Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
A preliminary amendment to the claims was received on 11/06/2024 and is acknowledged by the Examiner for examination purposes. In the preliminary amendment, Claims 1-62 have been cancelled, and Claims 63-77 are newly added.
Priority
The instant application 18863424 is a 371 of PCT Application PCT/EP/2023/062211, which claims priority to the foreign application IN 202241027671, which claims the priority filing date of 05/13/2022. Therefore, the effective filing date of the instant application is 05/13/2022.
Oath/Declaration
Applicant’s oath/declaration filed on 11/06/2024 has been reviewed by the examiner and is found to conform to the requirements prescribed in 37 C.F.R. 1.63.
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 12/16/2024 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Drawings
The drawings submitted on 11/06/2024 with the instant application are acceptable for examination purposes.
Specification
The specification submitted on 11/06/2024 with the instant application are acceptable for examination purposes.
Claim Rejections - 35 USC § 101
Claims 63-77 are rejected under 35 USC 101 because the claimed invention is directed to abstract ideas without significantly more.
Claim 63 is directed to an apparatus for conditionally prohibiting federated learning of a model by a terminal which performs steps including checking, monitoring, and prohibiting federated learning. The recited apparatus is a machine which falls within one of the statutory categories.
Claim 63 recites judicial exceptions in the checking, monitoring, and prohibiting steps performed by the apparatus. These processes, under broadest reasonable interpretation, cover performance of the limitations in the mind except for the recitation of generic computer components. That is, the recited limitations “checking whether an authorization for performing a federated learning of a model by a terminal is received from a first network element”; “monitoring whether a request for the performing the federated learning of the model by the terminal is received”; and “prohibiting the performing the federated learning of the model by the terminal … “ are all interpretable as mental processes. The recited steps of “checking …” and “monitoring …” merely list observable determinations performed by the generically stated apparatus. The “prohibiting …” step represents an application of conditional logic to decide whether or not to prevent federated learning by a terminal. Therefore, the claim, as a whole, represents an abstract idea as it only covers performance of simple observations and a rule-based decision to apply basic device functionality (not performing federated learning).
These judicial exceptions are not integrated into a practical application. Additional elements of the claim include the apparatus, processor, memory, terminal, first network element, model, and federated learning. All of the hardware elements are recited at a high level of generality. The model and federated learning are broadly stated additions and are not sufficient to narrow the interpretable environment or device functionality of the claim. Therefore, these additional elements do not serve to integrate the abstract ideas into a practical application because they do not serve to impose any meaningful limits on practicing the abstract ideas.
The claim does not incorporate the additional elements in a manner that is sufficient to amount to significantly more than the judicial exceptions. The additional elements, as stated above, are recited at a high level of generality—the claim language does not meaningfully connect the abstract checking, monitoring, and prohibiting steps to the operation of the apparatus. Therefore, nothing in the claim adds significantly more than the abstract ideas, and the claim is ineligible.
Claims 64-73 are also rejected due to their dependence on Claim 63.
Claim 74 is also directed to an apparatus for conditionally prohibiting federated learning of a model by a terminal which performs steps including monitoring, checking, and refusing to authorize federated learning. The recited apparatus is a machine which falls within one of the statutory categories.
Claim 74 recites judicial exceptions in the monitoring, checking, and refusing steps performed by the apparatus. These processes, under broadest reasonable interpretation, cover performance of the limitations in the mind except for the recitation of generic computer components. That is, the recited limitations “monitoring if a request for authorizing performing federated learning of a first model by a terminal is received from an application function …”; “checking whether the requirement fits to a relevant limitation for the performing the federated learning of the first model by the terminal if the request is received”; and “refusing the authorizing the performing the federated learning of the first model by the terminal if the requirement does not fit the relevant limitation“ are all interpretable as mental processes. The recited steps of “monitoring …” and “checking …” merely list observable determinations performed by the generically stated apparatus. Further, the checking step itself is performed based on conditional logic. The “refusing …” step is also an application of conditional logic to decide not to accept authorizing of federated learning by a terminal. Therefore, the claim, as a whole, represents an abstract idea as it only covers performance of simple observations and rule-based decision making to apply basic device functionality (checking a requirement and/or refusing authorizing of federated learning).
These judicial exceptions are not integrated into a practical application. Additional elements of the claim include the apparatus, processor, memory, terminal, application function, requirement, resource, limitation, model, and federated learning. All of the hardware elements are recited at a high level of generality. The application function, requirement, resource, limitation, model, and federated learning are broadly stated additions are not sufficient to narrow the interpretable environment or device functionality. Therefore, these additional elements do not serve to integrate the abstract ideas into a practical application because they do not serve to impose any meaningful limits on practicing the abstract ideas.
The claim does not incorporate the additional elements in a manner that is sufficient to amount to significantly more than the judicial exceptions. The additional elements, as stated above, are recited at a high level of generality—the claim language does not meaningfully connect the abstract monitoring, checking, and refusing steps to the operation of the apparatus. Therefore, nothing in the claim adds significantly more than the abstract ideas, and the claim is ineligible.
Claims 75 is also rejected due to its dependence on Claim 74.
Claim 76 is directed to an apparatus for monitoring terminal interactions with a database and providing a limitation upon request from a terminal. The recited apparatus is a machine which falls within one of the statutory categories.
Claim 76 recites judicial exceptions in the monitoring and supervising steps performed by the apparatus. These processes, under broadest reasonable interpretation, cover performance of the limitations in the mind except for the recitation of generic computer components. The recited limitations “monitoring whether a database receives an overall limitation for performing federated learning of any model by a terminal” and “supervising whether the database receives a request to provide a first limitation for performing federated learning of a first model by the terminal” are interpretable as mental processes. The recited steps of “monitoring …” and “supervising …” are human-capable observations. The additional recited limitations “storing the overall limitation in the database if the overall limitation is received” and “providing the first limitation in response to the receiving the request …” are merely data communicating recited at a high level of generality, and thus are insignificant extra-solution activity. Therefore, the claim, as a whole, represents an abstract idea as it only covers performance of observations except for the recitation of generic computer components and their capabilities.
These judicial exceptions are not integrated into a practical application. Additional elements of the claim include the apparatus, processor, memory, database, terminal, limitation (overall or relevant), model, federated learning, and the storing and providing steps. All of the hardware elements are recited at a high level of generality. The limitation(s), model, and federated learning are broadly stated additions and are not sufficient to narrow the interpretable environment or device functionality of the claim. As noted above, the storing and providing steps serve as insignificant extra-solution activity performable by generically stated computer components. Therefore, these additional elements do not serve to integrate the abstract ideas into a practical application because they do not serve to impose any meaningful limits on practicing the abstract ideas.
The claim does not incorporate the additional elements in a manner that is sufficient to amount to significantly more than the judicial exceptions. The additional elements, as stated above, are recited at a high level of generality—the claim language does not meaningfully connect the abstract monitoring and supervising steps to the operation of the apparatus. The storing and providing steps are conditional decisions that represent basic functionality of known devices, and thus do not amount to significantly more than the judicial exceptions. Therefore, the claim is ineligible.
Claims 77 is also rejected due to its dependence on Claim 76.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claim(s) 63-64 and 71-73 is/are rejected under 35 U.S.C. 103 as being unpatentable over Estevez et al. (US 20250274351 A1), hereinafter Estevez, in view of Choi et al. (US 20250141876 A1), hereinafter Choi.
Regarding Claim 63:
Estevez teaches an apparatus comprising: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to perform (Estevez – Paragraph [0243]: The apparatus may comprise a memory 1210, a transceiver 1220, and a controller 1230; and Paragraph [0244]: The memory 1210 may store at least one of information transmitted/received via the transceiver 1220 and information generated via the controller 1230. The memory may further store information/data/commands/instructions to be executed by the controller 1230 to cause the apparatus 1200 to perform operations described herein): checking whether an authorization for performing a federated learning of a model (Estevez – Paragraph [0013]: Federated learning (FL) is learning in distributed data environments. Referring to FIG. 1, in a federated learning system, multiple client data sources collaborate to learn a predictive model) by a terminal is received from a first network element (Estevez – Figure 7: illustration of a process for request performance of federated learning by a UE; and Paragraph [0191]: 702. The UE decides to request to join an AI/ML session for FL; and Paragraph [0194]: 705. The AMF requests authorization from the AFLSF entity for the AMSID to be used by the UE; and Paragraph [0200]: 708. The AFLSF entity provides a response to the AMF indicating whether the UE request to join the FL group is authorized); monitoring whether a request for the performing the federated learning of the model by the terminal is received (Estevez – Figure 7: illustration of a process for request performance of federated learning by a UE; and Paragraph [0191]: 702. The UE decides to request to join an AI/ML session for FL).
Estevez does not expressly teach and prohibiting the performing the federated learning of the model by the terminal if at least one of: the authorization for the federated learning of the model by the terminal is not received, and the request for the performing the federated learning of the model by the terminal is not received.
However, Choi teaches and prohibiting the performing the federated learning of the model by the terminal if at least one of: the authorization for the federated learning of the model by the terminal is not received, and the request for the performing the federated learning of the model by the terminal is not received (Choi – Paragraph [0062]: the server 200 may compare the separate signature variable calculated by the server 200 with the signature variable included in the update information of the signed update federated learning transmitted from the client terminal 101 to determine whether the signatures match; and Paragraph [0063]: if it confirms that the signature does not match, the server 200 may reject the request for participation in the federated learning update transmitted by the client terminal 101 and transmit a rejection message according to the rejection to the client terminal 101).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to modify Estevez, further incorporating Choi to arrive at the conclusion of the claimed invention. One would be motivated to incorporate Choi’s explicit rejection of federated learning participation by terminals determined to be unauthorized based on a required condition with Estevez’s techniques for ensuring secure and efficient federated learning of a model. This addition would ensure security of legitimate participating devices against unauthorized devices.
Regarding Claim 64:
The combination of Estevez and Choi teaches the apparatus according to claim 63.
Estevez further teaches further configured to perform: instructing the performing the federated learning of the model if the authorization for the federated learning of the model is received and the request for the federated learning of the model is received (Estevez – Figure 8: Additional illustration of UE requesting to join a FL session; and Paragraph [0207]-[0209]: [0207] 804. The SMF sends a validation request to the AFLSF entity for the AMSID to be used by the UE in order to determine whether the UE is allowed to engage in FL using the AMSID [0208] 805. When the AFLSF entity receives an AMSID from the SMF, the AFLSF entity validates if the AMSID can be used for the UE: [0209] The AFLSF entity may indicate whether the request should be granted or not; and Paragraph [0214]: 809. The UE starts the FL process after the establishment of the user plane resources).
The motivation to combine the arts is the same as that of Claim 63.
Regarding Claim 71:
The combination of Estevez and Choi teaches the apparatus according to claim 63.
Estevez further teaches wherein the checking comprises checking whether the authorization is received from the first network element via a non-access stratum container or as parameter update data for the terminal (Estevez – Paragraph [0152]: At any point in time, the network may decide to revoke the use of FL by the UE. The AF may indicate this using the NEF framework, and/or the UDM may inform the AMF about this, or the AFLSF entity may be informed by the AMF, and/or the AFLSF entity may inform the AMF or SMF, etc. The AMF/SMF may in turn inform the UE about the revocation to use FL using any new indication and/or NAS message. For example, the AMF may send an empty AMSID content which signifies the need to revoke previous AMSID information, or the AMF may send AMSID information with an operation code to delete the content in the UE which then means that the UE would not have this information and hence cannot use FL. Similarly, when sending valid AMSID information to the UE, the AMF may indicate an operation code such as “create new AMSID”, “update existing AMSID”, etc, which informs the UE about the actual action that needs to be taken. The UE should process the received action (i.e. operation code) and AMSID accordingly, e.g. if the operation code is to update or create an AMSID, then the UE updates the AMSID or creates a new AMSID respectively and stores/uses this information accordingly).
The motivation to combine the arts is the same as that of Claim 63.
Regarding Claim 72:
The combination of Estevez and Choi teaches the apparatus according to claim 63.
Estevez further teaches wherein the first network element comprises an access and mobility management function, AMF, or a session management function, SMF (Estevez – Paragraph [0152]: At any point in time, the network may decide to revoke the use of FL by the UE. The AF may indicate this using the NEF framework, and/or the UDM may inform the AMF about this, or the AFLSF entity may be informed by the AMF, and/or the AFLSF entity may inform the AMF or SMF, etc. The AMF/SMF may in turn inform the UE about the revocation to use FL using any new indication and/or NAS message. For example, the AMF may send an empty AMSID content which signifies the need to revoke previous AMSID information, or the AMF may send AMSID information with an operation code to delete the content in the UE which then means that the UE would not have this information and hence cannot use FL. Similarly, when sending valid AMSID information to the UE, the AMF may indicate an operation code such as “create new AMSID”, “update existing AMSID”, etc, which informs the UE about the actual action that needs to be taken. The UE should process the received action (i.e. operation code) and AMSID accordingly, e.g. if the operation code is to update or create an AMSID, then the UE updates the AMSID or creates a new AMSID respectively and stores/uses this information accordingly).
The motivation to combine the arts is the same as that of Claim 63.
Regarding Claim 73:
The combination of Estevez and Choi teaches the apparatus according to claim 63.
Estevez further teaches wherein the apparatus is included in the terminal, or the apparatus is the terminal (Estevez – Paragraph [0152]: At any point in time, the network may decide to revoke the use of FL by the UE. The AF may indicate this using the NEF framework, and/or the UDM may inform the AMF about this, or the AFLSF entity may be informed by the AMF, and/or the AFLSF entity may inform the AMF or SMF, etc. The AMF/SMF may in turn inform the UE about the revocation to use FL using any new indication and/or NAS message. For example, the AMF may send an empty AMSID content which signifies the need to revoke previous AMSID information, or the AMF may send AMSID information with an operation code to delete the content in the UE which then means that the UE would not have this information and hence cannot use FL. Similarly, when sending valid AMSID information to the UE, the AMF may indicate an operation code such as “create new AMSID”, “update existing AMSID”, etc, which informs the UE about the actual action that needs to be taken. The UE should process the received action (i.e. operation code) and AMSID accordingly, e.g. if the operation code is to update or create an AMSID, then the UE updates the AMSID or creates a new AMSID respectively and stores/uses this information accordingly).
The motivation to combine the arts is the same as that of Claim 63.
Claim(s) 65-66 and 68-70 is/are rejected under 35 U.S.C. 103 as being unpatentable over Estevez in view of Choi and 3GPP (“3rd Generation Partnership Project; Technical Specification Group Services and System Aspects; Study on 5G System Support for AI/ML-based Services (Release 18), 3GPP TR 23.700-80, V0.2.0, April 2022, pp. 1-93), hereinafter 3GPP.
Regarding Claim 65:
The combination of Estevez and Choi teaches the apparatus according to claim 64.
The combination of Estevez and Choi does not expressly teach further configured to perform: providing a limitation for the performing the federated learning of the model to a first application function.
However, 3GPP teaches further configured to perform: providing a limitation for the performing the federated learning of the model to a first application function (3GPP – P. 86-87: 1. AI/ML Application Server invokes Nnef_AIMLAssistanceInfo_Request to request 5G assistance information with specific conditions for FL operation. The AI/ML Application Server requests FL assistance information for the UE list included in the request. It also indicates that these UEs are authorized by AI/ML Application Server (application side) for FL operation. The input parameters of the Nnef_AIMLAssistanceInfo_Request service operation can be as below: - AF Identifier. - UE List (GPSIs, External Group ID, or IP addresses). - AF-Service-Identifier. - External Application Identifier. - QoS Reference. - Whether to request the list of candidate UEs (If indicated, AIML NF/NEF provides candidate UE list in the response). - Time when assistance information is needed (Indicates to the AIML NF/NEF the latest time the analytics consumer expects to receive assistance data provided by AIML NF/NEF). - Expected AIML operation time duration (Expected time duration by AI/ML Application Server based on the training information such as training model and training data size). - Preferred AIML operation time periods (Time periods that the AI/ML Application Server want to perform AIML operation). - Whether to request the recommended AIML operation time periods (If indicated, AIML NF/NEF provides recommended AIML operation time periods in the response). - Traffic volume for downloading global model and uploading training results. - DNN and S-NSSAI. - Expected UE Moving Trajectory for each UE in the UE list (e.g. A planned path of movement). - Area of Interest (AIML operation for the UEs in the area of interest). - Minimum number of UEs needed for FL operation; and P. 87-88: 8. AIML NF/NEF responds to the AI/ML Application Server with the derived assistance information. The output parameters of the Nnef_AIMLAssistanceInfo_Request service operation can be as below: - Set of the 5GS Assistance information requested by AI/ML Application Server. Each 5GS Assistance information can contain: - List of candidate UEs for FL operation. - Recommended time period to perform AIML operation. - Recommended area for AIML operation. The lists of candidate UEs for recommended FL operation can be different according to the recommended time period and/or the recommended area for AIML operation. If the "time when assistance information is needed" was provided in step 1, AIML NF/NEF takes this into account in responding to the AI/ML Application Server. Based on the received the 5GS assistance information, the AI/ML Application Server can select and manage the group of UEs, and determine the start time for FL operation).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to modify Estevez and Choi, further incorporating 3GPP to arrive at the conclusion of the claimed invention. One would be motivated to incorporate 3GPP’s imposition of conditional requirements for UEs to perform federated learning with Estevez and Choi’s techniques for ensuring secure and efficient federated learning of a model. This additional consideration would further enhance the efficacy of the federated learning by the terminal(s) and add security by reducing potentially unnecessary exposure of device data.
Regarding Claim 66:
The combination of Estevez, Choi, and 3GPP teaches the apparatus according to claim 65.
3GPP further teaches wherein the limitation comprises at least one of a limitation of a proportion of a first resource to be used for the federated learning of the model and a limitation of an access to data on the terminal to be used for the federated learning of the model (3GPP – P. 56-57: Federated learning (FL) is a group-based learning method and there will be many files and multiple rounds of interactions between the UEs and the FL server over 5GS during the FL process. Just like any other services, there should be SLA between the MNO and ASP to agree on the allowance of the maximum throughput that can be consumed between the end points (i.e. the group of UEs and the AS) of the given FL operation … Figure 6.16.1-1 below, illustrates an example of 10 UEs in total within the service area of a UPF that could be selected to participate in the given FL operation. In a certain round, AS selects 5 of them to be participated in the FL training. In order to ensure that the resources consumed by the 5 UEs are within the specified throughput limit, the UPF keeps track of aggregate bit rate against the Group-MBR among the set of QoS flows of the UEs who participate in the given FL operation. The aggregate bit rate shall not exceed the pre-determined Group-MBR for that specific FL operation. The following summarizes how the aggregated bit rate among the set of QoS flows of the UEs is monitored against Group-MBR: … As shown in the figure 6.16.1-2, Group-MBR is defined as the upper bound of the aggregate bit rate across all GBR and Non-GBR QoS Flows corresponding to the group of PDU Sessions of the UEs who participate in a group of transmissions concurrently (e.g. Federated Learning (FL) operation) with active user planes … In the following figures, the UEs participating in a given Federated Learning training session may consume different aggregate bit rate during each round of the FL operation, however, the aggregate bit rate among the corresponding set of QoS flows for each round of the FL operation should not exceed the contracted Group-MBR that was agreed between the ASP and the serving MNO. For example, the value of the contracted Group-MBR is 20 Mbps, the aggregated bit rate across the set of QoS flows for the participating UEs in each round of the FL operation can not exceed the Group-MBR i.e. 20 Mbps).
The motivation to combine the arts is the same as that of Claim 65.
Regarding Claim 68:
The combination of Estevez, Choi, and 3GPP teaches the apparatus according to claim 65.
3GPP further teaches further configured to perform: monitoring whether the performing the federated learning of the model violates the limitation; and discarding the performing the federated learning of the model if the performing the federated learning of the model violates the limitation (3GPP – P. 83: 2. The AF tasks Federated Learning Assistance Function (which could be integrated in NEF, NWDAF or be standalone) to propose UEs for FL operation. The AF indicates a list or group of UEs to choose from, the number of UEs required to perform federated learning cycle, possibly a desired location of UEs in a form of a target area for FL, duration for iteration and desired average latency for FL iterations. When providing a target area for FL, the AF may provide sub-areas, and provide a percentage of UEs that should take part in FL from each sub-area, and/or a minimum number of UEs and /or a maximum number of UEs that should take part in FL from each sub-area. Optionally, the AF may provide time when to start federated learning, duration of federated learning, size of input model and/or output model … 10.The AF sends to suggested UEs a request to start federated learning cycle(s) with input model and receives output model from each UE for each learning cycle. 11-12.The Federated Learning Assistance Function monitors whether suggested UEs remain available (registered, active, reachable), or still meet other selection criteria e.g. area of interest. If UEs become unavailable or no longer meet selection criteria the Federated Learning Assistance Function notifies the AF. The Federated Learning Assistance Function may also suggest new UEs for FL operation. 13-16.When the AF requests termination of federated learning session from Federated Learning Assistance Function, the Federated Learning Assistance Function terminates related events subscriptions and analytics subscriptions at AMF, UDM, and/or NWDAF).
The motivation to combine the arts is the same as that of Claim 65.
Regarding Claim 69:
The combination of Estevez and Choi teaches the apparatus according to claim 63.
The combination of Estevez and Choi does not expressly teach wherein the authorization indicates that a second application function is authorized to request the performing the federated learning of the model, and wherein the means are further configured to perform: informing the second application function that the authorization is received.
However, 3GPP teaches wherein the authorization indicates that a second application function is authorized to request the performing the federated learning of the model (3GPP – P. 89: 1. When an AF determines to perform the federated learning with UEs in a specific location, the AF request the PCF via NEF with QoS requirements, location information specific UE ID, external group ID if available. In addition, optionally, AF can also provide time window (e.g. start and end time) of FL operation. The NEF checks the authorization of the application server request with local policy or with a UDM. NEF forwards the request to PCF), and wherein the means are further configured to perform: informing the second application function that the authorization is received (3GPP – P.89: 2. Based on the policy, the PCF checks the AF request and determine whether the network can provide required QoS in specific location on required time window (if included) … 3. If PCF determine that it can allocate the required QoS, then the PCF sends the accept message to the AF with the agreed resources (QoS) information).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to modify Estevez and Choi, further incorporating 3GPP to arrive at the conclusion of the claimed invention. One would be motivated to incorporate 3GPP’s teaching to establish network-wide requirements or parameters for UEs regarding their performing of federated learning with Estevez and Choi’s techniques for ensuring secure and efficient federated learning of a model. This functionality would enhance the methods by creating a reference point for application functions in a 5G network to facilitate federated learning processes in corresponding terminals based on stored parameters enabling/preventing certain terminals from participating.
Regarding Claim 70:
The combination of Estevez and Choi teaches the apparatus according to claim 63.
The combination of Estevez and Choi does not expressly teach further configured to perform: informing a third application function that the performing the federated learning is prohibited if the request for the performing the federated learning of the model by the terminal is received from the third application function and the authorization for performing the federated learning of the model by the terminal is not received from the first network element.
However, 3GPP teaches further configured to perform: informing a third application function that the performing the federated learning is prohibited if the request for the performing the federated learning of the model by the terminal is received from the third application function (3GPP – P. 89: 1. When an AF determines to perform the federated learning with UEs in a specific location, the AF request the PCF via NEF with QoS requirements, location information specific UE ID, external group ID if available. In addition, optionally, AF can also provide time window (e.g. start and end time) of FL operation. The NEF checks the authorization of the application server request with local policy or with a UDM. NEF forwards the request to PCF) and the authorization for performing the federated learning of the model by the terminal is not received from the first network element (3GPP – P.89: 2. Based on the policy, the PCF checks the AF request and determine whether the network can provide required QoS in specific location on required time window (if included) … If PCF determine that it cannot allocate the required QoS on specified time or time window, the PCF rejects the AF request. In addition, PCF may provide alternative time or time window that network can provide required resources (QoS) in the rejection message).
The motivation to combine the arts is the same as that of Claim 69.
Claim(s) 67 is/are rejected under 35 U.S.C. 103 as being unpatentable over Estevez in view of Choi, 3GPP, and Ryden (US 20230262448 A1), hereinafter Ryden.
Regarding Claim 67:
The combination of Estevez, Choi, and 3GPP teaches the apparatus according to claim 65.
The combination of Estevez, Choi, and 3GPP does not expressly teach wherein the limitation is related to a category of the model.
However, Ryden teaches wherein the limitation is related to a category of the model (Ryden – Paragraph [0131]-[0137]: It will be appreciated that some wireless devices might support different ML models or types of ML models, and such support might limit the available model inputs. For example, some UEs might lack GNSS-support, and some UEs might only support a limited Neural Network, owing to memory limitations. Signalling by a UE of its capabilities to the network node may enable the node to select a suitable model based on the UE capability report. The capabilities signalled to the network node could include: [0132] UE manufacturer/model etc. [0133] Maximum consumed memory of model that can be supported. [0134] Floating point support, for example 8-bit/16-bit/32-bit float. [0135] UE computational capabilities, for example in terms of number of operations per second, type of processor (CPU, GPU), number of CPUs etc. This could be reported specifically for executing an ML model or more generally associated to the UE. [0136] Type of models supported, for example decision tree, decision forest, linear regression, feedforward neural network, recurrent neural network, convolutional neural network, etc. [0137] Maximum supported computational cost/load for executing a model. This could be expressed, for example, in terms of a number of operations and their type that the UE can perform for executing a model. The maximum supported computational cost/load can also be associated to a particular type of model. Therefore, for each model supported by the UE, the UE could report a maximum supported computational cost for executing a model. This may enable the network node to select the most appropriate model (type, dimension, etc.) for a specific UE based on the UE capabilities).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to modify Estevez, Choi, and 3GPP, further incorporating Ryden to arrive at the conclusion of the claimed invention. One would be motivated to incorporate Ryden’s teaching to consider UE limitations regarding types of models they are best suited to perform FL operations on into Estevez, Choi, and 3GPP’s apparatus for ensuring secure and efficient federated learning of a model. This addition would bolster the overall performance of the apparatus by optimizing UE selection for model training.
Claim(s) 74-75 is/are rejected under 35 U.S.C. 103 as being unpatentable over 3GPP in view of Estevez.
Regarding Claim 74:
3GPP teaches monitoring if a request for authorizing performing federated learning of a first model by a terminal is received from an application function, wherein the request comprises a requirement on a resource of the terminal or on data on the terminal for the performing the federated learning of the first model by the terminal (3GPP – P. 89: 1. When an AF determines to perform the federated learning with UEs in a specific location, the AF request the PCF via NEF with QoS requirements, location information specific UE ID, external group ID if available. In addition, optionally, AF can also provide time window (e.g. start and end time) of FL operation. The NEF checks the authorization of the application server request with local policy or with a UDM. NEF forwards the request to PCF; and P. 30: QoS requirements(e.g. includes 5QI, QoS Characteristics) indicates the QoS requirements of the QoS flow of the FL transmission … 5GS collects candidate FL members' data: … - Expected QoS estimate in the area for FL training. - QoS configuration in PCF / capability for the FL transmission QoS flow (if UEs can meet the required QoS (2MBPS) in the specified areas); and P. 32: Awareness of UE UL/DL data rate at AI/ML application server is beneficial for it to make decision on AI/ML operation … the required UL/DL data rate may be optionally provided as the QoS requirement, which may be contained in data rate monitoring information or not; Examiner’s Comment: Examiner respectfully submits that at least the listed teaching from 3GPP explicitly demonstrates requirements on at least a resource of the terminal. The QoS requirements (including 5QI, QoS characteristics, and/or required UL/DL data rates) represent parameters which define the resource needs to enable UE participation in FL operations); checking whether the requirement fits to a relevant limitation for the performing the federated learning of the first model by the terminal if the request is received (3GPP – P.89: 2. Based on the policy, the PCF checks the AF request and determine whether the network can provide required QoS in specific location on required time window (if included). The PCF may interact with the NWDAF and request the Network Performance analytics information for the Desired time window in the area of interest and other analytics such as UE related analytics as defined in TS 23.288 [6]); and refusing the authorizing the performing the federated learning of the first model by the terminal if the requirement does not fit the relevant limitation (3GPP – P. 89: If PCF determine that it cannot allocate the required QoS on specified time or time window, the PCF rejects the AF request. In addition, PCF may provide alternative time or time window that network can provide required resources (QoS) in the rejection message).
3GPP does not expressly teach an apparatus comprising: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to perform.
However, Estevez teaches an apparatus comprising: at least one processor; and at least one memory storing instructions that, when executed by the at least one processor, cause the apparatus at least to perform (Estevez – P. 7: an electronic device, as shown in FIG. 8, comprising a processor 801, a communication interface 802, a memory 803, and a communication bus 804, wherein the processor 801, the communication interface 802,and the memory 803 complete mutual communication by means of the communication bus 804,a memory 803 for storing a computer program; The processor 801 is configured to execute the program stored on the memory 803).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to modify 3GPP, further incorporating Estevez to arrive at the conclusion of the claimed invention. One would be motivated to incorporate Estevez’s organized structural elements for performing FL operations among 5G network functions and UEs in combination with 3GPP’s techniques for ensuring secure and efficient federated learning of a model. This addition would practically enable the performance of the methods taught by 3GPP in their intended environment.
Regarding Claim 75:
The combination of 3GPP and Estevez teaches the apparatus according to claim 74.
Estevez further teaches further configured to perform: authorizing the performing the federated learning of the first model by the terminal if the requirement fits the relevant limitation (Estevez – Paragraph [0109]: The AMSID may be provided either by the AF or another 5GC network function such as the AMF, SMF, or a new network function (NF) e.g. the AFLSF entity node. A UE performing FL may send the AMSID in a non-access stratum (NAS) message, which may be a 5G mobility management (5GMM) message, a 5G session management (5GSM) message or both; and Paragraph [0110]: A network node may verify if the AMSID is valid for use for FL. The verification may be performed against an existing FL UE context, which contains a set of valid AMSIDs, and/or against whether the UE is allowed to perform FL. Other conditions may also exist in the UE context, e.g. the location where FL is allowed to be performed by the UE, the time of the day, etc. This information may be available at the network either from the UDM (e.g. subscription) or from an interaction between the AF and the network, optionally via a network exposure function (NEF). The network determines if the UE is allowed to perform FL and, if so, the network grants the UE's request); and informing the terminal that the performing the federated learning of the first model is authorized if the requirement fits the relevant limitation (Estevez – Paragraph [0153]: The AMF may also inform the UE whether AI/ML (and/or FL in particular) is allowed/supported by the network).
The motivation to combine the arts is the same as that of Claim 74.
Claim(s) 76 is/are rejected under 35 U.S.C. 103 as being unpatentable over Suh et al. (US 20230308954 A1), hereinafter Suh, in view of 3GPP TS 29.503 (“3rd Generation Partnership Project; Technical Specification Group Core Network and Terminals; 5G System; Unified Data Management Services; Stage 3 (Release 17), 3GPP TS 29.503 V17.5.0, December 2021, pp. 1-466), hereinafter ‘503.
Regarding Claim 76:
Suh teaches [monitoring whether] a database receives an overall limitation for performing federated learning of any model by a terminal (Suh – Paragraph [0097]: FIG. 3 illustrates a procedure in which the AF 100 creates/updates/deletes parameters for an FL service to a UDM 130 according to an embodiment of the present disclosure; and Paragraph [0099]-[0102]: [0099] In step 301, the AF 100 may transmit the following parameters related to the FL service to the NEF 110 to request parameter creation/update/deletion: [0100] (1) an aggregated QoS request indicator, (2) a list of UE addresses or UE ids, requirement, [alternative requirements], (3) an FL group id, (4) S-NSSAI, (5) DNN, [selected aggregated QoS information], and the like. [0101] In the case that the request message transmitted by the AF 100 includes an aggregated QoS request indicator, the requirement may include a QoS reference (i.e., an identifier representing one of several predefined individual QoS parameters) or an individual QoS parameter. The AF 100 may include the following information in alternative requirements so that the network may select according to circumstances: [0102] a set composed of (QoS reference or individual QoS parameter, the number of minimum selected UEs) pairs; and Paragraph [0104]: In step 302, the NEF 110 may perform authorization on the request from the AF 100, and then transmit information included in the message received from the AF 100 to the UDM 130; and Paragraph [0107]: In step 304, if the verification is successful in step 303, the UDM 130 may store information received from the AF 100 in the UDR 140 using the following as a data key: [0108] (1) an FL group id, (2) a sub FL group id, and the like); storing the overall limitation in the database if the overall limitation is received (Suh – Paragraph [0107]: In step 304, if the verification is successful in step 303, the UDM 130 may store information received from the AF 100 in the UDR 140 using the following as a data key: [0108] (1) an FL group id, (2) a sub FL group id, and the like; and Paragraph [0119]: In step 402, the UDM 130 may receive information on a corresponding UE or PDU session from the UDR 140 based on information received from the PCF 120); supervising whether the database receives a request to provide a first limitation for performing federated learning of a first model by the terminal (Suh – Paragraph [0115]-[0116]: [0115] the PCF 120 may request subscriber information of the UE to the UDM 130. In this case, the request message may include the following information: [0116] (1) a UE id, (2) a PDU session id, (3) an FL group id, and the like); and providing the first limitation in response to the receiving the request, wherein the first limitation comprises at least one of the overall limitation and a relevant limitation for performing federated learning of the first model by the terminal, and the relevant limitation is based on the overall limitation (Suh – Paragraph [0119]: Upon recognizing the fact that the corresponding UE or the UE having the corresponding PDU session belongs to the FL group, the UDM 130 may include the following information in a response message to be transmitted to the PCF 120: [0120] (1) an aggregated QoS request indicator, (2) a list of UE addresses or UE ids, (3) requirement, [alternative requirements], (4) an FL group id, (5) S-NSSAI, (6) DNN, [selected aggregated QoS information], and the like).
Suh does not expressly teach monitoring whether a database receives an overall limitation.
However, ‘503 teaches monitoring whether a database receives an overall limitation (‘503 – P. 26: All scenarios shown in the following clauses assume that the UDM is stateful and stores information in local memory. However, the UDM may be stateless and stores information externally in the UDR. If so, the stateless UDM makes use of Nudr services as specified in 3GPP TS 29.504 [9] and 3GPP TS 29.505 [10] to retrieve required data from the UDR and store them locally before processing an incoming request. Processing the incoming request may then include updating data in the UDR or subscribing to data change notifications at the UDR by consuming the appropriate Nudr services; and P. 27: For the Nudm_SubscriberDataManagement service the following service operations are defined: - Get; Subscribe; ModifySubscription; Unsubscribe; Notification; Info; The Nudm_SubscriberDataManagement Service is used by Consumer NFs (AMF, SMF, SMSF, GMLC, NEF, 5G DDNMF) to retrieve the UE's individual subscription data relevant to the consumer NF from the UDM by means of the Get service operation. If the consumer NF supports the feature "sharedData" (see clause 6.1.8), the retrieved individual subscription data for the UE may contain shared data identifier identifying additional parts of subscription data shared by several UEs … It is also used by Consumer NFs to subscribe to notifications of data change by means of the Subscribe service operation. If the consumer NF supports the feature "sharedData" (see clause 6.1.8), the consumer NF may also subscribe to notifications of shared data change by means of the Subscribe service operation).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to modify Suh, further incorporating ‘503 to arrive at the conclusion of the claimed invention. One would be motivated to incorporate ‘503’s monitoring individual and shared UE data changes via subscription for notifications thereof with Suh’s apparatus for receiving, storing, and subsequently providing at least a limitation for performing federated learning by a UE. This additional functionality would allow network functions to identify data changes related to UE capabilities/functional limitations.
Claim(s) 77 is/are rejected under 35 U.S.C. 103 as being unpatentable over Suh in view of ‘503 and 3GPP.
Regarding Claim 77:
The combination of Suh and ‘503 teaches the apparatus according to claim 76.
The combination of Suh and ‘503 does not expressly teach wherein: the overall limitation comprises at least one of: an overall proportion of a resource to be used in total for performing the federated learning of any models by the terminal; and a limitation of an access to the data to be accessed for the performing the federated learning of any models by the terminal.
However, 3GPP teaches wherein: the overall limitation comprises at least one of: an overall proportion of a resource to be used in total for performing the federated learning of any models by the terminal; and a limitation of an access to the data to be accessed for the performing the federated learning of any models by the terminal (3GPP – P. 56-57: Federated learning (FL) is a group-based learning method and there will be many files and multiple rounds of interactions between the UEs and the FL server over 5GS during the FL process. Just like any other services, there should be SLA between the MNO and ASP to agree on the allowance of the maximum throughput that can be consumed between the end points (i.e. the group of UEs and the AS) of the given FL operation … Figure 6.16.1-1 below, illustrates an example of 10 UEs in total within the service area of a UPF that could be selected to participate in the given FL operation. In a certain round, AS selects 5 of them to be participated in the FL training. In order to ensure that the resources consumed by the 5 UEs are within the specified throughput limit, the UPF keeps track of aggregate bit rate against the Group-MBR among the set of QoS flows of the UEs who participate in the given FL operation. The aggregate bit rate shall not exceed the pre-determined Group-MBR for that specific FL operation. The following summarizes how the aggregated bit rate among the set of QoS flows of the UEs is monitored against Group-MBR: … As shown in the figure 6.16.1-2, Group-MBR is defined as the upper bound of the aggregate bit rate across all GBR and Non-GBR QoS Flows corresponding to the group of PDU Sessions of the UEs who participate in a group of transmissions concurrently (e.g. Federated Learning (FL) operation) with active user planes … In the following figures, the UEs participating in a given Federated Learning training session may consume different aggregate bit rate during each round of the FL operation, however, the aggregate bit rate among the corresponding set of QoS flows for each round of the FL operation should not exceed the contracted Group-MBR that was agreed between the ASP and the serving MNO. For example, the value of the contracted Group-MBR is 20 Mbps, the aggregated bit rate across the set of QoS flows for the participating UEs in each round of the FL operation can not exceed the Group-MBR i.e. 20 Mbps).
It would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to modify Suh and ‘503, further incorporating 3GPP to arrive at the conclusion of the claimed invention. One would be motivated to incorporate 3GPP’s teaching to monitor/cap individual and collective UE resource usage during FL operations with Suh and ‘503’s combined apparatus for receiving, storing, and subsequently providing at least a limitation for performing federated learning by a UE. This combination would enable efficient performance of federated learning among UEs without exhausting the resources of any given terminal.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Ouyang et al. (US 20230146099 A1) teaches techniques for performing federated learning in a telecom system
Kovacs et al. (US 20250247720 A1) teaches methods for performing federated learning operations with UEs, including configurations which consider the resource constraints of the participating UEs
Ali et al. (US 20240028961 A1) teaches enablement of federated learning for terminals wherein the terminals must complete federated learning tasks within at least a time limit using specified network performance related measurements
Drozd et al. (US 20220345484 A1) teaches a system for controlling FL-related access to network resources based on a risk evaluation of the requesting communication node(s)
Any inquiry concerning this communication or earlier communications from the examiner should be directed to NICHOLAS JOSEPH DILUZIO whose telephone number is (703)756-1229. The examiner can normally be reached Mon - Fri -- 7:30 AM - 5 PM.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Yin-Chen Shaw can be reached at 571-272-8878. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of published or unpublished applications may be obtained from Patent Center. Unpublished application information in Patent Center is available to registered users. To file and manage patent submissions in Patent Center, visit: https://patentcenter.uspto.gov. Visit https://www.uspto.gov/patents/apply/patent-center for more information about Patent Center and https://www.uspto.gov/patents/docx for information about filing in DOCX format. For additional questions, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
/NICHOLAS JOSEPH DILUZIO/Examiner, Art Unit 2498
/YIN CHEN SHAW/Supervisory Patent Examiner, Art Unit 2498
Prosecution Insights