Generative artificial intelligence prompt management in a communication network

Abstract

Techniques are provided for prompt management in a communication network environment. For example, a method includes obtaining a prompt and translating the obtained prompt to a translated prompt. The method then generates tracking data, the tracking data being indicative of the translation performed. The method then associates the tracking data with the translated prompt. An entity that performs the method is associated with a communication network and the obtained prompt and the translated prompt are generative artificial intelligence prompts.

Description

[0001] GENERATIVE ARTIFICIAL INTELLIGENCE PROMPT MANAGEMENT IN A COMMUNICATION NETWORK

[0002] Field

[0003] The field relates generally to communication networks, and more particularly, but not exclusively, to management of generative artificial intelligence functionalities in such communication networks.

[0004]

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

[0006] Advancements in communication network technologies have rapidly progressed over recent years.

[0007] Fourth generation (4G) wireless mobile telecommunications technology, also known as Long Term Evolution (LTE) technology, provided high-capacity mobile multimedia with high data rates particularly for human interaction, as compared with previous generations of communication networks.

[0008] Fifth generation (5G) technology currently provides not only for human interaction use cases, but also for machine type communications in so-called Internet of Things (loT) networks. While 5G networks enable massive loT services (e.g., very large numbers of limited capacity devices) and mission-critical loT services (e.g., requiring high reliability), improvements over 4G communication services are supported in the form of enhanced mobile broadband (eMBB) services providing improved wireless Internet access for mobile devices.

[0009] Sixth generation (6G) technology is now being developed for communication networks that differs from 5G technology by offering, inter alia, significant improvements in speed and latency (e.g., the Ultra-Reliable Low-Latency Communication (URLLC) service that began with 5G is being refined and improved in 6G to address more stringent connectivity requirements), as well as the capability to sense a physical environment through expanded spectrum band usage. Such sensing capability enables creation of a digital twin of the physical environment which leads to new applications such as, but not limited to, highly accurate localization and immersive experiences. Furthermore, in 6G technology, artificial intelligence (Al) applications, which mayinclude machine learning (ML) applications, are intended to be more readily utilized to facilitate various communication network functionalities.

[0010] However, security management is an important consideration in any communication network environment - and now especially ones that provide for applications such as localization, immersion, Al, and the like. Moreover, security management is an ongoing consideration due to continuing attempts to improve the architectures and protocols associated with communication networks in order to increase network efficiency and / or subscriber convenience. Accordingly, security management can present significant technical challenges.

[0011] Illustrative embodiments provide techniques for prompt management in a communication network environment. More particularly, for example, illustrative embodiments provide techniques for tracking generative artificial intelligence (Al) prompts in a communication network environment.

[0012] In one illustrative embodiment, a method includes obtaining a prompt and translating the obtained prompt to a translated prompt. The method then generates tracking data, the tracking data being indicative of the translation performed. The method then associates the tracking data with the translated prompt. An entity that performs the method is associated with a communication network and the obtained prompt and the translated prompt are generative artificial intelligence prompts.

[0013] In another illustrative embodiment, a method includes generating a prompt and generating tracking data indicative of the generated prompt. The method then associates the tracking data with the generated prompt and sends the generated prompt with the associated tracking data toward an entity in a communication network. In some embodiments, the method is performed by user equipment and the generated prompt is a generative artificial intelligence prompt.

[0014] Further illustrative embodiments are provided in the form of a non-transitory computer readable medium having embodied therein executable program code that when executed by a processor causes the processor to perform the above and / or other steps, operations, and the like. Still further illustrative embodiments comprise an apparatus with a processor and a memory configured to perform the above and / or other steps, operations, and the like. Some illustrative embodiments comprise a system configured to perform the above and / or other steps, operations,and the like. Further, some illustrative embodiments comprise an apparatus or a system comprising means for performing the above and / or other steps, operations, and the like.

[0015] In some illustrative embodiments, technical solutions described herein are particularly well suited for implementation in a 6G architecture that implements Al applications.

[0016] These and other features and advantages of embodiments described herein will become more apparent from the accompanying drawings and the following detailed description.

[0017]

[0018] FIG. 1 illustrates a communication network environment with which one or more illustrative embodiments may be implemented.

[0019] FIG. 2 illustrates user equipment and entities with which one or more illustrative embodiments may be implemented.

[0020] FIGS. 3 A and 3B illustrates a prompt management procedure in a communication network according to an illustrative embodiment.

[0021] FIG. 4 illustrates a prompt management procedure in a communication network with user equipment- sourced prompts according to an illustrative embodiment.

[0022] FIG. 5 illustrates a prompt management procedure in a communication network with user equipment- sourced prompts according to another illustrative embodiment.

[0023] FIG. 6 illustrates a prompt management procedure in a communication network with user equipment- sourced prompts according to yet another illustrative embodiment.

[0024] FIG. 7 illustrates a prompt management procedure in a communication network with user equipment- sourced prompts according to a further illustrative embodiment.

[0025] Detailed

[0026]

[0027] Embodiments will be illustrated herein in conjunction with example communication systems and associated techniques for security management in communication systems. It should be understood, however, that the scope of the claims is not limited to particular types of communication systems and / or processes disclosed. Embodiments can be implemented in a wide variety of other types of communication systems, using alternative processes and operations. For example, although illustrated in the context of wireless cellular systems utilizing 3rd GenerationPartnership Project (3GPP) system elements, such as 5G and 6G system elements, the disclosed embodiments can be adapted in a straightforward manner to a variety of other types of systems.

[0028] In accordance with illustrative embodiments, one or more 3GPP technical specifications (TS) and technical reports (TR) may provide further explanation of network elements / functions and / or operations that may interact with parts of the inventive solutions, for example, but not limited to, 3GPP TS 33.501 entitled “Technical Specification Group Services and System Aspects; Security Architecture and Procedures for 5G System,” the disclosure of which is incorporated by reference herein in its entirety. Note that 3GPP TS / TR documents are non-limiting examples of communication network standards (e.g., specifications, procedures, reports, requirements, recommendations, and the like). However, while well-suited for 3GPP standards, embodiments are not necessarily intended to be limited to any particular standards.

[0029] It is to be understood that the terms 5G network, 6G network, and the like (e.g., 5G or 6G system, 5G or 6G communication system, 5G or 6G environment, 5G or 6G communication environment, etc.), in some illustrative embodiments, may be understood to comprise all or part of an access network and all or part of a core network. However, the terms 5G network or 6G network, and the like, may also occasionally be used interchangeably herein with the terms 5GC network or 6GC network (as well as 5GS or 6GS), respectively, without any loss of generality.

[0030] Prior to describing illustrative embodiments, a general description of relevant aspects of generative Al functionalities in a communication network, and technical challenges thereof, will first be described below.

[0031] Generative Al is an approach to create new content of different types following the characteristics of a training data set. One generative Al approach utilizes large language models (LLMs) to generate plausible text / language based on an input query. There are currently numerous examples of proprietary and open-source models available and used in different applications e.g., ChatGPT, Llama, etc.

[0032] In order to improve model outputs, a retrieval augmented generation (RAG) approach is often used where the user input / prompt is enriched with knowledge external to the generative Al model. RAG typically has two phases:

[0033] 1) Retrieval phase: algorithms (e.g., similarity scoring with cosine calculation of user input and external information) search for and retrieve snippets of information most relevant to the user’ s prompt or question. This external information is appended to the user’s prompt and passed to the LLM.2) Generation phase: the LLM relies on the augmented prompt and its internal representation of its training data to generate the answer, and final response is a combination of the retrieved information and the model’s own generative capabilities.

[0034] LLMs use one input and output data modality and that is (natural) language. Large Multimodal Models (LMMs) combine various data modalities, e.g., text, audio, visual, sensor data, etc., capturing the correlations between different modalities. Such an LMM approach can be applicable to any kind of data, including communication network data. Finally, Small Language Models (SLMs) are less compute-intense than LLMs, both in training and inference, with sufficient performance especially if trained and used for a specific problem. Therefore, besides LLMs and LMMs, SLMs have also high relevance in telecommunication applications.

[0035] Generative Al models, e.g., LLMs, SLMs, LMMs, take as input user prompts.

[0036] LLM / LMM / SLM can be used to empower Al agents (also occasionally referred to herein as “GenAI agents”). In some illustrative embodiments, an Al agent is a set of program code (software) that resides on a device that is part of the system in which generative Al functionalities are being executed. However, in other illustrative embodiments, an Al agent can be implemented in hardware or a combination of software and hardware. Such Al agents are configured to leverage LLM / LMM / SLM capability to, inter alia, reason, use external tools, keep the memory of actions and environment perceptions, etc., thus serving as powerful entities to fulfill a task.

[0037] In the context of a communication network, e.g., a 6G network, Al agents can be configured with the following distinguishing features as compared to other types of network entities / functions :

[0038] (i) Al agents can be configured to act autonomously. For example, leveraging the capability of an LLM or LMM, an Al agent can understand its environment and generate actions according to the task that it is executing.

[0039] (ii) Multiple Al agents can collectively understand complex tasks that are not possible to solve with a single agent. Instead, a so-called planning agent may create a plan on how to fulfill the complex task. This may include a list of Al agents that need to be involved and collaborate among each other, and a list of actions (e.g., CRUD (create, read, update, delete) operations, using tools, services, etc.).

[0040] (iii) An Al agent can collaborate with other Al agents according to a plan.

[0041] Al agents in the 6G network can collaborate to fulfill a task. The task can be related to an end consumer (e.g., assistants in daily activities) or a network (e.g., fulfilling energy minimizationintent). Starting from the initial prompt describing a complex task, a series of prompt translations may be required in order to distil the complex task into simple prompts or commands, e.g., CRUD operations executable by simple, task specific Al agents or network functions.

[0042] FIG. 1 illustrates a communication network 100 with a plurality of Al agents 102-1 through 102-7 distributed therein. Starting from a complex task originating (sourced) at user equipment (UE) or a network function (NF), e.g., minimize energy consumption for XR service, translation of the complex prompt / task into different sub-prompts and sub-tasks is needed on different hierarchical levels of the plurality of Al agents 102-1 through 102-7, until the simple prompt / task / command or service request is derived and can be executed by NFs or Al agents. As shown along the left side branch in FIG. 1, from the complex prompt, translation occurs into prompt 1, prompt 3, and ultimately to an NF service call or request 1.

[0043] However, in this setup, one technical challenge is to ensure proper auditing and tracking of the generative Al operations. In the simple FIG. 1 example mentioned above (left side branch), the service call will be performed by Al agent 102-4, but that would be effectively the result of prompt translation from Al agent 102-2. Thus, while the UE is the complex prompt originator in this example, explainability or tracking of the generative Al operations at different nodes (Al agents) is desirable - but not available in existing generative Al systems.

[0044] Another technical challenge in existing generative Al systems is that service authorization does not need to be performed for Al agent 102-4 solely, but it needs to relate to all Al agents and UEs / NFs involved in the prompt translation.

[0045] For such a comprehensive service flow, taking into account all entities / AI agents involved in the prompt translation, there is a need for a prompt translation tracking mechanism which is currently missing in existing generative Al systems.

[0046] Illustrative embodiments overcome the above and other technical challenges by providing techniques for prompt management in a communication network. As illustratively used herein, the term “prompt” can include, for example, any data set that is used to convey a request, command, instruction, task, or the like, within a communication network. In some illustrative embodiments, such prompts may be exchanged between generative Al agents (also referred to herein as “nodes”) to fulfill an operation, functionality, step, process, procedure, or the like.

[0047] Referring initially to FIG. 2, a block diagram illustrates a computing architecture for various participants (e.g., UEs, NFs, other entities, etc.) in generative Al prompt management methodologies according to illustrative embodiments. More particularly, system 200 is showncomprising user equipment (UE) 202 and a plurality of entities 204-1, . . . . , 204-N. For example, in illustrative embodiments and with reference back to FIG. 1, UE 202 can represent the UE shown as the complex prompt originator, while entities 204- 1 , . . . , 204-N can represent network functions (e.g., an access and mobility management function (AMF), a policy control function (PCF), etc.) in a core network (CN) of a communication network, as well as system elements associated with a radio access network (RAN) that couples the UE to the CN. (e.g., gNB, etc.). It is to be appreciated that the UE 202 and entities 204-1, . . . . , 204-N are configured to interact to provide generative Al prompt management and other techniques described herein. In such illustrative embodiments, each of UE 202 and entities 204-1, . . . . , 204-N can be configured with an Al agent or node.

[0048] The user equipment 202 comprises a processor 212 coupled to a memory 216 and interface circuitry 210. The processor 212 of the user equipment 202 includes a prompt management processing module 214 that may be implemented at least in part in the form of software executed by the processor. The prompt management processing module 214 performs prompt management described in conjunction with subsequent figures and otherwise herein. The memory 216 of the user equipment 202 includes a prompt management storage module 218 that stores data generated or otherwise used during prompt management operations.

[0049] Each of the entities (individually or collectively referred to herein as 204) comprises a processor 222 (222-1, . . . , 222-N) coupled to a memory 226 (226-1, . . . , 226-N) and interface circuitry 220 (220-1, . . . , 220-N). Each processor 222 of each entity 204 includes a prompt management processing module 224 (224-1, . . . , 224-N) that may be implemented at least in part in the form of software executed by the processor 222. The prompt management processing module 224 performs prompt management operations described in conjunction with subsequent figures and otherwise herein. Each memory 226 of each entity 204 includes a prompt management storage module 228 (228- 1 , . . . , 228-N) that stores data generated or otherwise used during prompt management operations.

[0050] The processors 212 and 222 may comprise, for example, microprocessors such as central processing units (CPUs), application-specific integrated circuits (ASICs), digital signal processors (DSPs) or other types of processing devices, as well as portions or combinations of such elements.

[0051] The memories 216 and 226 may be used to store one or more software programs that are executed by the respective processors 212 and 222 to implement at least a portion of the functionality described herein. For example, prompt management operations and otherfunctionality as described in conjunction with subsequent figures and otherwise herein may be implemented in a straightforward manner using software code executed by processors 212 and 222.

[0052] A given one of the memories 216 and 226 may therefore be viewed as an example of what is more generally referred to herein as a computer program product or still more generally as a computer or processor readable (non-transitory or storage) medium that has executable program code embodied therein. Other examples of computer or processor readable media may include disks or other types of magnetic or optical media, in any combination. Illustrative embodiments can include articles of manufacture comprising such computer program products or other computer or processor readable media.

[0053] Further, the memories 216 and 226 may more particularly comprise, for example, electronic random-access memory (RAM) such as static RAM (SRAM), dynamic RAM (DRAM) or other types of volatile or non-volatile electronic memory. The latter may include, for example, non-volatile memories such as flash memory, magnetic RAM (MRAM), phase-change RAM (PC-RAM) or ferroelectric RAM (FRAM). The term “memory” as used herein is intended to be broadly construed, and may additionally or alternatively encompass, for example, a read-only memory (ROM), a disk-based memory, or other type of storage device, as well as portions or combinations of such devices.

[0054] The interface circuitries 210 and 220 illustratively comprise transceivers or other communication hardware or firmware that allows the associated system elements to communicate with one another in the manner described herein.

[0055] It is apparent from FIG. 2 that user equipment 202 and plurality of entities 204 are configured for communication with each other as prompt management participants via their respective interface circuitries 210 and 220. This communication involves each participant sending data to and / or receiving data from one or more of the other participants. The term “data” as used herein is intended to be construed broadly, so as to encompass any type of information that may be sent between participants including, but not limited to, identity data, key pairs, key indicators, tokens, secrets, prompt management messages, registration request / response messages and data, request / response messages, authorization and / or authentication request / response messages and data, metadata, control data, audio, video, multimedia, consent data, other messages, etc.It is to be appreciated that the particular arrangement of components shown in FIG. 2 is an example only, and numerous alternative configurations may be used in other embodiments. For example, any given network element / function and / or access point can be configured to incorporate additional or alternative components and to support other communication protocols.

[0056] Other network functions and system elements (not expressly mentioned above including those that will be further described herein in the context of FIGS. 3A-7) may each be configured to include components such as a processor, memory and network interface. Also, entities such as third-party applications and network operators can participate in methodologies described herein via computing devices configured to include components such as a processor, memory and network interface. These elements and devices need not be implemented on separate stand-alone processing platforms, but could instead, for example, represent different functional portions of a single common processing platform.

[0057] More generally, FIG. 2 can be considered to represent processing devices configured to provide respective prompt management functionalities and operatively coupled to one another in a communication system. By way of example only, all or parts of each of UE 202 and the plurality of entities 204 (e.g., processor and memory) can be considered examples of means for performing one or more operations, one or more steps, one or more functions, one or more processes, etc. as described herein.

[0058] Given the above general description of relevant features of a generative Al system, and technical challenges therewith, solutions proposed in accordance with illustrative embodiments will now be further described below.

[0059] Illustrative embodiments provide techniques for prompt translation tracking in a multilayered generative Al system environment where multiple Al agents deployed in different nodes (e.g., UEs, NFs, etc.) need to be involved in execution and translation of a complex task / prompt. Al agents can coordinate amongst each other on the complex task being executed by exchanging the prompts. More particularly, in accordance with illustrative embodiments, a prompt management approach can include:

[0060] (1) Distributed prompt translation tracking utilizing:

[0061] (i) A prompt complexity level indicator indicating if the prompt is “complex” and can be further distilled into simple prompts and correspondingly to commands / network function calls translation, or the prompt is “simple” and corresponds to a single service call or command / operation, e.g., CRUD.(ii) Prompt tracking metadata which stores the information on all involved Al agents on the path from distilling the complex prompt into a single simple prompt. In one example, such metadata can contain a list of all Al agent prompt originators on the way (e.g., along the hierarchical path) during the transition process from the complex to the simple prompt.

[0062] (2) A prompt complexity level indicator, in some illustrative embodiments, can be implemented via a custom header in a Service-Based Architecture (SB A) message framework. By way of one example, a format of the custom header can include:

[0063] 3GPP-Prompt-translator

[0064] { Indication: Complex / Simple

[0065] }

[0066] (3) Prompt metadata, in some illustrative embodiments, can also be implemented via a custom header in an SBA message framework. By way of one example, a format of the custom header:

[0067] 3GPP-Prompt-Info

[0068] { Level : 1...N, first translation will include level 1 and second translation will include level 2.. ,N

[0069] Source GenAI: Source Gen Al Identification

[0070] Destination GenAI: Destination Gen Al Identification

[0071] Source Prompt: Source prompt received at the Source NF

[0072] Source Prompt Type: Complex / Simple

[0073] Translated prompt: Translated prompt performed by the source NF (if translation leads to further prompt

[0074] Translated prompt type: Simple / Complex

[0075] Translated simple Operation List: Operation 1, Operation2..N (if translated prompt type is simple

[0076] }

[0077] (4) Centralized prompt translation tracking implemented with a centralized prompt translation tracking entity which collects, in a centralized manner, the information on Al agent prompt originators on the way during the transition process from the complex to the simple prompt.

[0078] Moreover, service authorization can be performed utilizing prompt translation metadata. The authorization can be use case dependent. For example, in a use case of collaborative Al agents fulfilling a task - the authorization entity needs to check all prompt originators if they are allowedto access required information. Additionally or alternatively, in a network RAG use case, only an initial prompt originator needs to have the authorization to access certain information. More particularly, in the case of collaborative Al agents, the authorization entity will check complete metadata containing all prompt origins on the path and if there is even a single Al agent on the path that is not authorized to consume certain service, the authorization request will not be granted. This can be due to avoiding unpreferred Al agents or locations where they are deployed. This feedback can be given back to the Al agents such that they avoid prompt translation and distillation in the future that will include such unpreferred Al agents.

[0079] Referring now to FIGS. 3 A and 3B, a prompt management procedure 300 (procedure 300) is depicted in a communication network according to an illustrative embodiment. More particularly, FIG. 3A depicts a distributed prompt management portion of procedure 300, while FIG. 3B depicts a centralized prompt management portion of procedure 300. In some illustrative embodiments, the distributed prompt management portion and the centralized prompt management portion of procedure 300 are performed separately (excluding the other), while in other illustrative embodiments, they are performed in combination.

[0080] As shown, procedure 300 may involve a plurality of Al agents 302-1, 302-2, 302-3, and 302-4 (e.g., GenAI Agent 1 on NF1, GenAI Agent 2 on NF2, GenAI Agent 3 on NF3, and GenAI Agent 4 on NF4), a network function (NF) 304, a network repository function (NRF) 306, and a prompt translation tracking function (PTTF) 308, as will be further described below in the context of steps 1-22.

[0081] Distributed prompt translation tracking portion of procedure 300:

[0082] Step 1: The GenAI agent 1 (302-1), which can be located at different network entities, a third party application, or a UE device, provides a complex prompt (task) towards GenAI agent 2 (302-2) on another network entity, e.g., such as “minimize energy consumption for XR service.” Such a prompt cannot be fulfilled by a simple action, e.g., configuration of a single NF, but needs to be distilled into several further prompts / tasks. The GenAI agentl (hosted on the NF1 provides a complex prompt to the NF2 / GenAI Agent 2 along with the prompt metadata containing following information:

[0083] { Level : 1 - since NF1 / GenAI Agent 1 is the originator of the prompt

[0084] Source GenAI: NF1 ID (GenAI 1 ID)

[0085] Destination GenAI: NF2 ID (GenAI 2 ID)

[0086] Source Prompt: Minimize energy consumption for XR service / devicesSource Prompt Type: Complex

[0087] Translated prompt: Monitor QoS of XR users / devices and idle / busy period of XR devices

[0088] Translated prompt type: Complex

[0089] Translated simple Operation List: Null

[0090] }

[0091] Further, the client credentials assertion (CCA) feature of the NF is enhanced to also include the translated prompt as well as translated prompt type, to ensure that the prompt is signed and is not tampered with during the path. This CCA is then sent along with the prompt lifecycle, and all the intermediate nodes add their CCAs.

[0092] Step 2: The NF2 / GenAI agent 2 comprehends the received prompt and distills it into (different) sub-prompt(s) and sub-task(s). The NF2 / GenAI agent 2 appends to the metadata received from the NFl / GenAI Agent 1 the metadata after the prompt translation / distillation. For example, the following metadata is appended to the metadata received from NFl / GenAI Agent 1:

[0093] { Level : 2 - since NF2 / GenAI Agent 2 received and distilled / translated Level 1 prompt from NFl / GenAI Agent 1

[0094] Source GenAI: NF2 ID (GenAI 2 ID)

[0095] Destination GenAI: NF3 ID (GenAI 3 ID)

[0096] Source Prompt: Monitor QoS of XR users / devices and idle / busy period of XR devices

[0097] Source Prompt Type: Complex

[0098] Translated prompt: Optimize idle-to-active transition states and adaptive energy profiles

[0099] Translated prompt type: Complex

[0100] Translated simple Operation List: Null

[0101] }

[0102] Steps 3 and 4: Similar to steps 1 and 2, GenAI Agent 2 (302-2) further sends a prompt(s) to the next GenAI Agent (302-3), which comprehends the prompt(s), further distills them, and adds the identify information to the prompt metadata. For example, in step 4, the NF3 / GenAI Agent 3 (302-3) will append the following metadata to already received metadata from step 3:

[0103] { Level : 3 - since NF2 / GenAI Agent 2 received and distilled / translated Level 1 prompt from NFl / GenAI Agent 1Source GenAI: NF3 ID (GenAI 3 ID)

[0104] Destination GenAI: NF4 ID (GenAI 4 ID)

[0105] Source Prompt: Optimize idle-to-active transition states and adaptive energy profiles

[0106] Source Prompt Type: Complex

[0107] Translated prompt: activate SON module at gNB for optimization of idle-to-active transition states and adaptive energy profiles

[0108] Translated prompt type: Simple

[0109] Translated simple Operation List: Create

[0110] }

[0111] Steps 3 and 4: Similar to steps 1 and 2, GenAI Agent 2 (302-2) further sends a prompt(s) to the next GenAI Agent (302-3), which comprehends the prompt(s), further distills them, and adds the identify information to the prompt metadata. For example, in step 4, the NF3 / GenAI Agent 3 (302-3) will append the following metadata to already received metadata from step 3:

[0112] { Level : 3 - since NF2 / GenAI Agent 2 received and distilled / translated Level 1 prompt from NF 1 / GenAI Agent 1

[0113] Source GenAI: NF3 ID (GenAI 3 ID)

[0114] Destination GenAI: NF4 ID (GenAI 4 ID)

[0115] Source Prompt: Optimize idle-to-active transition states and adaptive energy profiles

[0116] Source Prompt Type: Complex

[0117] Translated prompt: activate SON module at gNB for optimization of idle-to-active transition states and adaptive energy profiles

[0118] Translated prompt type: Simple

[0119] Translated simple Operation List: Create

[0120] }

[0121] Step 5: the GenAI Agent 3 (302-3) has distilled the prompt into a simple operation and sends the prompt to GenAI Agent 4 / NF4 (302-4). In this case, the prompt translation results in one simple operation of creating a JavaScript Object Notation (JSON) management object instance at the gNB. However, there can be other simple operations that can result from prompt translation, e.g., configuration change at an NF such as changing the antenna tilt parameter at the gNB orrequesting a management data analytics service (MDAS) for energy saving recommendations in order to contribute to task of energy minimization.

[0122] Step 6: Each of the GenAI Agents in the hierarchy of translating the complex task into simple tasks, i.e., GenAI Agents 1, 2, 3, 4, can request certain services based on the prompt they need to handle. For example, GenAI Agent 3 can request the MDAS service for energy saving recommendations or, as indicated in FIG. 3A, the GenAI Agent 4 (302-4) may request creation of specific JSON module at gNB. However, as the GenAI Agent 3 / GenAI Agent 4 (302-3 / 302-4) that are explicitly requesting the service are not the only entities that are involved in such service request, but rather the service request came as an outcome of complex prompt translation involving GenAI Agents 1 (302-1) and 2 (302-2), they can be considered as the implicit service requestors. Therefore, service authorization needs to take into account all involved entities, e.g., NRF 306 needs to verify if all GenAI Agents 1, 2, 3 and related NFs / UEs / applications (e.g., NF 304) are authorized to consume the requested services. Such comprehensive service authorization is performed based on the prompt tracking metadata that was updated by all the entities on the path from complex prompt towards simple prompt. NRF 306 then also verifies if the prompt tracking data matches the CCA(s) received from the intermediary GenAI NFs or not.

[0123] Step 7 : Only if the service request is authorized, the NF4 / GenAI Agent 4 (302-4) executes the needed operation, e.g., creation of JSON managed object instance.

[0124] Step 8: The authorization entity (e.g., NRF 306) can give feedback to all entities involved in the prompt translation process with respect to authorization-critical parts of the prompt metadata. For example, NRF 306 can help identify which prompt translation path is the most favorable from an authorization point of view. That is, if GenAI Agent 2 (302-2) resides in an unfavorable geographic location, and thus the authorization of the service request having GenAI Agent 2 (302-2) as one of the originators is critical, the GenAI Agent 2 (302-2) should be avoided in the future prompt translations. The authorization function can provide such feedback toward the GenAI Agents such that they avoid the critical GenAI Agent in the future.

[0125] Turning to FIG. 3B, the centralized prompt translation tracking portion (performed instead of the distributed portion in FIG. 3A or complementary thereto) of procedure 300 includes:

[0126] Step 9: The GenAI Agent 1 (302-1) provides a complex prompt, e.g., related to minimization of energy consumption of XR service to the GenAI Agent 2

[0127] Step 10: The GenAI Agent 1 (302-1) provides to the PTTF 308, the metadata of the prompt (as described in step 1).Step 11 : The GenAI Agent 2 (302-2) comprehends and distills the complex prompt into one or more simpler prompts and provides such prompt(s) to other GenAI Agents accordingly.

[0128] Step 12 through 16: Similar to steps 9 through 11 with the difference being that messages are exchanged between different network entities.

[0129] Step 17: PTTF 308 stores all the information related to prompt metadata. In such a way, complete information on prompt levels, source, destination, complexity, etc., on the path of the prompt being distilled step by step over different network entities is captured.

[0130] Step 18: Any network entity may request a service based on the prompt it handles, e.g., GenAI Agent 3 (302-3) or a related network entity / NF can request the MDAS service for energy saving recommendations.

[0131] Step 19: NRF 306, in order to process the service request and authorize the service request accordingly, fetches the information on prompt metadata related to all network entity / GenAI Agents that were involved in the translation / distillation from the complex prompt to the simple service request from the PTTF 308.

[0132] Step 20: NRF 306 performs service authorization based on the fetched prompt tracking information metadata from step 20 taking into account and checking authorization rights of all entities (e.g., GenAI Agents 1, 2, 3) involved in the translation of the complex prompt towards the simple service request.

[0133] Step 21: Only if the service request is authorized, the requestor NF / GenAI Agent executes the needed operation.

[0134] Step 22: The authorization function (NRF 306) can provide the feedback toward the entities on the path in translation from complex prompt to simple prompt in the case that there are entities critical from the authorization point of view. Such entities can be avoided in the future prompt translations.

[0135] FIGS. 4 and 5 show alternative embodiments, e.g., a prompt management procedure 400 and a prompt management procedure 500, which are variants of procedure 300 wherein a UE acts as the prompt source (originator) at the initial prompt path level (e.g., level 1 where the complex prompt is first introduced).

[0136] More particularly, FIG. 4 illustrates procedure 400 involving a plurality of Al agents 402-1, 402-3, and 402-4 (e.g., GenAI Agent 1 on UE1, GenAI Agent 3 on AMF, and GenAI Agent 4 on PCF), a User Plan Function (UPF) 404, and an NRF 406. There is no GenAI Agent 2 shown in this example. Steps 1-8 in procedure 400 are the same or similar to steps 1-8 in procedure 300with the GenAI Agent 1 (402-1) in UE1 providing prompt metadata, as part of a Non-Access Stratum (NAS) message containing all information related to the prompt, directly towards the AMF (GenAI Agent 3 (402-3)) of the communication network.

[0137] FIG. 5 illustrates procedure 500 involving a plurality of Al agents 502-1, 502-2, 502-3, and 502-4 (e.g., GenAI Agent 1 on UE1, GenAI Agent 2 on a RAN NF (e.g., gNB), GenAI Agent 3 on AMF, and GenAI Agent 4 on PCF), a UPF 504, and an NRF 506. Steps 1-8 of procedure 500 are a variation on steps 1-8 of procedure 400 with the difference being that the GenAI Agent 1 (502-1) in UE1 provides the prompt metadata containing all information related to the prompt as part of a NAS message to the gNB (GenAI Agent 2 (502-2)) which then forwards the message with the prompt data towards the AMF (GenAI Agent 3 (502-3)) of the communication network.

[0138] FIG. 6 illustrates a prompt management procedure 600 in a communication network with user equipment-sourced prompts according to yet another illustrative embodiment. More particularly, procedure 600 involves a plurality of Al agents 602-1, 602-3, and 602-4 (e.g., GenAI Agent 1 on UE1, a GenAI Agent 3 on a new NF, and GenAI Agent 4 on PCF), an Application Function (AF) 604, a Network Exposure Function (NEF) 606, a UPF 608, and an NRF 610. Steps 1-8 of procedure 600 are a variation on steps 1-8 of one or more of procedures 300, 400 or 500 with the difference being that GenAI Agent 1 on UE1 (602-1) can request the service via an AF / NEF based approach. More particularly, in this case, the GenAI Agent 1 on UE1 (602-1) sends the service request to AF 604, and AF 604 sends a service message to NEF 606 which then goes on to the GenAI Agent 3 on new NF 602-3. The remainder of the flow is the same as described above.

[0139] Eastly, FIG. 7 illustrates a prompt management procedure 700 in a communication network with user equipment- sourced prompts according to a further illustrative embodiment. Procedure 700 involves GenAI Agent 1 702 on UE1, a UE2704, and another NF 706 (AMF, AF, NEF, etc.). In this variation, UE1 and UE2 are assumed to have a direct communication link, independent of the cellular link between a UE and a communication network, e.g., a PC5 communication protocol. In such a case, GenAI Agent 1 702 UE1 (complex prompt generator / originator) sends the prompt (e.g., first level translation in the case that the translation is also done at the UE1 device) to the neighboring UE2704, which can then relay it to either AMF (NF 706) (via NAS message) or to AF / NEF.

[0140] In some embodiments, an apparatus comprises at least one processor and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatusat least to: generate a prompt; generate tracking data, the tracking data indicative of the generated prompt; associate the tracking data with the generated prompt; and send the generated prompt with the associated tracking data to an access and mobility management entity in a communication network. The apparatus is user equipment and the generated prompt is a generative artificial intelligence prompt. Further, in some embodiments, the generated prompt with the associated tracking data may be sent directly to the access and mobility management entity in a non-access stratum message.

[0141] In some embodiments, the tracking data further includes one or more of an indicator of a level of complexity of the generated prompt, a source of the generated prompt and a destination of the generated prompt, and an indicator of a translation level, relative to a hierarchy of translation levels, associated with the generated prompt.

[0142] In some embodiments, the apparatus is further caused to generate a set of credentials that include the generated prompt and at least a portion of the tracking data,

[0143] In some embodiments, the apparatus is further caused to participate in a service authorization with an authorization entity in the communication network based on at least a portion of the tracking data, and to receive a notification from the authorization entity indicative of service authorization results, the notification including an indication of any nodes in the communication network to avoid relative to subsequent translation of prompts.

[0144] In some embodiments, associating the tracking data with the generated prompt further includes appending the tracking data to the generated prompt.

[0145] In some embodiments, an apparatus and / or a method includes generating, at user equipment, a prompt; generating, at the user equipment, tracking data, the tracking data indicative of the generated prompt; associating, at the user equipment, the tracking data with the generated prompt; and sending, from the user equipment, the generated prompt with the associated tracking data to an access and mobility management entity in a communication network. The generated prompt is a generative artificial intelligence prompt. In some embodiments, the generated prompt with the associated tracking data is sent directly to the access and mobility management entity in a non-access stratum message.

[0146] In some embodiments, an apparatus and / or a method includes generating, at user equipment, a prompt; generating, at the user equipment, tracking data, the tracking data indicative of the generated prompt; associating, at the user equipment, the tracking data with the generated prompt; and sending, from the user equipment, the generated prompt with the associated trackingdata toward an access and mobility management entity in a communication network via a radio access node of a radio access network. The generated prompt is a generative artificial intelligence prompt.

[0147] In some embodiments, an apparatus and / or a method includes generating, at user equipment, a prompt; generating, at the user equipment, tracking data, the tracking data indicative of the generated prompt; associating, at the user equipment, the tracking data with the generated prompt; and sending, from the user equipment, the generated prompt with the associated tracking data toward an entity in a communication network via an application function and a network exposure function in the communication network. The generated prompt is a generative artificial intelligence prompt.

[0148] In some embodiments, an apparatus and / or a method includes generating, at first user equipment, a prompt; generating, at the first user equipment, tracking data, the tracking data indicative of the generated prompt; associating, at the first user equipment, the tracking data with the generated prompt; and sending, from the first user equipment, the generated prompt with the associated tracking data toward an entity in a communication network via second user equipment with which the first user equipment is connected via a non-cellular communication link. The generated prompt is a generative artificial intelligence prompt.

[0149] Given the inventive teachings herein, it is to be understood that any of the individual features or functionalities described above can be combined in a straightforward to realize other embodiments not expressly shown in the figures or not otherwise expressly described.

[0150] It is to be appreciated that the particular processing operations and other system functionality described in conjunction with the diagrams described herein are presented by way of illustrative example only and should not be construed as limiting the scope of the disclosure in any way. Alternative embodiments can use other types of processing operations and messaging protocols. For example, the ordering of the steps may be varied in other embodiments, or certain steps may be performed at least in part concurrently with one another rather than serially. Also, one or more of the steps may be repeated periodically, or multiple instances of the methods can be performed in parallel with one another.

[0151] It should again be emphasized that the various embodiments described herein are presented by way of illustrative example only and should not be construed as limiting the scope of the claims. For example, alternative embodiments can utilize different communication system configurations, user equipment configurations, base station configurations, authorization processes, messagingprotocols and message formats than those described above in the context of the illustrative embodiments. These and numerous other alternative embodiments within the scope of the appended claims will be readily apparent to those skilled in the art.

Claims

I / We Claim:

1. An apparatus comprising:at least one processor; andat least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to:generate a prompt;generate tracking data, the tracking data indicative of the generated prompt; associate the tracking data with the generated prompt; andsend the generated prompt with the associated tracking data to an access and mobility management entity in a communication network;wherein the apparatus is user equipment and the generated prompt is a generative artificial intelligence prompt.

2. The apparatus of claim 1, wherein the generated prompt with the associated tracking data is sent directly to the access and mobility management entity in a non-access stratum message.

3. The apparatus of claim 1, wherein the tracking data further includes an indicator of a level of complexity of the generated prompt.

4. The apparatus of claim 1, wherein the tracking data further includes a source of the generated prompt and a destination of the generated prompt.

5. The apparatus of claim 1, wherein the tracking data further includes an indicator of a translation level, relative to a hierarchy of translation levels, associated with the generated prompt.

6. The apparatus of claim 1, wherein the apparatus is further caused to generate a set of credentials that include the generated prompt and at least a portion of the tracking data.

7. The apparatus of claim 1, wherein the apparatus is further caused to participate in a service authorization with an authorization entity in the communication network based on at least a portion of the tracking data.

8. The apparatus of claim 7, wherein the apparatus is further caused to receive a notification from the authorization entity indicative of service authorization results, the notification including an indication of any nodes in the communication network to avoid relative to subsequent translation of prompts.

9. The apparatus of claim 1, wherein associating the tracking data with the generated prompt further includes appending the tracking data to the generated prompt.

10. An method comprising:generating, at user equipment, a prompt;generating, at the user equipment, tracking data, the tracking data indicative of the generated prompt;associating, at the user equipment, the tracking data with the generated prompt; and sending, from the user equipment, the generated prompt with the associated tracking data to an access and mobility management entity in a communication network;wherein the generated prompt is a generative artificial intelligence prompt.

11. The method of claim 10, wherein the generated prompt with the associated tracking data is sent directly to the access and mobility management entity in a non-access stratum message.

12. The method of claim 10, wherein the tracking data further includes an indicator of a level of complexity of the generated prompt.

13. The method of claim 1, wherein the tracking data further includes a source of the generated prompt and a destination of the generated prompt.

14. The method of claim 10, wherein the tracking data further includes an indicator of a translation level, relative to a hierarchy of translation levels, associated with the generated prompt.

15. The method of claim 10, further comprising generating a set of credentials that include the generated prompt and at least a portion of the tracking data.

16. The method of claim 10, further comprising participating, by the user equipment, in a service authorization with an authorization entity in the communication network based on at least a portion of the tracking data.

17. The method of claim 16, further comprising receiving, at the user equipment, a notification from the authorization entity indicative of service authorization results, the notification including an indication of any nodes in the communication network to avoid relative to subsequent translation of prompts.

18. The method of claim 10, wherein associating the tracking data with the generated prompt further includes appending the tracking data to the generated prompt.

Images