or UEs and the NW for both strict and relaxed privacy constraints are provided. Differential Privacy in AI / ML

[0189] As described above, DP is one way to perturb an AI model or its outcome without significantly affecting the model accuracy. By leveraging DP, correct statistical information of a dataset used for training can still be learned without revealing whether a specific UE contributed to the training dataset or not. In DP, it is possible to quantify the maximum amount of information that can be disclosed. This upper bound on “information leak” is referred to as the privacy budget. The privacy budget is typically set using a mathematical formula known as the "privacy loss function," which determines the amount of noise that needs to be added to the AI model or its outcome to achieve a certain level of privacy. That is, each UE mayhave a privacy budget^^which is a metric of privacy loss at a differential changein dataset used forThat is, the smaller the value of^^,the better theprivacy, which negatively impacts the accuracy of any results from analyzing thedata and the higher the value of^^, the more accurate are the results, but theprivacy is compromised.

[0190] Referring to FIG.12, in one example, DP may be implemented by each of UEs 106A-106C clipping updates to a model in order to limit a UE’s contribution. Further, aggregation unit 1202 may add noise proportional to a sensitivity when combining the updates. That is, a centralized differential private federated learning approach may be implemented, which may achieve high utility with good privacy preservation. Further, in one example, UEs 106A-106C may add noise locally and clip updates. This may occur in a case of low trust assumptions. Further, combinations of adding noise and clipping locally and adding noise duringClient Ref. No. P62335WO1 aggregation may be utilized.

[0191] As described above, there is a trade-off between privacy and utility. The noise added to data to preserve privacy can also reduce the utility of the data, making it less accurate or useful which can directly impact the achievable network KPIs. This trade-off can be difficult to manage and requires careful balancing to balance privacy and utility. For instance, AI for context-aware mobility may assume that the UEs can take advantage of their local data such as, mobility pattern, a user calendar, and UE location to guarantee seamless handover. In such a scenario, the UE either trains a model locally on its sensitive data for predicting the best next base station or shares it’s local information with the gNB for the end-to-end or coordinated training. Training the AI model at the UE and using differential privacy might result in longer training time for convergence and lower performance compared to non-privacy-preserving training.

[0192] Further, the lack of standardization and agreement on best practices is a challenge when considering DP. There is currently no standardized approach to implementing DP. This limits the ability to develop a common framework for DP for coordination between multiple UEs 106 and NW 1020. Further, different users can have different privacy needs, where some users are extremely restrictive while others are relatively loose. In DP, most of the approaches assume that all the UEs, as well as the NW, have the same privacy budget, which may impede UEs / NW from harnessing the information of datasets available at other UEs and / or the NW. Further, DP can cause over-protection over the UEs privacy which results in low accuracy of the AI models and hence poor NW utility.

[0193] In one example, UEs can be classified into at least three groups, demanding high, average, and low privacy protection for their data, respectively. These groups can be defined as follows: Privacy fundamentalists, Privacy pragmatists, and Privacy unconcerned. Privacy fundamentalists may include UEs that classify all the information generated by the device as sensitive. Hence, they are not willing to share any raw data or AI model trained on that data. Privacy pragmatists may include UEs that classify only some of the UE generated data as sensitive, but the rest can be shared with other UEs and / or the NW. PrivacyClient Ref. No. P62335WO1 unconcerned may include UEs where none of the UE generated data is classified as sensitive. Other classification levels can be defined depending on the group of individuals a given person is willing to share its information with, e.g., family members, friends, etc. Further, UEs can agree on sharing some or all their data samples with the NW or other UEs based on the privacy agreement between them. For instance, UEs belonging to the same individual might share freely all the data samples. UEs belonging to the same family / friends group might be willing to share data samples with a given privacy budget. A privacy level can also be defined at the data sample level for privacy pragmatists. Thus, a privacy budget may be associated not just per UE but per every data sample.

[0194] The techniques described herein enable both distributed and decentralized learning between multiple devices, and between devices and the NW in a privacy preserving manner, given that users might have different privacy requirements. It will be appreciated that the technical solutions provided herein may be incorporated into future specifications, including 3GPP Release 19. FIG.15, 16 and 17: Personalized Privacy Preserving Model Sharing

[0195] As described above, each UE may have a privacy budget^^andcollaboration is enabled in cellular networks based on the trustthat is established between devices. For example, three levels of trust can be considered: UEs that share models, UEs that share their data fully or partially with one another, including models, and UEs that only share the inference outcome. In some scenarios, the NW and all UEs have the same privacy budget. According to the techniques herein, the privacy of the shared AI models between the UEs or between the UEs may be preserved.

[0196] In one example, according to the techniques herein, a server-centric private model sharing approach may to utilized. In this approach, the UEs train a local machine learning model or gradients based on their sensitive dataset. That is, for example as described above with respect to FIG.12. The models are then shared with a trusted server that aggregates the AI models in a privacy preservingClient Ref. No. P62335WO1 manner before sending it to the gNB or sending it back to the UEs for updates.

[0197] FIG.15 illustrates an example of distributed collaboration for training an AI model with privacy preservation according to some embodiments. Referring to FIG.15, in accordance with an embodiment, procedure 1500 illustrates an example of signaling and data processing that may occur for distributed collaboration for training an AI model with privacy preservation according to some embodiments.

[0198] In FIG. 15, individual models are trained at N UEs 106, where N is a positive integer, which may include individual UEs, and / or UEs acting as a cluster head, and / or network (NW) 1020. The trained models are shared with the aggregation unit 1202. It should be noted that these models are trained on UEs 106 and NW 1020 sensitive data. If the sensitive data is shared, as is, with other entities, it may be possible to extract information about the private data used for training using techniques such as, deep inversion or matching techniques. These models may be referred to as teacher models or primary models and may include for example, AI and / or ML models. According to the techniques herein, privacy preserving aggregation during training can be achieved by generating a new model, which may be referred to as a student model or a secondary model, based on the private aggregate model and synthetic or publicly available data. Thus, privacy preserving training at the aggregation unit 1202 can be based on the following procedure: the aggregation unit 1202 collects unlabeled public or anonymized data from the NW 1020, UEs 106 or other sources. A synthetic dataset can be generated using AI techniques, for example, as described above. The data is labeled at the trusted aggregation unit 1202 as collected public / synthetic data. The aggregation unit 1202 trains a new model, which may be referred to as the student model, based on the collected labeled data. The student model can then be shared with untrusted entities (UEs 106 or NW 1020).

[0199] Procedure 1500 is initiated with NW 1020 sending a model update request to aggregation unit 1202. That is, NW 1020 requests to update a model that uses private user data. In some examples, it may be assumed that a public unlabeled dataset is stored at the aggregation unit 1202, and will be periodically updated by the NW 1020. In some examples, the training request can be attachedClient Ref. No. P62335WO1 with updated synthetic data that the NW is configured to use in training a student model at the aggregation unit 1202. The request can also designate the number (or minimum number) of users to be trained, and the input characteristics needed for a teacher model.

[0200] Referring to FIG. 15, aggregation unit 1202 can check whether it has enough users and data to satisfy the NW training request. The aggregation unit 1202 can then send a model update accept to the NW 1020 to start the training process, if the request can be satisfied. Along with the accept message, the aggregation unit 1202 may send the public data status (last update time) and model status. NW 1020 can send a refreshed version of the model (if needed) and the public data available at the aggregation unit 1202, based on the aggregation unit 1202 accept message details.

[0201] Aggregation unit 1202 can send a training request to the available UEs 106 to train their teacher model on-device (e.g. at each UE). This request may include the training characteristics needed. For example, the training characteristics can include a number of training iterations, a precision level, etc. Further, the training request may include the number of aggregation unit 1202 - UE 106 federated iterations in a case where federated learning is used. That is, in FIG.15, federated learning can be optional.

[0202] Once the UEs 106 finish all training iterations of the teacher model, the final teacher models are sent to the aggregation unit 1202. The aggregation unit 1202 aggregates all the teacher models and builds the final teacher model to be used for student model training (in a case federated learning is used). Otherwise, other aggregation techniques can be used such as weighting, described below, or voting mechanisms.

[0203] The student models are trained at the aggregation unit 1202 based on the public data shared by the NW 1020 and the teacher model provided by the UE 106 training. The aggregation unit 1202 sends a student model update to the NW 1020.

[0204] In some cases, UEs 106 may be attached to the NW 1020 and may haveClient Ref. No. P62335WO1 previously informed the NW 1020 of their training support capabilities (model, data available, etc.). In this case, the models to be trained can be previously agreed to with the NW 1020 (registered locally), or sent by the NW 1020 on the fly. NW 1020 and UE 106 vendors may previously agree on which features and models can be used and for what purpose. In this case, NW 1020 may be trusted and the trained models are shared with the NW 1020.

[0205] FIG.16 illustrates an example of distributed collaboration for training an AI model with privacy preservation according to some embodiments. Procedure 1600 can enable decentralized cooperation between untrusted UEs without relying on any trusted entity. Procedure 1600 is initiated with NW 1020 broadcasting a request to the UEs 106 that are registered with a training support option capability. The request can contain the training specifications, the model ID (if available), and the training data to be used to train the machine learning model. Each UE 106 can evaluate the NW 1020 request and, if the request is acceptable, send an accept for the model training (available or needs to be sent), and data status (public data already available / stored and update date). NW 1020 can send the refreshed models and public data. The refreshed model and public data may be requested by the each UEs 106. That is, UEs 106 have access to unlabeled public datasets. Further, UEs 106 may be able to generate synthetic data using AI / ML techniques, as previously discussed. In some examples, synthetic data can be generated in a privacy preserving manner by adding noise to the generative model while training, making it extremely difficult to determine the individual records in the original dataset from the newly generated data. Such data can be accessible to all UEs 106 and the trusted aggregation entity, if any. Further, in some cases, the data can be accessed via non-3GPP or 3GPP based links.

[0206] Each UE 106 can perform the student model training. In one example, UEs 106 can train a teacher AI model using the UE’s local private data and same private data can be used to generate a synthetic dataset. The teacher model can also be used for inference to label a publicly available dataset and / or synthetic dataset. A student AI model can be trained to mimic the teacher AI model. The student model can be trained either on labeled public data or on the generatedClient Ref. No. P62335WO1 synthetic data. Further, devices can exchange their models through device-to- device communication, as previously discussed. Thus, UEs 106 can train their models on their private data and then train a student model on public or synthetic data to mimic the teacher model.

[0207] Referring again to FIG.16, each UE 106 can send the trained student model to the NW 1020. The NW 1020 can aggregate all of the received student models and send the resulting aggregated model to the UEs 106. In this manner the aggregation process is privacy preserving for the UEs 106. Further, it should be noted that procedure 1600 is valid for all distributed AI techniques including, federated learning, split-learning, cooperative control (e.g., multi-agent reinforcement learning), and fully-distributed collaborative learning (i.e., with no centralized entity for aggregation).

[0208] FIG.17 illustrates a flow chart of an example method of AI training at a user equipment (UE), according to some embodiments. In method 1700, a UE trains an AI primary model on a private user data set, as shown at block 1710. It should be noted that in some cases, such models are not shared with others UEs or the NW since the gradients can reveal information about the data used for training. The UE uses the AI primary model for inference to label a public data set, as shown at block 1720. The data sets can include, for example, user activity and connectivity context (e.g., device, App, NW, and / or behavior data). The UE trains an AI secondary model using the label public data set (and possibility synthetic data), as shown at block 1730. The AI secondary model can mimic the AI primary model, such a model can be shared with any other entity since it was trained on public / synthetic data. The UE can share the secondary model directly with the NW, as shown in block 1740. That is, in some cases, the secondary model can be shared directly with the NW without any further processing. In this manner, an aggregation process that is privacy preserving for the UEs is provided.

[0209] It should be noted that in some examples, the on-device models may be trained directly on synthetic data. However, in some cases, the performance of models trained on synthetic data only, is not comparable with the performance of the models trained on real data. Typically, training on synthetic data is used forClient Ref. No. P62335WO1 weight initialization only. Moreover, defining a generative model for every sample can be computationally expensive. FIGS.18A-20B: Personalized Privacy Preserving Model Sharing

[0210] As described above, in some cases, each of UEs 106 may have the same privacy budget and in other cases, UEs 106 may have different privacy budgets which may be associated not just per UE but per every data sample. Thus, in the case of D2D scenarios, in some cases, it may be assumed that the data samples at each UE can be of different privacy budgets. Hence, for each privacy budget, a UE can define a training subset based on the sensitivity of the data samples in the training set. In one example, in order to enable UEs to share information among one another, the following may be implemented: UEs that are collocated in the same area can form a group, e.g., group of friends, family, colleagues, etc. and within a given a group, a single UE (cluster head) can be selected to collect sensitive data samples that have a given privacy budget for performing training. In this case, a model may then be trained on data samples with the same privacy budget that were collected from different UEs in the group. Further, in this case, a privacy group of budget ^^corresponds to all sensitive samples that belong to different UEs within the group i, but that have the same privacy budget. The UEs are assumed to be able to share their data with one another up to a given privacy budget depending on the group type. In one example, UEs belonging to the same person can share information for all privacy budgets, but UEs belonging to different individuals (e.g., colleagues) might be willing toshare samples up to a given budget threshold^^^.

[0211] As described above, aggregation unit 1202 may use a weighting aggregation technique. That is, according to the techniques herein, aggregation unit 1202 may use an aggregation mechanism of all the teacher models trained at the cluster heads for different privacy budget and a single group of UEs. In one example, weights may be assigned to teacher models based on their privacyClient Ref. No. P62335WO1 budget as well as the number of UEs contributing to the training in every group. Models trained on less sensitive data have higher utility and may receive higher weights compared to less accurate models trained on more sensitive data. Models generated from a large group of contributing UEs may also receive a high weight since more data makes the models more accurate.

[0212] In one example, a weight ^^is assigned to a group of UEs ^^for adataset of a privacy budget^^, and can be computed as follows:^ Relative privacy budget:^^=^^where ^ = ∑^^andℰis the set^∈ℰ of all possible privacy budgets.^ ^ Relative size of a group:^^= ^ , where ^ is the total number of clusters ^^The weight per group can be defined as follows: ^^ = ^^ . ^^

[0213] Given that there multiple privacy budgets, the may be redefined as follows: ^ ^^^= ^. where ^ = ∑^^.

[0214] Thus, according to the techniques herein a UE may be selected as a cluster head for AI model training for data samples of a particular privacy budget. For example, referring to FIG.14, UE 106B may be selected as a cluster head for UE 106B and UE 106C with respect to a privacy budget ^^and another UE (e.g., UE 106A) may be selected as a cluster head with respect to another privacy budget ^^. The UEs may also be selected as cluster heads based on additional UE such as the UE’s processing power, energy level (e.g. battery level),Client Ref. No. P62335WO1 etc.

[0215] It should be noted that although the example techniques described above, described with respect to one group of UEs, (e.g. family members, group of friends, etc.), in other examples, the techniques may be extended to multiple groups. For example, as follows: the same weight-based training process can be followed for each group and the weights may then be computed by taking into account all the groups and all the privacy budgets. In the single group case, the weights may be scaled based only on the privacy budgets.

[0216] It should be noted that a centralized aggregation unit may only be needed in distributed learning solutions that require model aggregation such as, federated learning and centralized critic, decentralized actor. In case of decentralized learning, the same privacy-preserving mechanism can be used by the UEs designated as cluster heads to share their models with one another, via D2D communication, without revealing any information about their datasets. In another example, a trusted UE may be designated as the model aggregator, if needed, e.g. federated learning among UEs. The models can be exchanged between UEs from other groups (e.g., groups of friends or family groups) given that the models are already trained in a privacy preserving manner.

[0217] FIGS.18A-18B illustrate examples where physical cluster of UEs form virtual clusters for supporting privacy levels, according to the techniques herein. In each of FIGS. 18A-18B, UEs 106A-106F support multiple privacy budgets asindicated (i.e., e.g., 106A supports ^^, ^^, ^^). In each of FIGS. 18A-18B, virtualclusters for particular privacy budget may be formed. For example, a UE may be designated as a virtual cluster head particular privacy budget. Thus, in oneexample, for privacy budget^^, UE 106E may be designated as the virtual clusterhead and a virtual cluster including UE 106E, UE 106B, and UE 106F may beformed. Further, in one example, for privacy budget^^, UE 106B may bedesignated as the virtual cluster head and a virtualincluding UE 106B, UE 106C, UE 106A, 106E, 106F, and UE 106F may be formed.Client Ref. No. P62335WO1

[0218] In the example of FIG.18A, UEs 106A-106F may form a physical cluster composed on the same family of UEs. In one example, UE 106B may be the physical cluster head (CH) that coordinates all the learning and virtual cluster creation. However, any UE (or a compute control node if any) can be used as a cluster head in P2P scenarios. Thus, in one example, UEs 106A-106F can share their data with one another, but not with UEs outside of this cluster. Virtual clusters which train models per privacy budget may be formed to enable devices to share their models with other UEs.

[0219] In the example of FIG. 18B, the models generated at the designated virtual CH UEs (e.g., UE 106B and UE 106E), can be shared directly with the NW 1020 (e.g., a BS). Further, in some examples, the same aggregation mechanisms performed in the device-centric scenario can be considered at the base station.

[0220] As described above, in the example of FIG.18A, UEs 106A-106F may form a physical cluster composed on the same family of UEs and virtual clusters for particular privacy budget may be formed. FIGS.19A-19B illustrates an example of training an AI model with privacy preservation according to some embodiments. With respect to FIGS.19A-19B, it is assumed that a subnetwork of local devices is already established and the cluster head or the compute control node determined. Further, in one example, there may be no support from the NW and the UEs can share the models with one another in a privacy-preserving manner.

[0221] Referring to FIGS.19A-19B, procedure 1900 generally includes Trusted UEs Group Formation and a UE Capabilities Exchange phase, a Training Data Collection among UEs phase, and a Training Session Phase. During the Trusted UEs Group Formation and a UE Capabilities Exchange phase, the Cluster Head UE (e.g., the physical cluster head UE 106B) can broadcast a request to form a group of trusted UEs and requests a maximum privacy budget of each of the other UEs. Each of the other UEs (e.g., UE 106A and UEs 106C-106F) can respond with its maximum privacy budget and the number of samples that can be sharedper supported privacy budget^^.During the Training Data Collection among UEsphase, the Cluster Head UEUE 106B) can requests the available number ofClient Ref. No. P62335WO1 samples per privacy budget and request capabilities. Each of the other UEs (e.g., UE 106A and UEs 106C-106F) can respond with UE capabilities and an indication of willingness to participate in the learning of the model.

[0222] Referring FIG. 19B, during the Train Session phase, for each privacybudget^^the Cluster Head UE identifies the UE that will perform the training. Forexample, physical cluster head 106B may identify that UE 106B will perform thetraining for^^and UE 106E will perform the training for^^. The Cluster Head UEinforms each UE of its selection for training for a given privacy budget ^^and requests confirmation. Each of the identified UEs provide confirmation it willhandle training for the given privacy budget ^^. The Cluster Head UE inform all UEs about the lead training UE (i.e., VirtualHead) for each privacy budget ^^and requests transmission of data samples to these UEs based on the personalized privacy budget of each sample. Each of the UEs share data samples and perform training, for example, as described above. The lead training UEs foreach privacy budget^^can send the models to Cluster Head UE.

[0223] Asabove, in the example of FIG.18B, the models generated at the designated virtual CH UEs can be shared directly with the NW 1020. FIGS. 20A-20B illustrates an example of training an AI model with privacy preservation according to some embodiments. With respect to FIGS. 20A-20B, it is assumed that a subnetwork of local devices is already established and the cluster head or the compute control node is determined. Referring to FIGS.20A-20B, procedure 2000 generally includes Trusted UEs Group Formation and a UE Capabilities Exchange phase, a Training Data Collection among UEs phase, and a Training Session and Model Sharing Phase. During the Trusted UEs Group Formation and a UE Capabilities Exchange phase, NW 1020 sends a request for group formation to the cluster head UE (e.g., the physical cluster head UE 106B). The Cluster Head UE can broadcast a request to form a group of trusted UEs and request a maximum privacy budget of each of the other UEs. Each of the other UEs can respond with its maximum privacy budget and the number of samples that can be shared perClient Ref. No. P62335WO1 supported privacy budget ^^. During the Training Data Collection among UEs phase, the Cluster Head UE can request the available number of samples per privacy budget and request capabilities. Each of the other UEs can respond with capabilities and an indication of willingness to participate in the learning.

[0224] Referring FIG.20B, during the Train Session and Model Sharing phase, for each privacy budget ^^the Cluster Head UE can identify the UE that will perform the training. The Cluster Head UE can inform each UE of its selection for training for a given privacy budget ^^and request confirmation. Each of the identified UEs can provide confirmation that these UEs will handle training for the given privacy budget ^^. The Cluster Head UE can inform all UEs about the leadtraining UE (i.e., Virtual Cluster Head) for each privacy budget^^and requesttransmission of data samples to these UEs based on the personalized privacy budget of each sample. Each of the UEs can share data samples and perform training. For example, as described above. The lead training UEs for each privacybudget^^can send the models to Cluster Head UE. The Cluster Head UE, canaggregate models, for example as described above, if necessary, and shares the aggregated models with the NW 1020. In this manner, according to the techniques herein, an AI / ML model may be trained with privacy preservation. FIG.21: Flow Chart of AI training at a user equipment (UE)

[0225] FIG.21 illustrates a flow chart of an example method of training at a user equipment (UE), according to some embodiments. The method shown in FIG.21 may be used in conjunction with any of the systems, methods, or devices illustrated in the Figures, among other devices. In various embodiments, some of the method elements shown may be performed concurrently, in a different order than shown, or may be omitted. Additional method elements may also be performed as desired.

[0226] Referring to FIG. 21, the method 2100 comprises for each privacy budget, a cluster head UE identifies the UE that will perform training, as shown atClient Ref. No. P62335WO1 block 2110. The cluster head UE informs each of the identified UEs that it will lead training, as shown at block 2120. The cluster head UE informs all of the other UEs of the UEs that will lead training, as shown at block 2130. The cluster head UE receives model updates from the UEs leading training, as shown at block 2140. The cluster head UE can send model updates, for example to a network, as shown in block 2150. In this manner, the formation of virtual cluster per privacy level and selection of a cluster head that performs the training and coordination per privacy level enables models, for example AI and / or ML models to be trained in a privacy preserving manner. FIG.22: Flow chart of AI training in a D2D architecture

[0227] FIG.22 illustrates a flow chart of an example method of training a model at a plurality of User Equipment (UEs) with different privacy budgets, via device to device communication, while preserving privacy at the plurality of UEs. The method shown in FIG.22 may be used in conjunction with any of the systems, methods, or devices illustrated in the Figures, among other devices. In various embodiments, some of the method elements shown may be performed concurrently, in a different order than shown, or may be omitted. Additional method elements may also be performed as desired.

[0228] Referring to FIG.22, the method 2200 comprises broadcasting a request from a cluster head UE, to the plurality of UEs to form a group of trusted UEs, as shown in block 2210. The cluster head UE can request a maximum privacy budget that is supported by each UE in the group of trusted UEs, as shown in block 2220. The method 2200 further comprises forming the group of trusted UEs into a plurality of privacy budget groups, wherein each UE in the group of trusted UEs is assigned to a privacy budget group in the plurality of privacy budget groups based on each UE’s maximum privacy budget, as shown in block 2230. The cluster head UE can identify a UE in each of the plurality of privacy budget groups as a training UE, as shown in block 2240. The cluster head UE can send an instruction to the training UE in each privacy budget group to request a transmission of data samples fromClient Ref. No. P62335WO1 other UEs in each corresponding privacy budget group to be sent to the training UE to train a model in each privacy budget group to form an updated model in each privacy budget group, as shown in block 2250. The cluster head UE can receive the updated model at the cluster head UE from each training UE for each privacy budget group, as shown in block 2260.

[0229] In some embodiments, the method 2200 can further comprise sending a request to each of the plurality of UEs for an available number of samples per privacy budget to be used to train the model. The cluster head UE can send a request for a capability of each UE in the plurality of UEs to identify UEs capable of participating in training of the model. The cluster head UE can receive the capability for each UE in the plurality of UEs. The cluster head UE can also receive a request from a network to form the group of trusted UEs. The cluster head UE can aggregate the updated model at the cluster head UE from each training UE. The cluster head UE can send the aggregated model to the network or to one or more UEs. For example, the cluster head UE may send the aggregated model to the UEs in the group of trusted UEs. Alternatively, the cluster head UE may send the aggregated model to UEs that are not within the group of trusted UEs, since the aggregated model is configured based on the UE’s privacy model so that information is not disclosed outside of each UE’s privacy model. The aggregated model may also be sent by the network to other UEs within or without of the trusted group. In some embodiments, a UE privacy budget can set an upper bound on information leak. In some embodiments, the UE privacy budget is set according to a privacy loss function.

[0230] In some embodiments, a baseband processor 604 can be configured to cause a user equipment UE 106 to assist with performing the operations disclosed herein, including but not limited to the operations disclosed in method 2100 and method 2200. The baseband processor 604 can be coupled to a memory.

[0231] In some embodiments, an apparatus can be configured to cause a user equipment (UE) to assist with performing the operations disclosed herein, including but not limited to the operations disclosed in method 2100 and method 2200. The UE can include one or more processors, coupled to a memory, configured toClient Ref. No. P62335WO1 perform the operations.

[0232] In some embodiments, a computer program product, comprising computer instructions which, when executed by one or more processors, perform any of the operations described herein.

[0233] Embodiments of the present disclosure may be realized in any of various forms. For example, some embodiments may be realized as a computer- implemented method, a computer readable memory medium, or a computer system. Other embodiments may be realized using one or more custom-designed hardware devices such as ASICs. Still other embodiments may be realized using one or more programmable hardware elements such as FPGAs.

[0234] In some embodiments, a non-transitory computer-readable memory medium may be configured so that it stores program instructions and / or data, where the program instructions, if executed by a computer system, cause the computer system to perform a method, e.g., any of the method embodiments described herein, or, any combination of the method embodiments described herein, or, any subset of any of the method embodiments described herein, or, any combination of such subsets.

[0235] In some embodiments, a device (e.g., a UE 106) may be configured to include a processor (or a set of processors) and a memory medium, where the memory medium stores program instructions, where the processor is configured to read and execute the program instructions from the memory medium, where the program instructions are executable to implement any of the various method embodiments described herein (or, any combination of the method embodiments described herein, or, any subset of any of the method embodiments described herein, or, any combination of such subsets). The device may be realized in any of various forms.

[0236] Any of the methods described herein for operating a user equipment (UE) may be the basis of a corresponding method for operating a base station, by interpreting each message / signal X received by the UE in the downlink as message / signal X transmitted by the base station, and each message / signal YClient Ref. No. P62335WO1 transmitted in the uplink by the UE as a message / signal Y received by the base station.

[0237] Although the embodiments above have been described in considerable detail, numerous variations and modifications will become apparent to those skilled in the art once the above disclosure is fully appreciated. It is intended that the following claims be interpreted to embrace all such variations and modifications.

Claims

Client Ref. No. P62335WO1 CLAIMS What is claimed is:

1. A method of training a model at a user equipment (UE), the method comprising: receiving a model training request from an aggregation unit; training a primary model on a private user data set to form a primary model update; and sending the primary model update to the aggregation unit to enable the aggregation unit to aggregate the primary model update with additional primary model updates from other UEs.

2. A method of training a model at an aggregation unit (AU), the method comprising: sending a model training request to a plurality of user equipment (UEs) to instruct each of the UEs to train a primary model on a private user data set of each of the UEs to form a primary model update for each of the UEs; receiving, at the AU, the primary model updates from one or more of each of the UEs in the plurality of UEs based on primary model training; generating a final primary model by aggregating the primary model updates; and training a secondary model using the final primary model and public data or synthetic data to enable the secondary model to be shared with untrusted entities.

3. The method of claim 2, wherein aggregating the primary model updates includes weighting the primary model updates based on UE privacy budgets of the plurality of UEs.

4. The method of claim 3, wherein a UE privacy budget sets an upper boundClient Ref. No. P62335WO1 on information leak.

5. The method of claim 3 or 4, wherein a UE privacy budget is set according to a privacy loss function.

6. The method of any of claims 1-5, wherein the training request indicates one or more of: a number of training iterations and training precision.

7. The method of any of claims 1-6, wherein the training request indicates a number of aggregation unit UE federated iterations.

8. A method of training a model at a user equipment (UE), the method comprising: receiving a public data set; training a primary model on a private user data set; using the primary model to label the public data set; training a secondary model on the labeled public data set to form a secondary model update; and sending the secondary model update based on the secondary model training to a network component.

9. A method of training a model at a user equipment (UE), the method comprising: receiving a public data set from a cluster head UE; training a primary model on a private user data set at the UE; using the primary model to label the public data set; training a secondary model at the UE on the labeled public data set to form a secondary model update; and sending the secondary model update based on the secondary model training to the cluster head UE.

10. The method of claim 8 or 9, further comprising training the secondaryClient Ref. No. P62335WO1 model on a synthetic dataset.

11. A method of training a model at a cluster head user equipment (UE), the method comprising: for each privacy budget, sending an indication to a UE that the UE is selected as a lead training UE for a given privacy budget; providing indications of selected lead training UEs to other UEs in a cluster; receiving model updates from lead training UEs; and sending model updates to a network.

12. The method of claim 11, further comprising: broadcasting a request to form a group of trusted UEs and requesting a maximum privacy budget for each UE; receiving a response including the maximum privacy budget for each UE; and forming trusted groups of trusted UEs.

13. The method of claim 11 or 12, wherein a UE privacy budget sets an upper bound on information leak.

14. The method of claim 13, wherein a UE privacy budget is set according to a privacy loss function.

15. A method of training a model at a user equipment (UE), the method comprising: receiving, at the UE, a request from a cluster head UE, to form a group of trusted UEs and a request for a maximum privacy budget that is supported by the UE, wherein the maximum privacy budget supported by the UE identifies a number of samples used for training a model that can be shared by the UE; sending, from the UE, a response to the cluster head UE includingClient Ref. No. P62335WO1 the maximum privacy budget and the number of samples that can be shared per supported privacy budget; receiving, from the cluster head UE, a request for the number of samples per privacy budget and a request for UE capabilities of the UE; and sending a response to the cluster head UE including an indication of a willingness to participate in training of a model and UE capabilities of the UE.

16. A method of training a model at a plurality of User Equipment (UEs) with different privacy budgets, via device to device communication, while preserving privacy at the plurality of UEs, the method comprising: broadcasting a request from a cluster head UE, to the plurality of UEs to form a group of trusted UEs; requesting a maximum privacy budget that is supported by each UE in the group of trusted UEs; forming the group of trusted UEs into a plurality of privacy budget groups, wherein each UE in the group of trusted UEs is assigned to a privacy budget group in the plurality of privacy budget groups based on each UE’s maximum privacy budget; identifying a UE in each of the plurality of privacy budget groups as a training UE; sending an instruction to the training UE in each privacy budget group to request a transmission of data samples from other UEs in each corresponding privacy budget group to be sent to the training UE to train a model in each privacy budget group to form an updated model in each privacy budget group; and receiving the updated model at the cluster head UE from each training UE for each privacy budget group.

17. The method of claim 16, further comprising:Client Ref. No. P62335WO1 sending a request to each of the plurality of UEs for an available number of samples per privacy budget to be used to train the model; sending a request for a capability of each UE in the plurality of UEs to identify UEs capable of participating in training of the model; and receiving the capability for each UE at the cluster head UE.

18. The method of claims 16 or 17, further comprising: receiving, at the cluster head UE, a request from a network to form the group of trusted UEs.

19. The method of claims 16 to 18, further comprising: aggregating the updated model at the cluster head UE from each training UE; and sending the aggregated model to the network or to one or more UEs.

20. The method of any of claims 15 to 19, wherein a UE privacy budget sets an upper bound on information leak.

21. The method of any of claims 15 to 19, wherein a UE privacy budget is set according to a privacy loss function.

22. The method of claims 1 to 21, wherein a model includes a machine learning model.

23. The method of claims 1 to 21, wherein a model includes an artificial intelligence model.

24. A baseband processor configured to cause a user equipment (UE) to assist with performing the method of any of claims 1 to 21.Client Ref. No. P62335WO1 25. An apparatus configured to cause a user equipment (UE) to assist with performing the method of any of claims 1 to 21.

26. A computer program product, comprising computer instructions which, when executed by one or more processors, perform any of the operations described herein.

Images