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 .
Status of Claims
This action is in reply to the application filed on 09/06/2024. Claims 1-20 are currently pending and have been examined.
Information Disclosure Statement
The information disclosure statements (IDS) submitted on 09/06/2024 and 02/04/2026 have been considered by the examiner and initialed copies of the IDS are hereby attached.
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.
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 1-3, 5-6, 9, 11-13, 17-19 are rejected under 35 U.S.C 103 as being unpatentable over Wang (CN117331021A) in view of Youn (US20230305111A1).
Regarding claim 1 Wang discloses: A radar system comprising (Page 1: “The Direction of Arrival (DOA) estimation refers to the direction of the incoming wave of the signal estimated from the observation of the plurality of array receiving antennas. DOA estimation is a major problem in array signal processing, and has wide application in various fields such as radar, sonar, wireless communication and so on.”): a plurality of transmitter modules configured to transmit a plurality of transmitted radar signals ; a plurality of receiver modules configured to receive reflections of the plurality of transmitted radar signals reflected by at least one object and to generate signals based on the received reflections (Pages 10-11: “The present invention also provides a robust direction-of-arrival estimation system based on ADMM-Net, comprising: a receiving module for receiving the signal source power and multiple beat signals from multiple array receiving antennas; a processing module for executing the robust direction-of-arrival estimation method based on ADMM-Net to realize DOA estimation; and a display module for displaying the DOA estimation result. The receiving module of the invention can be radar array antenna, microphone array, sonar array and so on. The processing module of the invention can be a computer processor CPU or GPU capable of calculating in parallel. The display module of the invention can be radar display screen, microphone receiver display screen and so on.“);
Wang does not teach “and a processor configured to: determine a measurement vector using signals received by the plurality of receiver modules, implement a neural network comprising a plurality of nodes arranged in a plurality of layers”
However, Youn in the analogous arts teaches: and a processor configured to: determine a measurement vector using signals received by the plurality of receiver modules (Para 0002: “ Embodiments of systems and methods for estimating DOA are disclosed. In one or more embodiments, a radar system includes an array of antennas, a radar transceiver connected to the array of antennas, and a signal processing unit connected to the radar transceiver and that includes processing circuitry and memory coupled to the processing circuitry, wherein the processing circuitry includes multiple vector processing units, each vector processing unit configured to receive an antenna input vector that is a representation of radar return signal amplitudes received by each antenna of the array of antennas”), implement a neural network comprising a plurality of nodes arranged in a plurality of layers (Para 0075: “ FIG. 4 is a functional block diagram of an example network structure 450, referred to as a circular convolutional neural network, that can be used to implement CC-LISTA. In the example of FIG. 4, the process for estimating a DOA is implemented as an L-layer CC-LISTA in which each layer 436 implements at least one circular convolution and a non-linear activation function. As is described below, the circular convolutional neural network of FIG. 4 can be implemented via the cluster 234 of vector processing units 236 and memory 238 of the signal processing unit 214 in the radar system 202 of FIG. 2.”).
Wang further teaches: determine an optimized output amplitude vector, wherein the optimized output amplitude vector defines a radar signal spectrum having peaks associated with the at least one object (Page 5: “Firstly, the multi-snapshot data received by the information source and array is sparsely transformed through the space complete dictionary, and the DOA estimation is transformed into the compressed sensing sparsity recovery problem; then expanding the ADMM algorithm to form the model driving depth network ADMM-Netwith interpretability; The ADMM-Net is used for reconstructing the information source power spectrum for DOA estimation.”) , by performing steps including: determining an expression defining an iteration of an optimization problem configured to determine the optimized output amplitude, wherein the expression includes a plurality of learnable parameters (Page 9: “Step 3.1, in the kth iteration of the model-driven deep network ADMM-Net, the first step of the algorithm is as follows: pk represents the result of the first step of the algorithm in the k-th iteration, p is the learning parameter of the deep network, zk + 1 represents the result of the second step of the algorithm in the (k-1) - th iteration, and beta k-1 represents the result of the third step of the algorithm in the (k-1) - th iteration;“), causing each node of the plurality of nodes in the neural network to iteratively solve the expression to determine optimized values of the plurality of learnable parameters (Page 9: “step 3.4, using the residual network to perform residual learning on the result obtained in the step 3.2;“), and determining, by a final node in the plurality of nodes (Page 9: “Step 3.3, in the k-th iteration of the model-driven deep network ADMM-Net, the step of updating the Lagrange multiplier (.beta. k) in the third step of the algorithm is as follows: [beta] k + [beta] k-1 + [beta] k-zk step 3.4, using the residual network to perform residual learning on the result obtained in the step 3.2“), the optimized output amplitude vector based on the optimized values of the plurality of learnable parameters determined by another node in the plurality of nodes, wherein the optimized output amplitude vector is a sparse signal vector (Page 7: “The power spectrum p is only a non-zero element at the source position, i.e., the spatial spectrum p is sparse. Thereby, the covariance matrix can be obtained by using the sample covariance matrix The power spectrum p is recovered to solve the DOA estimation problem, and the position corresponding to the peak value in p represents the information source direction.After converting into a sparse recovery problem, the problem is a typical sparse linear inverse problem, which can be solved by optimizing the following objective function:“); and determine an estimated direction of arrival of a first object using the optimized output amplitude vector (Page 10: “The present invention also provides a robust direction-of-arrival estimation system based on ADMM-Net”).
It would have been obvious to someone in the art prior to the effective filing date of the claimed invention to modify Wang with Youn to incorporate the feature of: and a processor configured to: determine a measurement vector using signals received by the plurality of receiver modules. Wang and Youn are all considered analogous arts as they all disclose the use of deep learning methods to process radar sensor data. However, Wang fails to disclose a feature of multiple receive antennas. This feature is disclosed by Youn. It would have been obvious to someone in the art prior to the effective filling date of the claimed invention to modify Wang with Youn to incorporate the feature of: and a processor configured to: determine a measurement vector using signals received by the plurality of receiver modules as such a feature would increase data quality and the efficiency of the system.
Regarding 2 the combination of Wang and Youn disclose all the limitations of claim 1. Youn further teaches: wherein the signals received by the plurality of receiver modules are associated with a MIMO virtual array with antenna elements positioned at integer multiples of unit value (Para 0051: “The array of antennas 210 may be an antenna array as is known in the field. FIG. 3 depicts an example of an array of antennas 310 for the radar system that includes an array of transmit antennas 318 (transmit antenna array) that includes multiple transmit antennas 320 and an array of receive antennas 322 (receive antenna array) that includes multiple receive antennas 324. For example, the transmit and receive antennas are patch antennas that are configured for a particular wavelength range such as 75-76 GHz. In the example of FIG. 3, the transmit antennas are configured in a linear array of four antennas along the z-dimension and the receive antennas are configured in a linear array of twelve antennas along the y-dimension. In one or more embodiments, the transmit antennas are evenly spaced along the z-dimension at, for example, λ/2 (where λ is the center wavelength of the wavelength range that is used for linear frequency modulation) and the receive antennas are evenly spaced along the y-dimension at, for example, λ/2 intervals. The configuration of the transmit and receive antennas could be used with different MIMO processing techniques to estimate the DOA in both elevation and azimuth. Although an example antenna array is described with reference to FIG. 3, other configurations of the antenna array are possible, including numbers of antennas, spacing (uniform or sparse), and location of the antennas. In one embodiment, there are four transmit antennas and eight receive antennas, which have sparse spacing, however, other combinations of antennas are possible.”).
Claims 11 and 17 recites limitations that are similar to those of claim 2, therefore claims 11 and 16 are rejected under the same rationale.
Regarding 3 the combination of Wang and Youn disclose all the limitations of claim 2. Youn further teaches: wherein the unit value is equal to half of a wavelength of the signals received by the plurality of receiver modules (Para 0051: “In the example of FIG. 3, the transmit antennas are configured in a linear array of four antennas along the z-dimension and the receive antennas are configured in a linear array of twelve antennas along the y-dimension. In one or more embodiments, the transmit antennas are evenly spaced along the z-dimension at, for example, λ/2 (where λ is the center wavelength of the wavelength range that is used for linear frequency modulation) and the receive antennas are evenly spaced along the y-dimension at, for example, λ/2 intervals.”).
Claims 12 and 18 recites limitations that are similar to those of claim 3, therefore claims 12 and 18 are rejected under the same rationale.
Regarding 5 the combination of Wang and Youn disclose all the limitations of claim 3. Youn further teaches: wherein a first learnable parameter in the plurality of learnable parameters is a circulant matrix (Para 0073: “the learning parameters, collectively referred to as weighting vectors, with l (layer) ranging from 0 to L−1, and W.sub.c is a circulant matrix. The derivation follows from the fact that the product of circulant matrices is also a circulant matrix.”).
Regarding 6 the combination of Wang and Youn disclose all the limitations of claim 1. Wang further teaches: wherein the neural network is implemented using an alternating direction method of multipliers (ADMM) model (Page 1: “The invention belongs to the technical field of array signal processing and deep learning, specifically to a robust direction-of-arrival estimation method based on ADMM-Net.”).
Claims 13 and 19 recites limitations that are similar to those of claim 6, therefore claims 13 and 19 are rejected under the same rationale.
Regarding 9 the combination of Wang and Youn disclose all the limitations of claim 1. Wang further teaches: wherein the expression includes activation functions that are executed by each node in the neural network (Page 9: “The specific structure of the residual network is: the first layer is 64 9 * 9 convolution cores and ReLU activation functions; the second layer is 32 convolution cores and ReLU activation functions of 1*1 size; The third layer is a 5 x 5 convolution kernel and a ReLU activation function.“).
Claims 15 and 20 recites limitations that are similar to those of claim 9, therefore claims 15 and 20 are rejected under the same rationale.
Claim 4 are rejected under 35 U.S.C 103 as being unpatentable over Wang (CN117331021A) in view of Youn (US20230305111A1) and further in view of Zeng (CN111767791A)
Regarding 4 the combination of Wang and Youn disclose all the limitations of claim 3. Wang does not teach “wherein a first learnable parameter in the plurality of learnable parameters is a hermitian-centrohermitian matrix “.
However, Zeng in the analogous art teaches wherein a first learnable parameter in the plurality of learnable parameters is a hermitian-centrohermitian matrix (Page 8: “S2. obtaining the characteristic extraction matrix of the covariance matrix R as the input of the deep neural network; it needs to be noted that the covariance matrix R is a Hermitian matrix; the upper triangular part and the lower triangular part have the same information; therefore, the upper triangular element of the covariance matrix R is placed in a column vector, and constructing feature extraction matrix according to the real part and the imaginary part of the column vector”).
It would have been obvious to someone in the art prior to the effective filing date of the claimed invention to modify Wang with Zeng to incorporate the feature of: wherein a first learnable parameter in the plurality of learnable parameters is a hermitian-centrohermitian matrix. Wang and Zeng are all considered analogous arts as they all disclose deep learning algorithms to process remote sensing data for direction of arrival estimation. However, Wang fails to disclose a feature of using a Hermitian matrix in the neural network training. This feature is disclosed by Zeng. It would have been obvious to someone in the art prior to the effective filling date of the claimed invention to modify Wang with Zeng to incorporate the feature of: wherein a first learnable parameter in the plurality of learnable parameters is a hermitian-centrohermitian matrix as such a feature would increase the efficiency of the system.
Claims 7-8 and 14 are rejected under 35 U.S.C 103 as being unpatentable over Wang (CN117331021A) in view of Youn (US20230305111A1) and further in view of Chetlur (US20210133583A1).
Regarding 7 the combination of Wang and Youn disclose all the limitations of claim 1. Wang does not teach “wherein the processor includes a plurality of processor cores and each node in the plurality of nodes is implemented by a processor core out of the plurality of processor cores “
However, Chetlur in the analogous arts teaches: wherein the processor includes a plurality of processor cores and each node in the plurality of nodes is implemented by a processor core out of the plurality of processor cores (Para 0052: “This document describes a system and method that improves training of a machine learning model by allowing node weights to be updated in parallel. In at least one embodiment, a plurality of workers perform forward propagation and back propagation to produce, in parallel, a set of gradients. In at least one embodiment, a worker may be a thread, process, processor, processor core, or parallel processing circuit that executes instructions in parallel with other workers. In at least one embodiment, gradients are distributed across workers, and each worker is assigned a subset of weights to which to apply gradients. In at least one embodiment, each worker applies gradients to its assigned subset of weights in parallel with other workers. In at least one embodiment, after gradients are applied, updated weights are distributed among workers so that each worker has a complete set of updated weights. In at least one embodiment, by updating weights in parallel with a plurality of workers, speed with which network can be trained is improved. In at least one embodiment, back propagation process is repeated iteratively until training is complete.”).
It would have been obvious to someone in the art prior to the effective filing date of the claimed invention to modify Wang with Chetlur to incorporate the feature of: wherein the processor includes a plurality of processor cores and each node in the plurality of nodes is implemented by a processor core out of the plurality of processor cores. Wang and Chetlur are all considered analogous arts as they all disclose deep learning algorithms to process remote sensing data. However, Wang fails to disclose a feature of model training using multiple processor cores. This feature is disclosed by Chetlur. It would have been obvious to someone in the art prior to the effective filling date of the claimed invention to modify Wang with Chetlur to incorporate the feature of: wherein the processor includes a plurality of processor cores and each node in the plurality of nodes is implemented by a processor core out of the plurality of processor cores as such a feature would increase the computational efficiency of the system.
Regarding 8 the combination of Wang, Youn and Chetlur disclose all the limitations of claim 7. Youn further teaches: wherein each processor core of the plurality of processor cores is implemented by an application-specific integrated circuit (ASIC) (Para 0098: “Embodiments of the invention may be implemented entirely in hardware or in an implementation containing both hardware and software elements. In one or more embodiments, the vector processing units are implemented in hardware as ASICs with specific hardware circuits (including logic and memory) configured to implement the circular convolution engine, the vector summing engine, and the activation engine of each vector processing unit. In embodiments which use software, the software may include but is not limited to firmware, resident software, microcode, etc.”).
Claim 14 recites limitations that are similar to those of claim 8, therefore claim 14 is rejected under the same rationale.
Claims 10 and 16 are rejected under 35 U.S.C 103 as being unpatentable over Wang (CN117331021A) in view of Youn (US20230305111A1) in view of Zeng (CN111767791A) and further in view of Chetlur (US20210133583A1).
Regarding claim 10 Wang discloses: A radar system comprising: a plurality of transmitter modules configured to transmit a plurality of transmitted radar signals comprising (Page 1: “The Direction of Arrival (DOA) estimation refers to the direction of the incoming wave of the signal estimated from the observation of the plurality of array receiving antennas. DOA estimation is a major problem in array signal processing, and has wide application in various fields such as radar, sonar, wireless communication and so on.”)
Youn in the analogous arts teaches: a plurality of receiver modules configured to receive reflections of the plurality of transmitted radar signals reflected by at least one object modules (Para 0002: “ Embodiments of systems and methods for estimating DOA are disclosed. In one or more embodiments, a radar system includes an array of antennas, a radar transceiver connected to the array of antennas, and a signal processing unit connected to the radar transceiver and that includes processing circuitry and memory coupled to the processing circuitry, wherein the processing circuitry includes multiple vector processing units, each vector processing unit configured to receive an antenna input vector that is a representation of radar return signal amplitudes received by each antenna of the array of antennas”);
Chetlur in the analogous arts teaches: a plurality of processor cores, wherein each processing core in the plurality of processing cores is associated with layers of a neural network (Para 0052: “This document describes a system and method that improves training of a machine learning model by allowing node weights to be updated in parallel. In at least one embodiment, a plurality of workers perform forward propagation and back propagation to produce, in parallel, a set of gradients. In at least one embodiment, a worker may be a thread, process, processor, processor core, or parallel processing circuit that executes instructions in parallel with other workers. In at least one embodiment, gradients are distributed across workers, and each worker is assigned a subset of weights to which to apply gradients. In at least one embodiment, each worker applies gradients to its assigned subset of weights in parallel with other workers. In at least one embodiment, after gradients are applied, updated weights are distributed among workers so that each worker has a complete set of updated weights. In at least one embodiment, by updating weights in parallel with a plurality of workers, speed with which network can be trained is improved. In at least one embodiment, back propagation process is repeated iteratively until training is complete.”)
Zeng in the analogous arts teaches: and a processor configured to: determine a measurement vector using signals received by the plurality of receiver modules, determining an expression defining an iteration of an optimization problem configured to determine an optimized output amplitude vector, wherein the expression includes a first parameter that is a hermitian-centrohermitian matrix or a circulant matrix (Page 8: “S2. obtaining the characteristic extraction matrix of the covariance matrix R as the input of the deep neural network; it needs to be noted that the covariance matrix R is a Hermitian matrix; the upper triangular part and the lower triangular part have the same information; therefore, the upper triangular element of the covariance matrix R is placed in a column vector, and constructing feature extraction matrix according to the real part and the imaginary part of the column vector”).
Wang further teaches: causing each processor core of the plurality of processor cores to solve iterations of the expression to determine an optimized value of the first parameter, causing a final processor core of the plurality of processor cores to determine optimized output amplitude vector based on the optimized value of the first parameter(Page 9: “Step 3.1, in the kth iteration of the model-driven deep network ADMM-Net, the first step of the algorithm is as follows: pk represents the result of the first step of the algorithm in the k-th iteration, p is the learning parameter of the deep network, zk + 1 represents the result of the second step of the algorithm in the (k-1) - th iteration, and beta k-1 represents the result of the third step of the algorithm in the (k-1) - th iteration;“) , wherein the optimized output amplitude vector is a sparse signal vector (Page 7: “The power spectrum p is only a non-zero element at the source position, i.e., the spatial spectrum p is sparse. Thereby, the covariance matrix can be obtained by using the sample covariance matrix The power spectrum p is recovered to solve the DOA estimation problem, and the position corresponding to the peak value in p represents the information source direction. After converting into a sparse recovery problem, the problem is a typical sparse linear inverse problem, which can be solved by optimizing the following objective function:“), and determine an estimated direction of arrival of a first object using the optimized output amplitude vector (Page 10: “The present invention also provides a robust direction-of-arrival estimation system based on ADMM-Net”).
The reason for combining Wang with Youn, Zeng and Chetlur is the same as one given in claims 1, 4 and 7.
Claim 16 recites limitations that are similar to those of claim 10, therefore claim 16 is rejected under the same rationale.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Bongani J. Mashele whose telephone number is (703)756-5861. The examiner can normally be reached Monday-Friday, 8:00AM-5:00PM (CT).
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 Resha H. Desai, can be reached on 571-270-7792. 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.
/BONGANI JABULANI MASHELE/Examiner, Art Unit 3648
/VLADIMIR MAGLOIRE/Supervisory Patent Examiner, Art Unit 3648
Prosecution Insights