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 October 21, 2025 has been entered. Claims 1-2, 4-5, 8-9, 12, 14-15, 24, 32, 34-36, 38-40, 50, 54-55, and 57 are pending in the application. Applicant has submitted amendments to the claims along with other remarks. Applicant’s amendments to claim 36 overcome the objection. Claims 1-2, 4-5, 8-9, 12, 14-15, 24, 32, 34-36, 38-40, 50, 54-55, and 57 are still rejected by prior art references, refer to the following rejection for details.
Response to Arguments
Applicant’s arguments and amendments, see pp. 9-13 of the response, filed October 21, 2025, with respect to the rejection(s) of claim(s) 1-2, 4-5, 8-9, 12, 14-15, 24, 32, 34-36, 38-40, 50, 54-55, and 57 under §§ 102, 103 have been fully considered but are not persuasive.
At the outset, independent claim 32 was not amended and an argument was not provided to clarify the reasons for patentability. Independent claim 32 is rejected for the reasons previously provided and those provided herein.
Regarding amended claim 1, Applicant attacks the references individually, which is insufficient to overcome a prima facie case of obviousness. MPEP § 2145(IV). For example, Applicant states that “The BS uses this information for instantaneous downlink transmission decisions (like precoding), not for a systematic, high-level configuration action that requires initiation across the network.” Remarks at 11. This statement only addresses the teachings of Mashhadi instead of the combination of Mashhadi and Wu.
Further, Wu teaches: wherein using the received state representation to generate a configuration action for the RAN operation comprises using a Machine Learning, ML, process to generate the configuration action as a function of the state of the second node ((1) State space: A state s ∈ S is characterized by a 6-tuple, i.e., s = CQIreported, RIreported, RIprev, (3) MCSprev 1 , MCSprev 2 , HARQprev) (4) where CQIreported and RIreported are the most recent channel quality indicator (CQI) and rank indicator (RI) reported by the CSI feedback, respectively).).
Wu goes on to describe the action space in the next paragraph (e.g., the claimed configuration action): (2) Action space: An action a ∈ A is a 3-tuple with a = (Nlayers, MCS1, MCS2) representing the Nlayers, MCS1, and MCS2 to be assigned to the upcoming transmission. It should also be noted that TBS was directly determined by MCS1 and MCS2 [10]).
Applicant has emphasized the claimed elements “generated using an ML process.” The teachings of Mashhadi or Wu fall within an ML process when Mashhadi relates to “deep learning” for MIMO and Wu relates to “Q-Learning based Link Adaptation in 5G.”
For these reasons, the rejection is maintained.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
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.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
Claims 1, 2, 4, 5, 8, 9, 12, 24, 32, 34-36, 38-39, 50, 54, 55, and 57 are rejected under 35 U.S.C. 103 as being unpatentable over Non-patent Literature entitled, “Deep Learning for Massive MIMO Channel State Acquisition and Feedback” (hereinafter “Mashhadi”) in view of Non-patent Literature entitled, “Q-Learning based Link Adaptation in 5G” (hereinafter “Wu”).
Regarding claim 1, Mashhadi teaches: A computer implemented method for managing a Radio Access Network, RAN, operation performed by a first node (BS) in a communication network that comprises a Radio Access Network, the method, performed by the first node, comprising: receiving a representation of a state of a second node with respect to the RAN operation wherein the state of the second node (UE) comprises a compressed representation of parameter values that describe at least one of a physical state, a radio environment or a physical environment experienced by the second node or experienced by at least one node that is connected to the communication network via the second node using the received state representation to generate a configuration action for the RAN operation (p. 6/14 col. 2) In 40, 41, a DL-based CSI matrix compression technique, called DeepCMC, is proposed, which employs entropy coding to further compress the quantizer outputs. Figure 2 provides the endto-end block diagram for a downlink digital CSI feedback scheme based on DeepCMC40, 41. In this fgure, Hd and H ¬d denote the downlink CSI matrix at the UE and its estimate at the BS, respectively, and the two model input matrices represent R(Hd) and I(Hd). The UE applies a CNN-based feature encoder on Hd to obtain its low-dimensional representation, which is subsequently quantized and compressed using contextadaptive binary arithmetic coding (CABAC) 42. The resulting bit stream passes through channel coding and digital modulation. The modulation output is then mapped over OFDM subcarriers and transmitted back to the BS over the uplink channel.); and initiating configuration of the RAN operation in accordance with the generated configuration action (p. 1/14 col. 1) which can be used to shape the transmitted signals in a specific direction or to null interference. This yields a beamforming gain that translates into increased energy efficiency, reduced interference, or improved coverage.).
Mashhadi does not teach: wherein using the received state representation to generate a configuration action for the RAN operation comprises using a Machine Learning, ML, process to generate the configuration action as a function of the state of the second node.
However, in the same field of endeavor, Wu teaches: wherein using the received state representation to generate a configuration action for the RAN operation comprises using a Machine Learning, ML, process to generate the configuration action as a function of the state of the second node (1) State space: A state s ∈ S is characterized by a 6-tuple, i.e., s = CQIreported, RIreported, RIprev, (3) MCSprev 1 , MCSprev 2 , HARQprev) (4) where CQIreported and RIreported are the most recent channel quality indicator (CQI) and rank indicator (RI) reported by the CSI feedback, respectively).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Mashhadi to include the feature of generation of the configuration action as a function of the state of the second node and a combination of Mashhadi with Wu renders the claim prima facie obvious within the described scope of the prior art and any indicated differences within the level of one of ordinary skill in the art (e.g., telecommunications engineer) according to a combination of known prior art elements with known methods to yield predictable results. MPEP 2143(I)(A) (e.g., generating the configuration action as a function of the state of the second node).
Regarding claim 2, Mashhadi does not teach: obtaining a measure of success of the RAN operation configured in accordance with the generated configuration action and updating, based on the obtained success measure, how the received state representation is used to generate a configuration action for the RAN operation.
However, in the same field of endeavor, Wu teaches: obtaining a measure of success of the RAN operation configured in accordance with the generated configuration action and updating, based on the obtained success measure, how the received state representation is used to generate a configuration action for the RAN operation (p. 3/6, col. 1 - As shown in Fig. 2, the decision and feedback processes in the network are a natural fit for the reinforcement learning framework. In this paper we considered the Qlearning algorithm for LA (QLA). p. 3/6 col. 2 - Reward: The reward function is measured by how many bits are successfully delivered to the UE.).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Mashhadi to include the feature of reward based feedback and a combination of Mashhadi with Wu renders the claim prima facie obvious within the described scope of the prior art and any indicated differences within the level of one of ordinary skill in the art (e.g., telecommunications engineer) according to a combination of known prior art elements with known methods to yield predictable results. MPEP 2143(I)(A) (e.g., providing feedback to indicate performance of the generated configuration).
Regarding claim 5, Mashhadi does not teach: wherein using an ML process to generate the configuration action as a function of the state of the second node comprises at least one of: inputting a representation of the state of the second node to an ML model trained for use in generating a configuration action ;executing a Reinforcement Learning, RL, process by: using an ML model to predict a success measure for each of a plurality of possible configuration actions; using a selection function to select the configuration action based on the predicted success measures and an exploration component; and following execution of the RAN operation configured according to the configuration action, updating the ML model for predicting success measures; executing a Reinforcement Learning, RL, process by: using an ML model to predict the probability of executing each of a plurality of possible configuration actions; using a selection function to select the configuration action based on the predicted probability for each action; and following execution of the RAN operation configured according to the configuration action, updating the ML model generating the probability for each possible action based on an obtained measure of success of the RAN operation.
However, in the same field of endeavor, Wu teaches: wherein using an ML process to generate the configuration action as a function of the state of the second node comprises at least one of: inputting a representation of the state of the second node to an ML model trained for use in generating a configuration action ((1) State space: A state s ∈ S is characterized by a 6-tuple, i.e., s = CQIreported, RIreported, RIprev, (3) MCSprev 1 , MCSprev 2 , HARQprev) (4)); executing a Reinforcement Learning, RL, process by: using an ML model to predict a success measure for each of a plurality of possible configuration actions ((4) Reward: The reward function is measured by how many bits are successfully delivered to the UE); using a selection function to select the configuration action (Action space: An action a ∈ A is a 3-tuple with a = (Nlayers, MCS1, MCS2) representing the Nlayers, MCS1, and MCS2 to be assigned to the upcoming transmission. It should also be noted that TBS was directly determined by MCS1 and MCS2 [10]) based on the predicted success measures and an exploration component ((3) Policy: A policy π(s) is a mapping from a state to an action, which is commonly used in a greedy manner. The greedy algorithm [6] consists of two steps. O); and following execution of the RAN operation configured according to the configuration action, updating the ML model for predicting success measures (The proposed QLA starts with the BS choosing an action, i.e., deciding (Nlayers, MCS1, MCS2), according to π(s). Then, the UE feeds back HARQ information and CSI reports to the BS, forming the new state s. The BS computes the reward r according to HARQ and TBS. Next, the BS finds the new state a using the policy π(s), i.e., a = π(s), and updates the state-action function according to (7).); executing a Reinforcement Learning, RL, process by: using an ML model to predict the probability of executing each of a plurality of possible configuration actions; using a selection function to select the configuration action based on the predicted probability for each action; and following execution of the RAN operation configured according to the configuration action, updating the ML model generating the probability for each possible action based on an obtained measure of success of the RAN operation (Action space: An action a ∈ A is a 3-tuple with a = (Nlayers, MCS1, MCS2) representing the Nlayers, MCS1, and MCS2 to be assigned to the upcoming transmission. It should also be noted that TBS was directly determined by MCS1 and MCS2 [10].)).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Mashhadi to include the feature of reinforcement learning applied to CSI data and a combination of Mashhadi with Wu renders the claim prima facie obvious within the described scope of the prior art and any indicated differences within the level of one of ordinary skill in the art (e.g., telecommunications engineer) according to a combination of known prior art elements with known methods to yield predictable results. MPEP 2143(I)(A) (e.g., feature of reinforcement learning applied to CSI data).
Regarding claim 8, Mashhadi does not teach: wherein executing an RL process for generating a configuration action comprises: inputting a representation of the state of the second node to the ML model for predicting a success measure for possible configuration actions; selecting a configuration action based on the predicted success measures for possible actions; obtaining a measure of success of the RAN operation configured in accordance with the generated configuration action; and updating the ML model on the basis of the obtained measure of success.
However, in the same field of endeavor, Wu teaches: wherein executing an RL process for generating a configuration action comprises: inputting a representation of the state of the second node to the ML model for predicting a success measure for possible configuration actions (p. 3/6 (1) State space: A state s ∈ S is characterized by a 6-tuple, i.e.,); selecting a configuration action based on the predicted success measures for possible actions (p. 3/6 (2) Action space: An action a ∈ A is a 3-tuple with a =(Nlayers,MCS1,MCS2) representing the Nlayers, MCS1,); obtaining a measure of success of the RAN operation configured in accordance with the generated configuration action (p. 3/6 (4) Reward: The reward function is measured by how); and updating the ML model on the basis of the obtained measure of success (Fig. 2).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Mashhadi to include the feature of executing an RL process for generating a configuration action and a combination of Mashhadi with Wu renders the claim prima facie obvious within the described scope of the prior art and any indicated differences within the level of one of ordinary skill in the art (e.g., telecommunications engineer) according to a combination of known prior art elements with known methods to yield predictable results. MPEP 2143(I)(A) (e.g., executing an RL process for generating a configuration action).
Regarding claim 9, Mashhadi does not teach: obtaining an ML model for use in generating a configuration action; training an ML model for use in generating a configuration action.
However, in the same field of endeavor, Wu teaches: obtaining an ML model for use in generating a configuration action (Fig. 4); training an ML model for use in generating a configuration action (p. 4/6 During the training of the NN, a regular choice of objective function is root mean squared error (RMSE).).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Mashhadi to include the feature of training an ML model for use in generating a configuration action and a combination of Mashhadi with Wu renders the claim prima facie obvious within the described scope of the prior art and any indicated differences within the level of one of ordinary skill in the art (e.g., telecommunications engineer) according to a combination of known prior art elements with known methods to yield predictable results. MPEP 2143(I)(A) (e.g., training an ML model for use in generating a configuration action).
Regarding claim 12, Mashhadi teaches: wherein the received representation of a state of the second node comprises at least one of :a state identifier for the state of the second node; the compressed representation of parameter values that comprises the state (The UE applies a CNN-based feature encoder on Hd to obtain its low-dimensional representation, which is subsequently quantized and compressed using contextadaptive binary arithmetic coding (CABAC) 42. The resulting bit stream passes through channel coding and digital modulation.); or an indication of difference from a previous state of the second node; and wherein a state identifier for a state of a second node comprises an identifier that is unique to a method used to generate the compressed representation of parameter values that comprises the state.
Regarding claim 24, Mashhadi does not teach: obtaining a measure of usefulness of the received representation of a state of the second node for configuration of the RAN operation ; and updating at least one of :a process for using the received state representation to generate a configuration action for the RAN operation; or a configuration for receipt of the state representation on the basis of the obtained measure of usefulness.
However, in the same field of endeavor, Wu teaches: obtaining a measure of usefulness of the received representation of a state of the second node for configuration of the RAN operation (p. 3/6 (4) Reward: The reward function is measured by how many bits are successfully delivered to the UE. The BS acquires the reward via the HARQ information fed back from the UE, ); and updating at least one of :a process for using the received state representation to generate a configuration action for the RAN operation; or a configuration for receipt of the state representation on the basis of the obtained measure of usefulness ((2) Action space: An action a ∈ A is a 3-tuple with a =(Nlayers,MCS1,MCS2) representing the Nlayers, MCS1, and MCS2 to be assigned to the upcoming transmission.).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Mashhadi to include the feature of usefulness and a combination of Mashhadi with Wu renders the claim prima facie obvious within the described scope of the prior art and any indicated differences within the level of one of ordinary skill in the art (e.g., telecommunications engineer) according to a combination of known prior art elements with known methods to yield predictable results. MPEP 2143(I)(A) (e.g., providing an indication of usefulness (e.g., reward)).
Regarding claim 32, Mashhadi teaches: A computer implemented method for facilitating a Radio Access Network, RAN, operation performed by a first node (BS) in a communication network that comprises a Radio Access Network, the method, performed by a second node (UE), comprising: generating a state of the second node with respect to the RAN operation ((p. 6/14 col. 2) In 40, 41, a DL-based CSI matrix compression technique, called DeepCMC, is proposed, which employs entropy coding to further compress the quantizer outputs. Figure 2 provides the endto-end block diagram for a downlink digital CSI feedback scheme based on DeepCMC40, 41. In this figure, Hd and H ¬d denote the downlink CSI matrix at the UE and its estimate at the BS, respectively, and the two model input matrices represent R(Hd) and I(Hd). The UE applies a CNN-based feature encoder on Hd to obtain its low-dimensional representation, which is subsequently quantized and compressed using contextadaptive binary arithmetic coding (CABAC) 42. The resulting bit stream passes through channel coding and digital modulation. The modulation output is then mapped over OFDM subcarriers and transmitted back to the BS over the uplink channel.), wherein the state of the second node comprises a compressed representation of parameter values that describe at least one of a physical state, a radio environment or a physical environment experienced by the second node or experienced by at least one node that is connected to the communication network via the second node (The UE applies a CNN-based feature encoder on Hd to obtain its low-dimensional representation, which is subsequently quantized and compressed using contextadaptive binary arithmetic coding (CABAC) 42.); and transmitting a representation of the generated state to at least one of the first node or a node of the communication network that is operable to communicate with the first node (The modulation output is then mapped over OFDM subcarriers and transmitted back to the BS over the uplink channel).
Regarding claim 34, Mashhadi teaches: wherein generating a state of the second node with respect to the RAN operation comprises: assembling parameter values for inclusion in the state (p. 6/14 col. 2 - denote the downlink CSI matrix at the UE); and generating a compressed representation of the parameter values using a Machine Learning, ML, process (Figure 2).
Regarding claim 35, Mashhadi teaches: wherein generating a compressed representation of the parameter values using an ML process comprises: reducing a dimensionality of the assembled parameter values using a trained ML model (p. 5/14 col. 2 - Recently, dimensionality reducing autoencoders have shown significant success for lossy compression of such sources with a data-driven approach).
Regarding claim 36, Mashhadi teaches: wherein the trained ML model comprises at least one of: an encoder part of an Autoencoder; a model trained to execute a Principal Component Analysis process (Figure 2 – Encoder).
Regarding claim 38, Mashhadi teaches: preparing a representation of the generated state for transmission (p. 5/14 col. 2 - These autoencoder architectures can be trained to learn a lower dimensional representation of the original CSI matrix to be transmitted over the feedback channel with a reduced overhead.).
Regarding claim 39, Mashhadi teaches: wherein preparing a representation of the generated state for transmission comprises at least one of: mapping the generated state to a state identifier for transmission; assembling the compressed representation for transmission (Figure 2 – decoder); computing a difference between the generated state and a previous state of the second node (p. 7/14 The training cost for DeepCMC is a weighted sum of the mean square error (MSE) of the CSI reconstruction and the quantizer’s output Entropy).
Regarding claim 50, Mashhadi teaches: obtaining a measure of usefulness of the transmitted representation of a state of the second node for configuration of the RAN operation (p. 7/14 - However, if the feedback channel capacity is smaller than the resulting bit rate, the feedback channel will fail to deliver the CSI. To avoid this, a network trained to work at a lower bit rate (trained with smaller -) should be used. Different - values will provide networks that work on different points on a rate distortion curve. The UE will store different networks, and use the proper one depending on the uplink channel state and the capacity achievable for CSI feedback.); and updating at least one of :a process for generating the state of the second node (Different - values will provide networks that work on different points on a ratedistortion curve); or a parameter included with the transmitted representation of the generated state on the basis of the obtained measure of usefulness.
Regarding claim 54, Mashhadi teaches: A non-transitory computer readable medium comprising having computer readable code embodied therein, the computer readable code being configured such that, on execution by a suitable computer or processor, the computer or processor is caused to, for managing a Radio Access Network, RAN, operation performed by a first node (BS) in a communication network that comprises a RAN: receive a representation of a state of a second node (UE) with respect to the RAN operation, wherein the state of the second node comprises a compressed representation of parameter values that describe at least one of a physical state, a radio environment or a physical environment experienced by the second node or experienced by at least one node that is connected to the communication network via the second node; use the received state representation to generate a configuration action for the RAN operation (p. 6/14 col. 2) In 40, 41, a DL-based CSI matrix compression technique, called DeepCMC, is proposed, which employs entropy coding to further compress the quantizer outputs. Figure 2 provides the endto-end block diagram for a downlink digital CSI feedback scheme based on DeepCMC40, 41. In this fgure, Hd and H ¬d denote the downlink CSI matrix at the UE and its estimate at the BS, respectively, and the two model input matrices represent R(Hd) and I(Hd). The UE applies a CNN-based feature encoder on Hd to obtain its low-dimensional representation, which is subsequently quantized and compressed using contextadaptive binary arithmetic coding (CABAC) 42. The resulting bit stream passes through channel coding and digital modulation. The modulation output is then mapped over OFDM subcarriers and transmitted back to the BS over the uplink channel.); and initiate configuration of the RAN operation in accordance with the generated configuration action (p. 1/14 col. 1) which can be used to shape the transmitted signals in a specific direction or to null interference. This yields a beamforming gain that translates into increased energy efficiency, reduced interference, or improved coverage.).
Mashhadi does not teach: wherein using the received state representation to generate a configuration action for the RAN operation comprises using a Machine Learning, ML, process to generate the configuration action as a function of the state of the second node.
However, in the same field of endeavor, Wu teaches: wherein using the received state representation to generate a configuration action for the RAN operation comprises using a Machine Learning, ML, process to generate the configuration action as a function of the state of the second node (1) State space: A state s ∈ S is characterized by a 6-tuple, i.e., s = CQIreported, RIreported, RIprev, (3) MCSprev 1 , MCSprev 2 , HARQprev) (4) where CQIreported and RIreported are the most recent channel quality indicator (CQI) and rank indicator (RI) reported by the CSI feedback, respectively).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify Mashhadi to include the feature of generation of the configuration action as a function of the state of the second node and a combination of Mashhadi with Wu renders the claim prima facie obvious within the described scope of the prior art and any indicated differences within the level of one of ordinary skill in the art (e.g., telecommunications engineer) according to a combination of known prior art elements with known methods to yield predictable results. MPEP 2143(I)(A) (e.g., generating the configuration action as a function of the state of the second node).
Regarding claim 55, Mashhadi teaches: A first node (BS) in a communication network comprising a Radio Access Network, RAN, the first node for managing a RAN operation performed by the first node and comprising processing circuitry and memory, wherein the memory may contain instructions executable by the processing circuitry, and wherein the processing circuitry is configured to: receive a representation of a state of a second node (UE) with respect to the RAN operation, wherein the state of the second node comprises a compressed representation of parameter values that describe at least one of a physical state, a radio environment or a physical environment experienced by the second node or by at least one node that is connected to the communication network via the second node(p. 6/14 col. 2) In 40, 41, a DL-based CSI matrix compression technique, called DeepCMC, is proposed, which employs entropy coding to further compress the quantizer outputs. Figure 2 provides the endto-end block diagram for a downlink digital CSI feedback scheme based on DeepCMC40, 41. In this fgure, Hd and H ¬d denote the downlink CSI matrix at the UE and its estimate at the BS, respectively, and the two model input matrices represent R(Hd) and I(Hd). The UE applies a CNN-based feature encoder on Hd to obtain its low-dimensional representation, which is subsequently quantized and compressed using contextadaptive binary arithmetic coding (CABAC) 42. The resulting bit stream passes through channel coding and digital modulation. The modulation output is then mapped over OFDM subcarriers and transmitted back to the BS over the uplink channel.); use the received state representation to generate a configuration action for the RAN operation; and initiate configuration of the RAN operation in accordance with the generated configuration action (p. 1/14 col. 1) which can be used to shape the transmitted signals in a specific direction or to null interference. This yields a beamforming gain that translates into increased energy efficiency, reduced interference, or improved coverage.).
Regarding claim 57, Mashhadi teaches: A second node (UE) in a communication network comprising a Radio Access Network, RAN, the second node for facilitating a RAN operation performed by a first node (BS) in the communication network and comprising processing circuitry and memory, wherein the memory may contain instructions executable by the processing circuitry, and wherein the processing circuitry is configured to: generate a state of the second node with respect to the RAN operation ((p. 6/14 col. 2) In 40, 41, a DL-based CSI matrix compression technique, called DeepCMC, is proposed, which employs entropy coding to further compress the quantizer outputs. Figure 2 provides the endto-end block diagram for a downlink digital CSI feedback scheme based on DeepCMC40, 41. In this figure, Hd and H ¬d denote the downlink CSI matrix at the UE and its estimate at the BS, respectively, and the two model input matrices represent R(Hd) and I(Hd). The UE applies a CNN-based feature encoder on Hd to obtain its low-dimensional representation, which is subsequently quantized and compressed using contextadaptive binary arithmetic coding (CABAC) 42. The resulting bit stream passes through channel coding and digital modulation. The modulation output is then mapped over OFDM subcarriers and transmitted back to the BS over the uplink channel.), wherein the state of the second node comprises a compressed representation of parameter values that describe at least one of a physical state, a radio environment or a physical environment experienced by the second node or by at least one node that is connected to the communication network via the second node (The UE applies a CNN-based feature encoder on Hd to obtain its low-dimensional representation, which is subsequently quantized and compressed using contextadaptive binary arithmetic coding (CABAC) 42.); and transmit a representation of the generated state to at least one of the first node or a node of the communication network that is operable to communicate with the first node (The modulation output is then mapped over OFDM subcarriers and transmitted back to the BS over the uplink channel).
Claims 14 and 40 are rejected under 35 U.S.C. 103 as being unpatentable over Mashhadi in view of Wu and further in view of U.S. Publication No. 2021/0314036 (hereinafter “Baknina”).
Regarding claim 14, the combination of Mashhadi and Wu does not teach: wherein obtaining an ML model for use in generating a configuration action comprises obtaining an ML model that corresponds to the received state identifier.
However, in the same field of endeavor, Baknina teaches: wherein obtaining an ML model for use in generating a configuration action comprises obtaining an ML model that corresponds to the received state identifier ([0050] Referring to FIG. 4, the experience can be gathered for 5 different channel states. For each different state, the system saves a spectral efficiency (SE) an RI, a CQI, and a state identifier. The SE is the number of bits successfully received at the UE on a certain channel).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify the combination of Mashhadi and Wu to include the feature of an identifier to indicate the state and a combination of Mashhadi and Wu with Baknina renders the claim prima facie obvious within the described scope of the prior art and any indicated differences within the level of one of ordinary skill in the art (e.g., telecommunications engineer) according to a combination of known prior art elements with known methods to yield predictable results. MPEP 2143(I)(A) (e.g., indicating the state with an identifier).
Regarding claim 40, the combination of Mashhadi and Wu does not teach: wherein a state identifier for a state of a second node comprises an identifier that is unique to a method used to generate the compressed representation of parameter values that comprises the state.
However, in the same field of endeavor, Baknina teaches: wherein a state identifier for a state of a second node comprises an identifier that is unique to a method used to generate the compressed representation of parameter values that comprises the state ([0050] Referring to FIG. 4, the experience can be gathered for 5 different channel states. For each different state, the system saves a spectral efficiency (SE) an RI, a CQI, and a state identifier. The SE is the number of bits successfully received at the UE on a certain channel).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify the combination of Mashhadi and Wu to include the feature of an identifier to indicate the state and a combination of Mashhadi and Wu with Baknina renders the claim prima facie obvious within the described scope of the prior art and any indicated differences within the level of one of ordinary skill in the art (e.g., telecommunications engineer) according to a combination of known prior art elements with known methods to yield predictable results. MPEP 2143(I)(A) (e.g., indicating the state with an identifier).
Claim 15 is rejected under 35 U.S.C. 103 as being unpatentable over Mashhadi in view of Wu and Baknina and further in view of WIPO Publication No. 2020/091543 (hereinafter “Park”)
Regarding claim 15, the combination of Mashhadi, Wu, and Baknina does not teach: if an ML model that corresponds to the received state identifier cannot be obtained, and if fewer than a threshold number of second nodes have reported the received state identifier: instructing the second node to use a legacy reporting procedure for the RAN operation; if an ML model that corresponds to the received state identifier cannot be obtained, and if at least a threshold number of second nodes have reported the received state identifier: training a new ML model for use in generating a configuration action from the received state identifier.
However, in the same field of endeavor, Park teaches: if an ML model that corresponds to the received state identifier cannot be obtained, and if fewer than a threshold number of second nodes have reported the received state identifier: instructing the second node to use a legacy reporting procedure for the RAN operation; if an ML model that corresponds to the received state identifier cannot be obtained, and if at least a threshold number of second nodes have reported the received state identifier: training a new ML model for use in generating a configuration action from the received state identifier ([1075-1078] That is, the terminal can receive a plurality of parameter values including fallback values for each of the codebook configuration parameters for configuring a codebook from the base station, and if the CSI calculated by the received parameters is greater than the resources allocated for CSI reporting, the terminal can recalculate the CSI by applying the fallback values and report the CSI on the allocated resources.).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to modify the combination of Mashhadi, Wu, and Baknina to include the feature of a fallback model and a combination of Mashhadi, Wu, and Baknina with Park renders the claim prima facie obvious within the described scope of the prior art and any indicated differences within the level of one of ordinary skill in the art (e.g., telecommunications engineer) according to a combination of known prior art elements with known methods to yield predictable results. MPEP 2143(I)(A) (e.g., providing a fallback model and configuration).
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
U.S. Publication No. 2023/0422054 (Jeon) Method for performing federated learning in wireless communication system, and apparatus therefor
U.S. Publication No. 2023/0261909 (Namgoong) Federated learning for classifiers and autoencoders for wireless communication
U.S. Patent No. 12,273,854 (Koteshwar) Apparatus and method for user equipment positioning and network node using the same
THIS ACTION IS MADE FINAL. 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 JUSTIN BARRY whose telephone number is (571)272-0201. The examiner can normally be reached 8:00am EST to 5:00pm EST.
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, Jinsong HU can be reached at (571) 272-3965. 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.
/JAB/ Examiner, Art Unit 2643
/SAN HTUN/ Primary Examiner, Art Unit 2643