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 .
Priority
1. Receipt is acknowledged of certified copies of papers required by 37 CFR 1.55.
Claim Interpretation
The following is a quotation of 35 U.S.C. 112(f):
(f) Element in Claim for a Combination. – An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
The following is a quotation of pre-AIA 35 U.S.C. 112, sixth paragraph:
An element in a claim for a combination may be expressed as a means or step for performing a specified function without the recital of structure, material, or acts in support thereof, and such claim shall be construed to cover the corresponding structure, material, or acts described in the specification and equivalents thereof.
2. The claims in this application are given their broadest reasonable interpretation using the plain meaning of the claim language in light of the specification as it would be understood by one of ordinary skill in the art. The broadest reasonable interpretation of a claim element (also commonly referred to as a claim limitation) is limited by the description in the specification when 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is invoked.
As explained in MPEP § 2181, subsection I, claim limitations that meet the following three-prong test will be interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph:
(A) the claim limitation uses the term “means” or “step” or a term used as a substitute for “means” that is a generic placeholder (also called a nonce term or a non-structural term having no specific structural meaning) for performing the claimed function;
(B) the term “means” or “step” or the generic placeholder is modified by functional language, typically, but not always linked by the transition word “for” (e.g., “means for”) or another linking word or phrase, such as “configured to” or “so that”; and
(C) the term “means” or “step” or the generic placeholder is not modified by sufficient structure, material, or acts for performing the claimed function.
Use of the word “means” (or “step”) in a claim with functional language creates a rebuttable presumption that the claim limitation is to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites sufficient structure, material, or acts to entirely perform the recited function.
Absence of the word “means” (or “step”) in a claim creates a rebuttable presumption that the claim limitation is not to be treated in accordance with 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph. The presumption that the claim limitation is not interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, is rebutted when the claim limitation recites function without reciting sufficient structure, material or acts to entirely perform the recited function.
Claim limitations in this application that use the word “means” (or “step”) are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action. Conversely, claim limitations in this application that do not use the word “means” (or “step”) are not being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, except as otherwise indicated in an Office action. This language is recited in the limitations of claim 30.
Claim Rejections - 35 USC § 102
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 the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale, or otherwise available to the public before the effective filing date of the claimed invention.
3. Claims 1, 8, 9 and 30 are rejected under 35 U.S.C. 102(a) (1) as being anticipated by MOHADES ZOHREH ET AL: "Deep Neural Network for Compressive Sensing and Application to Massive MIMO Channel Estimation", CIRCUITS, SYSTEMS AND SIGNAL PROCESSING, CAMBRIDGE, MS, US, vol. 40, no. 9, 9 March 2021 (2021-03-09), pages 4474-4489, XP037528052, ISSN: 0278-081X, DOI: 10.1007/S00034-021-01675-Z.
Regarding claims 1 and 30, Mohades discloses a processor-implemented method (Heading 3: Proposed methods. The method will be executed by a means. The means for executing the methods is a processor.), comprising:
generating a current sparsifying dictionary by processing a sensing matrix and a current channel observation for a digital communication channel using a posterior neural network in a first iteration of a machine learning model (Heading 2.3 Dictionary Learning. A learned dictionary can represent a given signal with a smaller representation error than a predefined dictionary. In this paper we use a block sparce dictionary learning to learn a dictionary which can offer satisfying offer a sparse representation of a massive MIMO channel. Abstract: in the first algorithm…a four-layer feed-forward neural network is applied.); and
generating a current sparse channel representation by processing the current sparsifying dictionary, the sensing matrix, and the current channel observation using a likelihood neural network in the first iteration (Heading 4 channel estimation in massive MIMO system with proposed algorithms. For this purpose, we utilize channel measurements H and consider equation 11. Abstract: in the second algorithm. Recurrent neural networks are employed to extract the sparsity structure.).
Regarding claim 8, Mohades discloses wherein the posterior neural network and the likelihood neural network are shared across each iteration of the machine learning model (Abstract: in the first algorithm…a four-layer feed-forward neural network is applied. In the second algorithm. Recurrent neural networks are employed to extract the sparsity structure. The dictionary and matrix are used in each algorithm, sharing the information.).
Regarding claim 9, Mohades discloses wherein each respective iteration of the machine learning model has a respective posterior neural network and a respective likelihood neural network (Abstract: in the first algorithm…a four-layer feed-forward neural network is applied. In the second algorithm. Recurrent neural networks are employed to extract the sparsity structure. Each step has a respective NN.).
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.
4. Claims 2-5 are rejected under 35 U.S.C. 103 as being unpatentable over MOHADES ZOHREH ET AL: "Deep Neural Network for Compressive Sensing and Application to Massive MIMO Channel Estimation", CIRCUITS, SYSTEMS AND SIGNAL PROCESSING, CAMBRIDGE, MS, US, vol. 40, no. 9, 9 March 2021 (2021-03-09), pages 4474-4489, XP037528052, ISSN: 0278-081X, DOI: 10.1007/S00034-021-01675-Z in view of Li et al (US 2023/0053588).
Regarding claim 2, Mohades discloses the method stated above. Mohades does not disclose wherein generating the current sparsifying dictionary using the posterior neural network comprises: generating a mean value and a variance value by processing the sensing matrix, the current channel observation, and a previous sparse channel representation generated in a previous iteration of the machine learning model, using the posterior neural network; and sampling a distribution having the mean value and the variance value.
Li discloses a neural network as stated in the abstract. Paragraph 0060 discloses the multi-resolution generator neural network then utilizes a first linear neural network layer to generate a mean value associated with the reduced resolution set. The multi-resolution generator also utilizes a second linear layer to generate a variance associated with the resolution feature set. Accordingly, the multi-resolution generator neural network generates a mean value and the variance value to represent the distribution of features.
It would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to utilize the teaching of Li into the method and system of Mohades. Using the mean and the variance to determine or generate the distribution to represent the output is well known in neural network operation. Utilizing well known steps allows for a simpler method or algorithm, reducing costs and complexity.
Regarding claim 3, the combination discloses wherein the distribution is a posterior distribution defined as stated in the claim wherein: the recited distribution includes the current sparsifying dictionary, the previous sparse channel representation generated in a previous iteration of the machine learning model, the current channel observation, and the sensing matrix (Mohades: Heading 2.3 Dictionary Learning. A learned dictionary can represent a given signal with a smaller representation error than a predefined dictionary. In this paper we use a block sparce dictionary learning to learn a dictionary which can offer satisfying offer a sparse representation of a massive MIMO channel. Abstract: in the first algorithm…a four-layer feed-forward neural network is applied. Heading 3.1: sparse matrix. Heading 4 channel estimation in massive MIMO system with proposed algorithms. For this purpose, we utilize channel measurements H and consider equation 11. Abstract: in the second algorithm. Recurrent neural networks are employed to extract the sparsity structure. The input and the previous input are reflected in the output.).
Regarding claim 4, Mohades discloses the method stated above. Mohades does not disclose wherein generating the current sparse channel representation using the likelihood neural network comprises: generating a mean value by processing the current sparsifying dictionary, the sensing matrix, and the current channel observation using the likelihood neural network; and generating a distribution having the mean value and a variance value.
Li discloses a neural network as stated in the abstract. Paragraph 0060 discloses the multi-resolution generator neural network then utilizes a first linear neural network layer to generate a mean value associated with the reduced resolution set. The multi-resolution generator also utilizes a second linear layer to generate a variance associated with the resolution feature set. Accordingly, the multi-resolution generator neural network generates a mean value and the variance value to represent the distribution of features.
It would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to utilize the teaching of Li into the method and system of Mohades. Using the mean and the variance to determine or generate the distribution to represent the output is well known in neural network operation. Utilizing well known steps allows for a simpler method or algorithm, reducing costs and complexity.
Regarding claim 5, the combination discloses wherein the distribution is a likelihood model defined in the claim, wherein: the recited distribution includes the current sparse channel representation, a ground truth channel state, sparsifying dictionaries generated in one or more previous iterations in the machine learning model, the current channel observation, and the sensing matrix (Mohades: Heading 2.3 Dictionary Learning. A learned dictionary can represent a given signal with a smaller representation error than a predefined dictionary. In this paper we use a block sparce dictionary learning to learn a dictionary which can offer satisfying offer a sparse representation of a massive MIMO channel. Abstract: in the first algorithm…a four-layer feed-forward neural network is applied. Heading 3.1: sparse matrix. Heading 4 channel estimation in massive MIMO system with proposed algorithms. For this purpose, we utilize channel measurements H and consider equation 11. Abstract: in the second algorithm. Recurrent neural networks are employed to extract the sparsity structure. The input and the previous channel state are reflected in the output.).
5. Claims 6, 7, 21 and 25-28 are rejected under 35 U.S.C. 103 as being unpatentable over MOHADES ZOHREH ET AL: "Deep Neural Network for Compressive Sensing and Application to Massive MIMO Channel Estimation", CIRCUITS, SYSTEMS AND SIGNAL PROCESSING, CAMBRIDGE, MS, US, vol. 40, no. 9, 9 March 2021 (2021-03-09), pages 4474-4489, XP037528052, ISSN: 0278-081X, DOI: 10.1007/S00034-021-01675-Z in view of Ahuja et al (US 2020/0326667).
Regarding claim 6, Mohades discloses the method and system stated above. Mohades does not disclose generating an uncertainty measurement using the likelihood neural network; determining that the uncertainty measurement satisfies one or more defined criteria; and in response to determining that the uncertainty measurement satisfies the one or more defined criteria, initiating re-training of the posterior neural network and the likelihood neural network.
Ahuja discloses a neural network architecture as stated in the abstract. Paragraph 0079 discloses one or more processors are configured to output the sets of distribution data in accordance with the neural network architecture in which each one of the neural network outputs a respective one of the sets of distribution data and wherein one or more of the processors are configured to monitor the uncertainty estimate values over time and to initiate a re-training sequence for the plurality of neural networks when an epistemic uncertainty value associated with the one or more of the plurality of neural networks exceeds a threshold epistemic value.
It would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teaching of Ahuja into the method and system of Mohades. By generating uncertainty values, one can determine when the NN is not providing acceptable information and take appropriate actions. This will improve the function of the method and system.
Regarding claim 7, Mohades discloses the method and system stated above. Mohades does not disclose determining an entropy based on the likelihood neural network; determining that the entropy satisfies one or more defined criteria; and in response to determining that the entropy satisfies the one or more defined criteria: refraining from processing the sensing matrix and the current channel observation using a subsequent iteration of the machine learning model; generating a current channel estimation based on the current sparse channel representation; and outputting the current channel estimation.
Ahuja discloses a neural network architecture as stated in the abstract. Paragraph 0047 discloses uncertainty estimating has been a focus of researchers in the Bayesian deep learning field. Parameters are learned using variational training from which various predictive uncertainty measures can be extracted (e.g., entropies). Paragraph 0079 discloses one or more processors are configured to output the sets of distribution data in accordance with the neural network architecture in which each one of the neural network outputs a respective one of the sets of distribution data and wherein one or more of the processors are configured to monitor the uncertainty estimate values over time and to initiate a re-training sequence for the plurality of neural networks when an epistemic uncertainty value associated with the one or more of the plurality of neural networks exceeds a threshold epistemic value. When the uncertainty does not exceed the threshold and the training is completed, the data is output.
It would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to combine the teaching of Ahuja into the method and system of Mohades. By generating uncertainty values, one can determine when the NN is providing acceptable information and take appropriate actions, such as outputting that information. This will improve the function of the method and system by limiting the amount of time a system is trained.
Regarding claim 21, Mohades discloses a system for executing a processor-implemented method (Heading 3: Proposed methods. The method will be executed by a means. The means for executing the methods is a processor.), comprising:
generating a current sparsifying dictionary by processing a sensing matrix and a current channel observation for a digital communication channel using a posterior neural network in a first iteration of a machine learning model (Heading 2.3 Dictionary Learning. A learned dictionary can represent a given signal with a smaller representation error than a predefined dictionary. In this paper we use a block sparce dictionary learning to learn a dictionary which can offer satisfying offer a sparse representation of a massive MIMO channel. Abstract: in the first algorithm…a four-layer feed-forward neural network is applied.); and
generating a current sparse channel representation by processing the current sparsifying dictionary, the sensing matrix, and the current channel observation using a likelihood neural network in the first iteration (Heading 4 channel estimation in massive MIMO system with proposed algorithms. For this purpose, we utilize channel measurements H and consider equation 11. Abstract: in the second algorithm. Recurrent neural networks are employed to extract the sparsity structure.).
Mohades does not explicitly discloses a memory for storing and one or more processors to execute computer executable instructions.
Ahuja discloses a neural network architecture as stated in the abstract. Paragraph 0079 discloses one or more processors are configured to output the sets of distribution data in accordance with the neural network architecture in which each one of the neural network outputs a respective one of the sets of distribution data and wherein one or more of the processors are configured to monitor the uncertainty estimate values over time and to initiate a re-training sequence for the plurality of neural networks when an epistemic uncertainty value associated with the one or more of the plurality of neural networks exceeds a threshold epistemic value.
It would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to combine the memory and processors of Ahuja for executing instructions into the system of Mohades. By using processors and memories, the size of the circuit can be reduced, minimizing the cost and complexity of the system.
Regarding claim 25, the combination of Mohades and Ahuja discloses the method and system stated above. The combination does not disclose generating an uncertainty measurement using the likelihood neural network; determining that the uncertainty measurement satisfies one or more defined criteria; and in response to determining that the uncertainty measurement satisfies the one or more defined criteria, initiating re-training of the posterior neural network and the likelihood neural network.
Ahuja discloses a neural network architecture as stated in the abstract. Paragraph 0079 discloses one or more processors are configured to output the sets of distribution data in accordance with the neural network architecture in which each one of the neural network outputs a respective one of the sets of distribution data and wherein one or more of the processors are configured to monitor the uncertainty estimate values over time and to initiate a re-training sequence for the plurality of neural networks when an epistemic uncertainty value associated with the one or more of the plurality of neural networks exceeds a threshold epistemic value.
It would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to combine this teaching of Ahuja into the method and system of the combination of Mohades and Ahuja. By generating uncertainty values, one can determine when the NN is not providing acceptable information and take appropriate actions. This will improve the function of the method and system.
Regarding claim 26, the combination of Mohades and Ahuja discloses the method and system stated above. The combination does not disclose determining an entropy based on the likelihood neural network; determining that the entropy satisfies one or more defined criteria; and in response to determining that the entropy satisfies the one or more defined criteria: refraining from processing the sensing matrix and the current channel observation using a subsequent iteration of the machine learning model; generating a current channel estimation based on the current sparse channel representation; and outputting the current channel estimation.
Ahuja discloses a neural network architecture as stated in the abstract. Paragraph 0047 discloses uncertainty estimating has been a focus of researchers in the Bayesian deep learning field. Parameters are learned using variational training from which various predictive uncertainty measures can be extracted (e.g., entropies). Paragraph 0079 discloses one or more processors are configured to output the sets of distribution data in accordance with the neural network architecture in which each one of the neural network outputs a respective one of the sets of distribution data and wherein one or more of the processors are configured to monitor the uncertainty estimate values over time and to initiate a re-training sequence for the plurality of neural networks when an epistemic uncertainty value associated with the one or more of the plurality of neural networks exceeds a threshold epistemic value. When the uncertainty does not exceed the threshold and the training is completed, the data is output.
It would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to combine this teaching of Ahuja into the method and system of the combination of Mohades and Ahuja. By generating uncertainty values, one can determine when the NN is providing acceptable information and take appropriate actions, such as outputting that information. This will improve the function of the method and system by limiting the amount of time a system is trained.
Regarding claim 27, the combination discloses wherein the posterior neural network and the likelihood neural network are shared across each iteration of the machine learning model (Mohades: Abstract: in the first algorithm…a four-layer feed-forward neural network is applied. In the second algorithm. Recurrent neural networks are employed to extract the sparsity structure. The dictionary and matrix are used in each algorithm, sharing the information.).
Regarding claim 28, the combination discloses wherein each respective iteration of the machine learning model has a respective posterior neural network and a respective likelihood neural network (Mohades: Abstract: in the first algorithm…a four-layer feed-forward neural network is applied. In the second algorithm. Recurrent neural networks are employed to extract the sparsity structure. So each step has a respective NN.).
6. Claim 10 is rejected under 35 U.S.C. 103 as being unpatentable over MOHADES ZOHREH ET AL: "Deep Neural Network for Compressive Sensing and Application to Massive MIMO Channel Estimation", CIRCUITS, SYSTEMS AND SIGNAL PROCESSING, CAMBRIDGE, MS, US, vol. 40, no. 9, 9 March 2021 (2021-03-09), pages 4474-4489, XP037528052, ISSN: 0278-081X, DOI: 10.1007/S00034-021-01675-Z in view of Son et al (US 2017/0257230).
Regarding claim 10, Mohades discloses the method and system stated above. Mohades does not disclose performing one of analog beamforming, beam selection, or spectral efficiency prediction based on the current sparse channel representation.
Son discloses a channel estimation method and apparatus in a wireless communication system as stated in the abstract. Paragraph 0085 discloses the MS 105 measures the reception signal strengths of the received reference beams, selects reference beams for which the received signal strengths are equal to or greater than a predetermined reference value as effective beams and configures the sparse channel suing the selected effective beams so as to estimate the channel. The MS 105 generates feedback based on the selected effective beams. Therefore, Son discloses performing beam selection based on the current sparse channel representation.
It would have been obvious for one of ordinary skill in the art before the effective filing date to incorporate the steps of determine the effective channel and selecting beams accordingly as taught by Son into the method and system of Mohades. By utilizing the determined channel estimate appropriate actions can be taken that improve the function of the communication system, improving efficiency and effectiveness of the system.
7. Claims 11, 13, 19 and 20 are rejected under 35 U.S.C. 103 as being unpatentable over MOHADES ZOHREH ET AL: "Deep Neural Network for Compressive Sensing and Application to Massive MIMO Channel Estimation", CIRCUITS, SYSTEMS AND SIGNAL PROCESSING, CAMBRIDGE, MS, US, vol. 40, no. 9, 9 March 2021 (2021-03-09), pages 4474-4489, XP037528052, ISSN: 0278-081X, DOI: 10.1007/S00034-021-01675-Z in view of Agrawal et al (US 2021/0142158).
Regarding claim 11, Mohades discloses the method and system as stated above. Mohades does not disclose generating a first loss by processing the current sparsifying dictionary using a prior neural network; generating a second loss based on the current sparse channel representation; and refining the posterior neural network, the likelihood neural network, and the prior neural network based on the first loss and the second loss.
Agarawal discloses neural networks for forwarding error correction decoding as stated in the abstract. Agrawal further discloses optimizing trainable parameters of the neural network to minimize loss functions as stated in paragraph 0012. Paragraph 0018 discloses the loss function. By training the parameters, the neural network can be refined to minimize that loss.
It would have been obvious for one of ordinary skill in the art to incorporate determining the loss function in a neural network and optimizing trainable parameters to minimize that loss as taught by Agrawal into the method and system of Mohades. By minimizing the loss in the system, the system can minimize errors and other issues, improving the operation of the system.
Regarding claim 13, Mohades discloses a processor-implemented method (Heading 3: Proposed methods. The method will be executed by a means. The means for executing the methods is a processor.), comprising:
Receiving a sensing matrix and a current channel observation for a digital communication channel (Heading 3.1: sparse matrix. After obtaining network parameters, the iterative algorithm is applied to reconstruct the sparse signal.);
generating a current sparsifying dictionary by processing a sensing matrix and a current channel observation for a digital communication channel using a posterior neural network in a first iteration of a machine learning model (Heading 2.3 Dictionary Learning. A learned dictionary can represent a given signal with a smaller representation error than a predefined dictionary. In this paper we use a block sparce dictionary learning to learn a dictionary which can offer satisfying offer a sparse representation of a massive MIMO channel. Abstract: in the first algorithm…a four-layer feed-forward neural network is applied.); and
generating a current sparse channel representation by processing the current sparsifying dictionary, the sensing matrix, and the current channel observation using a likelihood neural network in the first iteration (Heading 4 channel estimation in massive MIMO system with proposed algorithms. For this purpose, we utilize channel measurements H and consider equation 11. Abstract: in the second algorithm. Recurrent neural networks are employed to extract the sparsity structure.).
Mohades does not disclose generating a first loss by processing the current sparsifying dictionary using a prior neural network; generating a second loss based on the current sparse channel representation; and refining the posterior neural network, the likelihood neural network, and the prior neural network based on the first loss and the second loss.
Agarawal discloses neural networks for forwarding error correction decoding as stated in the abstract. Agrawal further discloses optimizing trainable parameters of the neural network to minimize loss functions as stated in paragraph 0012. Paragraph 0018 discloses the loss function. By training the parameters, the neural network can be refined to minimize that loss.
It would have been obvious for one of ordinary skill in the art to incorporate determining the loss function in a neural network and optimizing trainable parameters to minimize that loss as taught by Agrawal into the method and system of Mohades. By minimizing the loss in the system, the system can minimize errors and other issues, improving the operation of the system.
Regarding claim 19, the combination discloses wherein the posterior neural network and the likelihood neural network are shared across each iteration of the machine learning model (Mohades: Abstract: in the first algorithm…a four-layer feed-forward neural network is applied. In the second algorithm. Recurrent neural networks are employed to extract the sparsity structure. The dictionary and matrix are used in each algorithm, sharing the information.).
Regarding claim 20, the combination discloses wherein each respective iteration of the machine learning model has a respective posterior neural network and a respective likelihood neural network (Mohades: Abstract: in the first algorithm…a four-layer feed-forward neural network is applied. In the second algorithm. Recurrent neural networks are employed to extract the sparsity structure. So each step has a respective NN.).
8. Claims 12, 14, 15, 17 and 18 are rejected under 35 U.S.C. 103 as being unpatentable over MOHADES ZOHREH ET AL: "Deep Neural Network for Compressive Sensing and Application to Massive MIMO Channel Estimation", CIRCUITS, SYSTEMS AND SIGNAL PROCESSING, CAMBRIDGE, MS, US, vol. 40, no. 9, 9 March 2021 (2021-03-09), pages 4474-4489, XP037528052, ISSN: 0278-081X, DOI: 10.1007/S00034-021-01675-Z in view of Agrawal et al (US 2021/0142158) further in view of Li et al (US 2023/0053588).
Regarding claims 12 and 14, the combination of Mohades and Agrawal discloses the method stated above. The combination does not disclose wherein generating the first loss using the prior neural network comprises: generating a mean value and a variance value by processing a previous sparsifying dictionary generated in a previous iteration of the machine learning model using the prior neural network; and generating a distribution having the mean value and the variance value.
Li discloses a neural network as stated in the abstract. Paragraph 0060 discloses the multi-resolution generator neural network then utilizes a first linear neural network layer to generate a mean value associated with the reduced resolution set. The multi-resolution generator also utilizes a second linear layer to generate a variance associated with the resolution feature set. Accordingly, the multi-resolution generator neural network generates a mean value and the variance value to represent the distribution of features.
It would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to utilize the teaching of Li into the method and system of the combination of Mohades and Agrawal. Using the mean and the variance to determine or generate the distribution to represent the output is well known in neural network operation. Utilizing well known steps allows for a simpler method or algorithm, reducing costs and complexity.
Regarding claim 15, the combination discloses wherein the distribution is a prior distribution defined in the claim including the current sparsifying dictionary, and the previous sparsifying dictionary generated in the previous iteration of the machine learning model (Mohades: heading 4: the dictionary is found through learning from the given channel measurements as discussed in section 2.3. When the channel measurements change during each iteration, the dictionary will be found based on those channel measurements.).
Regarding claim 17, the combination of Mohades and Agrawal discloses the method stated above. The combination does not disclose wherein generating the current sparsifying dictionary using the posterior neural network comprises: generating a mean value and a variance value by processing the sensing matrix, the current channel observation, and a previous sparse channel representation generated in a previous iteration of the machine learning model, using the posterior neural network; and sampling a distribution having the mean value and the variance value.
Li discloses a neural network as stated in the abstract. Paragraph 0060 discloses the multi-resolution generator neural network then utilizes a first linear neural network layer to generate a mean value associated with the reduced resolution set. The multi-resolution generator also utilizes a second linear layer to generate a variance associated with the resolution feature set. Accordingly, the multi-resolution generator neural network generates a mean value and the variance value to represent the distribution of features.
It would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to utilize the teaching of Li into the method and system of the combination of Mohades and Agrawal. Using the mean and the variance to determine or generate the distribution to represent the output is well known in neural network operation. Utilizing well known steps allows for a simpler method or algorithm, reducing costs and complexity.
Regarding claim 18, the combination of Mohades and Agrawal discloses the method stated above. Mohades does not disclose wherein generating the current sparse channel representation using the likelihood neural network comprises: generating a mean value by processing the current sparsifying dictionary, the sensing matrix, and the current channel observation using the likelihood neural network; and generating a distribution having the mean value and a variance value.
Li discloses a neural network as stated in the abstract. Paragraph 0060 discloses the multi-resolution generator neural network then utilizes a first linear neural network layer to generate a mean value associated with the reduced resolution set. The multi-resolution generator also utilizes a second linear layer to generate a variance associated with the resolution feature set. Accordingly, the multi-resolution generator neural network generates a mean value and the variance value to represent the distribution of features.
It would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to utilize the teaching of Li into the method and system of the combination of Mohades and Agrawal. Using the mean and the variance to determine or generate the distribution to represent the output is well known in neural network operation. Utilizing well known steps allows for a simpler method or algorithm, reducing costs and complexity.
9. Claim 16 is rejected under 35 U.S.C. 103 as being unpatentable over MOHADES ZOHREH ET AL: "Deep Neural Network for Compressive Sensing and Application to Massive MIMO Channel Estimation", CIRCUITS, SYSTEMS AND SIGNAL PROCESSING, CAMBRIDGE, MS, US, vol. 40, no. 9, 9 March 2021 (2021-03-09), pages 4474-4489, XP037528052, ISSN: 0278-081X, DOI: 10.1007/S00034-021-01675-Z in view of Agrawal et al (US 2021/0142158) in view of Li et al (US 2023/0053588) further in view of Ahuja et al (US 2020/0326667).
Regarding claim 16, the combination of Mohades, Agrawal and Li discloses the method stated above and discloses wherein the previous iteration corresponds to an input iteration of the machine learning model (Mohades: heading 4: the dictionary is found through learning from the given channel measurements as discussed in section 2.3. When the channel measurements change during each iteration, the dictionary will be found based on those channel measurements.). The combination does not disclose the previous sparsifying dictionary is sampled from a Gaussian distribution.
Ahuja discloses a neural network architecture as stated in the abstract. Paragraph 0079 discloses one or more processors are configured to output the sets of distribution data in accordance with the neural network architecture in which each one of the neural network outputs a respective one of the sets of distribution data and wherein one or more of the processors are configured to monitor the uncertainty estimate values over time and to initiate a re-training sequence for the plurality of neural networks when an epistemic uncertainty value associated with the one or more of the plurality of neural networks exceeds a threshold epistemic value. Paragraph 0062 discloses any suitable combination of multivariate Gaussians and Gaussian mixture models (GMMs) may be implemented to model the distributions at the various layers of the DNN architecture 400. This is advantageous as the features of the outer layers of DNNs tend to be well represented by simper Gaussian distributions. Using simpler distributions that are well represented reduces the complexity and processing of the system.
It would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to combine this teaching of Ahuja into the method and system of the combination of Mohades, Agrawal and Li. Using simpler distributions that are well represented reduces the complexity and processing of the system.
10. Claims 22-24 are rejected under 35 U.S.C. 103 as being unpatentable over MOHADES ZOHREH ET AL: "Deep Neural Network for Compressive Sensing and Application to Massive MIMO Channel Estimation", CIRCUITS, SYSTEMS AND SIGNAL PROCESSING, CAMBRIDGE, MS, US, vol. 40, no. 9, 9 March 2021 (2021-03-09), pages 4474-4489, XP037528052, ISSN: 0278-081X, DOI: 10.1007/S00034-021-01675-Z in view of Ahuja et al (US 2020/0326667) in view of Li et al (US 2023/0053588).
Regarding claim 22, the combination of Mohades and Ahuja discloses the method stated above. The combination does not disclose wherein generating the current sparsifying dictionary using the posterior neural network comprises: generating a mean value and a variance value by processing the sensing matrix, the current channel observation, and a previous sparse channel representation generated in a previous iteration of the machine learning model, using the posterior neural network; and sampling a distribution having the mean value and the variance value.
Li discloses a neural network as stated in the abstract. Paragraph 0060 discloses the multi-resolution generator neural network then utilizes a first linear neural network layer to generate a mean value associated with the reduced resolution set. The multi-resolution generator also utilizes a second linear layer to generate a variance associated with the resolution feature set. Accordingly, the multi-resolution generator neural network generates a mean value and the variance value to represent the distribution of features.
It would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to utilize the teaching of Li into the method and system of the combination of Mohades and Ahuja. Using the mean and the variance to determine or generate the distribution to represent the output is well known in neural network operation. Utilizing well known steps allows for a simpler method or algorithm, reducing costs and complexity.
Regarding claim 23, the combination of Mohades and Ahuja discloses the method stated above. The combination does not disclose wherein generating the current sparse channel representation using the likelihood neural network comprises: generating a mean value by processing the current sparsifying dictionary, the sensing matrix, and the current channel observation using the likelihood neural network; and generating a distribution having the mean value and a variance value.
Li discloses a neural network as stated in the abstract. Paragraph 0060 discloses the multi-resolution generator neural network then utilizes a first linear neural network layer to generate a mean value associated with the reduced resolution set. The multi-resolution generator also utilizes a second linear layer to generate a variance associated with the resolution feature set. Accordingly, the multi-resolution generator neural network generates a mean value and the variance value to represent the distribution of features.
It would have been obvious for one of ordinary skill in the art before the effective filing date of the claimed invention to utilize the teaching of Li into the method and system of the combination of Mohades and Ahuja. Using the mean and the variance to determine or generate the distribution to represent the output is well known in neural network operation. Utilizing well known steps allows for a simpler method or algorithm, reducing costs and complexity.
Regarding claim 24, the combination discloses wherein the distribution is a likelihood model defined in the claim, wherein: the recited distribution includes the current sparse channel representation, a ground truth channel state, sparsifying dictionaries generated in one or more previous iterations in the machine learning model, the current channel observation, and the sensing matrix (Mohades: Heading 2.3 Dictionary Learning. A learned dictionary can represent a given signal with a smaller representation error than a predefined dictionary. In this paper we use a block sparce dictionary learning to learn a dictionary which can offer satisfying offer a sparse representation of a massive MIMO channel. Abstract: in the first algorithm…a four-layer feed-forward neural network is applied. Heading 3.1: sparse matrix. Heading 4 channel estimation in massive MIMO system with proposed algorithms. For this purpose, we utilize channel measurements H and consider equation 11. Abstract: in the second algorithm. Recurrent neural networks are employed to extract the sparsity structure. The input and the previous channel state are reflected in the output.).
11. Claim 29 is rejected under 35 U.S.C. 103 as being unpatentable over MOHADES ZOHREH ET AL: "Deep Neural Network for Compressive Sensing and Application to Massive MIMO Channel Estimation", CIRCUITS, SYSTEMS AND SIGNAL PROCESSING, CAMBRIDGE, MS, US, vol. 40, no. 9, 9 March 2021 (2021-03-09), pages 4474-4489, XP037528052, ISSN: 0278-081X, DOI: 10.1007/S00034-021-01675-Z in view of Ahuja et al (US 2020/0326667) in view of Son et al (US 2017/0257230).
Regarding claim 29, the combination of Mohades and Ahuja discloses the method and system stated above. The combination does not disclose performing one of analog beamforming, beam selection, or spectral efficiency prediction based on the current sparse channel representation.
Son discloses a channel estimation method and apparatus in a wireless communication system as stated in the abstract. Paragraph 0085discloses the MS 105 measures the reception signal strengths of the received reference beams, selects reference beams for which the received signal strengths are equal to or greater than a predetermined reference value as effective beams and configures the sparse channel suing the selected effective beams so as to estimate the channel. The MS 105 generates feedback based on the selected effective beams. . Therefore, Son discloses performing beam selection based on the current sparse channel representation.
It would have been obvious for one of ordinary skill in the art before the effective filing date to incorporate the steps of determine the effective channel and selecting beams accordingly as taught by Son into the method and system of the combination of Mohades and Ahuja. By utilizing the determined channel estimate appropriate actions can be taken that improve the function of the communication system, improving efficiency and effectiveness of the system.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to KEVIN M. BURD whose telephone number is (571)272-3008. The examiner can normally be reached 9:30 - 5:00.
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, Chieh Fan can be reached at 571-272-3042. 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.
/KEVIN M BURD/Primary Examiner, Art Unit 2632 6/5/2026
Prosecution Insights