DETAILED ACTION
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 .
Response to Amendment
The Amendment filed on 12/4/2025 has been entered. Claims 1 and 3-20 remain pending in the application. Applicant’s amendments to Claims have overcome each and every objection and 112(b) rejection previously set forth in the Non-Final Office Action mailed on 9/4/2025.
Response to Arguments
Applicant's arguments filed on 12/4/2025 regarding claims 1, 9 and 17 have been fully considered but they are not persuasive.
In response to applicant’s argument on pages 7-14 that the office action failed to address independent claims 1, 9 and 17 in the manner required by the MPEP. The finality of the next office action is prohibited.
On the contrary to the applicant’s arguments, independent claims 1, 9 and 17 have been amended to overcome the 35 USC 112(b) rejection set forth in the recent office action, thereby changing the scope of the invention and necessitating the reconsideration of the invention. Therefore, the action is made final.
Applicant’s arguments on pages 7-9 with respect to claim 9 have been considered but are moot upon a further consideration and a new ground of rejection made under 35 U.S.C. 103 as being unpatentable over Cheng (WO 2024/040476) in view of Zheng (US PGPub 2024/0314591), as set forth below.
Applicant argues that Cheng (WO 2024/040476) fails to teach the limitation of “a receiver circuit that receives a first Artificial Intelligence / Machine Learning (AI/ML) model that is broadcasted or groupcasted to multiple UEs by the first base station via one or more downlink RRC messages” as required by amended claim 9.
While Cheng does not expressly disclose the issued claim limitation, Zheng (US PGPub 2024/0314591) teaches that a gNB may transmit the AI model information in a system information message. The system information message may be transmitted by a SIB (system information block) in a broadcast channel. The system information message may be transmitted to UE in a RRC message when the UE is in RRC connected state. Multiple AI models may be included in system information message (Zheng, see paragraphs 0136-0138 and 0140). Therefore, Zheng teaches the limitation of “a receiver circuit that receives a first Artificial Intelligence / Machine Learning (AI/ML) model that is broadcasted or groupcasted to multiple UEs by the first base station via one or more downlink RRC messages”.
Applicant’s arguments on pages 9-10 with respect to claim 1 have been considered but are moot upon a further consideration and a new ground of rejection made under 35 U.S.C. 103 as being unpatentable over Cheng (WO 2024/040476) in view of Wang (US PGPub 2024/0146582), as set forth below.
Applicant argues that Cheng (WO 2024/040476) fails to teach the limitation of “negotiating an AI/ML model encoding method, wherein the negotiation is between the UE and the first base station” as required by amended claim 1.
While Cheng does not expressly disclose the issued claim limitation, Wang (US PGPub 2024/0146582) teaches that the network device determines, through negotiation with the terminal, to enable an information encoding feedback mechanism based on an AI encoder. Subsequently, the network device may send an AI encoder to the terminal, and the terminal may perform encoding by using the AI encoder at a negotiated moment (Wang, see paragraph 0160). Therefore, Wang teaches the limitation of “negotiating an AI/ML model encoding method, wherein the negotiation is between the UE and the first base station”.
Applicant’s arguments on pages 13-14 with respect to claim 17 have been considered but are moot upon a further consideration and a new ground of rejection made under 35 U.S.C. 103 as being unpatentable over Cheng (WO 2024/040476) in view of Pantelidou (US PGPub 2023/0300686), as set forth below.
Applicant argues that Cheng (WO 2024/040476) fails to teach the limitation of “transferring the first AI/ML model from the first base station to a second base station during a handover operation from the first base station to the second base station” as required by amended claim 17.
While Cheng does not expressly disclose the issued claim limitation, Pantelidou (US PGPub 2023/0300686) teaches that This decision/wish to receive at least one ML model may be indicated to the source network apparatus 802 as part of a Handover Request acknowledgement 8005. For example, the target network apparatus may introduce a flag value (nw-based transfer flag) in the Handover Request Acknowledgment with which it informs the source whether the ML models should be transferred over the Xn interface (nw-based transfer=1). the source network apparatus 902 may send the ML Model Set and/or the corresponding MDT Configurations used to train the Models in the ML Model Set to the target network apparatus 903. The ML model set may be the ML model(s) identified in the acknowledgment message. The ML model Set and/or the corresponding MDT configurations may be sent through the Xn interface (Pantelidou, see paragraphs 0191 and 0214). Therefore, Pantelidou teaches the limitation of “transferring the first AI/ML model from the first base station to a second base station during a handover operation from the first base station to the second base station”.
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.
Claims 1 and 3-8 are rejected under 35 U.S.C. 103 as being unpatentable over Cheng (WO 2024/040476) in view of Wang (US PGPub 2024/0146582).
Regarding claim 1, Cheng teaches a method (Cheng, see abstract, Rrc procedure design for wireless ai/ml) comprising:
establishing a radio resource control (RRC) connection by a user equipment (UE) with a first base station in a network (Cheng, see paragraphs 0031 and 0064, The UE 102 and UE 104 utilize connections (or channels) (shown as connection 108 and connection 110, respectively) with the RAN 106, each of which comprises a physical communications interface. the gNB may transfer the transparent container including the AI/ML model to the UE via a downlink RRC message);
receiving a first Artificial Intelligence / Machine learning (AI/ML) model onto the UE from the first base station via one or more downlink RRC messages (Cheng, see paragraph 0071, in step S23, the gNB may configure the trained model to the UE via a downlink RRC message);
storing the first AI/ML model, wherein the first AI/ML model belongs to a set of AI/ML models comprising: Convolution Neural Network (CNN), a Recurrent Neural Network (RNN), a Long Short Term Memory (LSTM), and a Transformer (Cheng, see paragraph 0052, A typical implementation of AI/ML is neural network (NN) , such as Conventional Neural Network (CNN) , Recurrent/Recursive neural network (RNN) , Generative Adversarial Network (GAN) , or the like); and
executing the first AI/ML model (Cheng, see paragraph 0072, the UE may store the trained AI/ML model in a memory of its modem (modulator-demodulator), and configure the modem for inference of corresponding use case).
Cheng teaches the above yet fails to teach negotiating an AI/ML model encoding method, wherein the negotiation is between the UE and the first base station.
Then Wang teaches negotiating an AI/ML model encoding method, wherein the negotiation is between the UE and the first base station (Wang, see paragraph 0160, The network device determines, through negotiation with the terminal, to enable an information encoding feedback mechanism based on an AI encoder. Subsequently, the network device may send an AI encoder to the terminal, and the terminal may perform encoding by using the AI encoder at a negotiated moment).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Cheng with Information encoding control method and related apparatus of Wang, because doing so would make Cheng more efficient in quickly and efficiently selecting an appropriate information encoding solution, avoiding distortion after to-be-transmitted information is encoded and decoded, avoiding impact on performance of a communication system, and improving transmission quality (Wang, see paragraph 0005).
Regarding claim 3, Cheng in view of Wang teaches further comprising:
decompressing the first AI/ML model based on the negotiated AI/ML encoding method (Cheng, see paragraph 0061, the AI/ML model data may be compressed for storage and/or transfer, for example, by using standard compression methods provided in ISO-IEC 15938-17 or any other possible compression methods).
Regarding claim 4, Cheng in view of Wang teaches wherein the decompressing of the first AI/ML model includes decompressing or decoding a set of weights included in the first AI/ML model utilizing at least one of a Binary Weight Network (BNN), a Ternary Weight Network (TNN), a B-bit compression, a K-means Clustering, a model sparsification, a Huffman Coding, and a Golomb Coding (Cheng, see paragraph 0061, the AI/ML model data may be compressed for storage and/or transfer, for example, by using standard compression methods provided in ISO-IEC 15938-17 or any other possible compression methods).
Regarding claim 5, Cheng in view of Wang teaches, further comprising:
updating the first AI/ML model when a second AI/ML model is received from a second base station, wherein the first base station communicates one or more characteristics of the first AI/ML model to the second base station during a handover procedure (Cheng, see paragraph 0066, In step S14, the gNB extracts the trained AI/ML model from the received RRC message. The trained AI/ML model may be used for various purposes. For example, the gNB may store the trained AI/ML model in a memory of its modem (modulator-demodulator) , and configure the modem for inference of corresponding use case, such as the CSI feedback enhancement, the beam management, the positioning accuracy enhancement, or the like).
Regarding claim 6, Cheng in view of Wang teaches further comprising:
updating the first AI/ML model when a performance metric value is greater or less than a threshold value (Cheng, see paragraph 0066, In step S14, the gNB extracts the trained AI/ML model from the received RRC message. The trained AI/ML model may be used for various purposes. For example, the gNB may store the trained AI/ML model in a memory of its modem (modulator-demodulator) , and configure the modem for inference of corresponding use case, such as the CSI feedback enhancement, the beam management, the positioning accuracy enhancement, or the like).
Regarding claim 7, Cheng in view of Wang teaches wherein the performance metric value is a Block Error Rate (BLER) value, an Uplink Packet Throughput (UPT) value, a predicted accuracy value, a below user-specific threshold value, or a below application-specific threshold value (Cheng, see paragraph 0066, In step S14, the gNB extracts the trained AI/ML model from the received RRC message. The trained AI/ML model may be used for various purposes. For example, the gNB may store the trained AI/ML model in a memory of its modem (modulator-demodulator) , and configure the modem for inference of corresponding use case, such as the CSI feedback enhancement, the beam management, the positioning accuracy enhancement, or the like).
Regarding claim 8, Cheng in view of Wang teaches further comprising:
updating the first AI/ML model in response to a scenario change, a configuration change, a site change, or a performance metric change (Cheng, see paragraph 0066, In step S14, the gNB extracts the trained AI/ML model from the received RRC message. The trained AI/ML model may be used for various purposes. For example, the gNB may store the trained AI/ML model in a memory of its modem (modulator-demodulator) , and configure the modem for inference of corresponding use case, such as the CSI feedback enhancement, the beam management, the positioning accuracy enhancement, or the like).
Claims 9-16 are rejected under 35 U.S.C. 103 as being unpatentable over Cheng (WO 2024/040476) in view of Zheng (US PGPub 2024/0314591).
Regarding claim 9, Cheng teaches a User Equipment (UE) (Cheng, see abstract, Rrc procedure design for wireless ai/ml), comprising:
a radio resource control (RRC) connection handling circuit that establishes an RRC connection with a first base station (Cheng, see paragraphs 0031 and 0064, The UE 102 and UE 104 utilize connections (or channels) (shown as connection 108 and connection 110, respectively) with the RAN 106, each of which comprises a physical communications interface. the gNB may transfer the transparent container including the AI/ML model to the UE via a downlink RRC message);
a memory circuit that stores the first AI/ML model, wherein the first AI/ML model belongs to a set of AI/ML models comprising: Convolution Neural Network (CNN), a Recurrent Neural Network (RNN), a Long Short Term Memory (LSTM), and a Transformer (Cheng, see paragraphs 0052 and 0068, a gNB may download an AI/ML model from its model server which is possibly built or rent by a gNB vendor. The downloading of the AI/ML may be initiated by a request from the gNB (S20). A typical implementation of AI/ML is neural network (NN) , such as Conventional Neural Network (CNN) , Recurrent/Recursive neural network (RNN) , Generative Adversarial Network (GAN) , or the like); and
a processor circuit that executes the first AI/ML model (Cheng, see paragraph 0072, the UE may store the trained AI/ML model in a memory of its modem (modulator-demodulator), and configure the modem for inference of corresponding use case).
Cheng teaches the above yet fails to teach a receiver circuit that receives a first Artificial Intelligence/Machine Learning (AI/ML) model that is broadcasted or groupcasted to multiple UEs by the first base station via one or more downlink RRC messages.
Then Zheng teaches a receiver circuit that receives a first Artificial Intelligence/Machine Learning (AI/ML) model that is broadcasted or groupcasted to multiple UEs by the first base station via one or more downlink RRC messages (Zheng, see paragraphs 0136-0138 and 0140, gNB may transmit the AI model information in a system information message. The system information message may be transmitted by a SIB (system information block) in a broadcast channel. The system information message may be transmitted to UE in a RRC message when the UE is in RRC connected state. Multiple AI models may be included in system information message).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Cheng with Model control and management for wireless communications of Zheng, because doing so would make Cheng more efficient in controlling and/or managing models applied to wireless communication systems, where the models can relate to Artificial Intelligence (AI) and/or Machine Learning (ML) (Zheng, see paragraph 0005).
Regarding claim 10, Cheng in view of Zheng teaches wherein the UE negotiates an AI/ML model encoding method, wherein the negotiation is performed between the UE and the first base station (Cheng, see paragraph 0061, the AI/ML model data may be compressed for storage and/or transfer, for example, by using standard compression methods provided in ISO-IEC 15938-17 or any other possible compression methods).
Regarding claim 11, Cheng in view of Zheng teaches wherein the UE decompresses the first AI/ML model before executing the first AI/ML model (Cheng, see paragraph 0061, the AI/ML model data may be compressed for storage and/or transfer, for example, by using standard compression methods provided in ISO-IEC 15938-17 or any other possible compression methods).
Regarding claim 12, Cheng in view of Zheng teaches wherein the decompressing of the first AI/ML model includes decompressing or decoding a set of weights included in the first AI/ML model utilizing at least one of a Binary Weight Network (BNN), a Ternary Weight Network (TNN), a B-bit compression, a K-means Clustering, a model sparsification, a Huffman Coding, and a Golomb Coding (Cheng, see paragraph 0061, the AI/ML model data may be compressed for storage and/or transfer, for example, by using standard compression methods provided in ISO-IEC 15938-17 or any other possible compression methods).
Regarding claim 13, Cheng in view of Zheng teaches wherein the UE receives a second AI/ML model from a second base station, wherein the UE executes the second AI/ML model in response to receiving the second AI/ML model from the second base station, and wherein the first base station communicates one or more characteristics of the first AI/ML model to the second base station during a handover procedure (Cheng, see paragraph 0066, In step S14, the gNB extracts the trained AI/ML model from the received RRC message. The trained AI/ML model may be used for various purposes. For example, the gNB may store the trained AI/ML model in a memory of its modem (modulator-demodulator) , and configure the modem for inference of corresponding use case, such as the CSI feedback enhancement, the beam management, the positioning accuracy enhancement, or the like).
Regarding claim 14, Cheng in view of Zheng teaches wherein the UE updates the first AI/ML model when a performance metric value is greater or less than a threshold value (Cheng, see paragraph 0066, In step S14, the gNB extracts the trained AI/ML model from the received RRC message. The trained AI/ML model may be used for various purposes. For example, the gNB may store the trained AI/ML model in a memory of its modem (modulator-demodulator) , and configure the modem for inference of corresponding use case, such as the CSI feedback enhancement, the beam management, the positioning accuracy enhancement, or the like).
Regarding claim 15, Cheng in view of Zheng teaches wherein the performance metric value is a Block Error Rate (BLER) value, an Uplink Packet Throughput (UPT) value, a predicted accuracy value, a below user-specific threshold value, or a below application-specific threshold value (Cheng, see paragraph 0066, In step S14, the gNB extracts the trained AI/ML model from the received RRC message. The trained AI/ML model may be used for various purposes. For example, the gNB may store the trained AI/ML model in a memory of its modem (modulator-demodulator) , and configure the modem for inference of corresponding use case, such as the CSI feedback enhancement, the beam management, the positioning accuracy enhancement, or the like).
Regarding claim 16, Cheng in view of Zheng teaches wherein the UE updates the first AI/ML model in response to a scenario change, a configuration change, a site change, or a performance metric change (Cheng, see paragraph 0066, In step S14, the gNB extracts the trained AI/ML model from the received RRC message. The trained AI/ML model may be used for various purposes. For example, the gNB may store the trained AI/ML model in a memory of its modem (modulator-demodulator) , and configure the modem for inference of corresponding use case, such as the CSI feedback enhancement, the beam management, the positioning accuracy enhancement, or the like).
Claims 17-20 are rejected under 35 U.S.C. 103 as being unpatentable over Cheng (WO 2024/040476) in view of Pantelidou (US PGPub 2023/0300686).
Regarding claim 17, Cheng teaches a method for improving the efficiency of wireless communications in fifth-generation / sixth-generation (5G/6G) networks by using artificial intelligence and machine learning (AI/ML) models (Cheng, see abstract, Rrc procedure design for wireless ai/ml), the method comprising:
storing a first AI/ML model in a first base station, wherein the first AI/ML model belongs to a set of AI/ML models comprising: a Convolution Neural Network (CNN), a Recurrent Neural Network (RNN), a Long Short Term Memory (LSTM), and a Transformer (Cheng, see paragraphs 0052 and 0068, a gNB may download an AI/ML model from its model server which is possibly built or rent by a gNB vendor. The downloading of the AI/ML may be initiated by a request from the gNB (S20). A typical implementation of AI/ML is neural network (NN) , such as Conventional Neural Network (CNN) , Recurrent/Recursive neural network (RNN) , Generative Adversarial Network (GAN) , or the like);
configuring the first AI/ML model in the first base station (Cheng, see paragraph 0071, Then in step S22, the gNB may train the AI/ML model with its local data by using various methods);
delivering the first AI/ML model to a user equipment device (UE) via a first set of downlink Radio Resource Control (RRC) messages (Cheng, see paragraph 0071, After performing the training, in step S23, the gNB may configure the trained model to the UE via a downlink RRC message).
Cheng teaches the above yet fails to teach transferring the first AI/ML model from the first base station to a second base station during a handover operation from the first base station to the second base station.
Then Pantelidou teaches transferring the first AI/ML model from the first base station to a second base station during a handover operation from the first base station to the second base station (Pantelidou, see paragraphs 0191 and 0214, This decision/wish to receive at least one ML model may be indicated to the source network apparatus 802 as part of a Handover Request acknowledgement 8005. For example, the target network apparatus may introduce a flag value (nw-based transfer flag) in the Handover Request Acknowledgment with which it informs the source whether the ML models should be transferred over the Xn interface (nw-based transfer=1). the source network apparatus 902 may send the ML Model Set and/or the corresponding MDT Configurations used to train the Models in the ML Model Set to the target network apparatus 903. The ML model set may be the ML model(s) identified in the acknowledgment message. The ML model Set and/or the corresponding MDT configurations may be sent through the Xn interface).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to modify Cheng with communication system for machine learning metadata of Pantelidou, because doing so would make Cheng more efficient in introducing signalling for handover operations when a terminal has ML models available in Training, Execution or Idle. This means that network resources may be used more efficiently as only desired ML models are kept when handover is completed (Pantelidou, see paragraph 0217).
Regarding claim 18, Cheng in view of Pantelidou teaches wherein the first base station communicates a second AI/ML model to the UE via a second set of downlink RRC messages (Cheng, see paragraph 0071, in step S23, the gNB may configure the trained model to the UE via a downlink RRC message).
Regarding claim 19, Cheng in view of Pantelidou teaches wherein during the handover operation, the first base station provides the second base station with an indicator indicating a set of capabilities of the first AI/ML model, and identification information indicating at least one of a UE type, a UE vendor, a UE modem type, and a UE chipset (Pantelidou, see paragraphs 0157-0167, the content of the metadata information).
Regarding claim 20, Cheng in view of Pantelidou teaches wherein during the handover operation, the first base station communicates an AI/ML indicator that indicates if the first AI/ML model is to be activated (Pantelidou, see paragraph 0135, The following discusses potential signalling that may be performed between a source network apparatus, a target network apparatus and/or a terminal being handed over between the source network apparatus and the target network apparatus when the terminal has at least one ML model available. The at least one ML model available may be in training, in execution, or in an idle mode).
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to CHONG G KIM whose telephone number is (571)270-0619. The examiner can normally be reached Mon-Fri @ 9am - 5pm.
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, Nicholas R. Taylor can be reached at 571-272-3889. 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.
/CHONG G KIM/Examiner, Art Unit 2443
/NICHOLAS R TAYLOR/Supervisory Patent Examiner, Art Unit 2443
Prosecution Insights