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 .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 05 May 2023 is in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statement is being considered by the examiner.
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 2, 4, 5, 15, 16 is/are rejected under 35 U.S.C. 103 as being unpatentable over “One-Bit-Over-the-Air Aggregation for Communication-Efficient federated Edge Learning: Desing and Convergence Analysis” (Zhu et al.) in view of “Collaborative Machine Learning at the Wireless Edge with Blind Transmitters” (Amir et al).
Regarding claim 1, Zhu et al. discloses: “a non-coherent over-the-air computation methodology (Page 2120, Column 2: “One promising solution is over-the-air aggregation also called over-the-air computation (AirComp) that exploits the waveform superposition property of the wireless medium to support simultaneous transmission by all the devices”) occurring in both uplink (UL) and downlink (DL), sequentially, in a multi-cell environment for federated edge learning (FEEL) (Page 2120, Column 2: “A popular framework, called federated edge learning (FEEL), distributes the task of model training
over many edge devices”) without using channel state information (CSI) (Page 2121, Column 2: “channel-state information (CSI) requirement for over the-air aggregation [i.e., “Air Comp’) is relaxed by exploiting multiple antennas at the edge server”), comprising: providing a distributed machine-learning model to be trained with the update vectors received at a plurality of edge servers (ESs) as transmitted from a plurality of edge devices (EDs) (Page 2122, Column 1, C. Contributions and Organization: “In this paper, we consider the implementation of over-the air aggregation for FEEL over a practical wireless system with digital modulation”; “we design an elaborate FEEL scheme, called OBDA, which features one-bit gradient quantization and QAM modulation at devices, and over-the-air majority-vote based gradient-decoding at the edge server”); and performing methodology operations comprising: transmitting local updates vectors (Page 2123, Column 2, B. Communication Model, “Local gradient estimates of edge devices are transmitted to the edge server over a broadband Multiple Access Channel (MAC)”; [i.e., updates of EDs/ESs are sent through wireless] ) as weighted votes with respective of the plurality of edge servers (ESs) functioning as aggregation nodes in the UL via a wireless multi-cell environment (Page 2123, Footnote 2: “the global gradient estimate is a weighted average of the local ones … the desired weighted aggregation of the local gradient estimate can also be attained by the proposed over-the-air aggregation with additional pre-processing” [i.e., the local gradients are essentially the votes and these local gradients are weighted]), independently detecting orthogonal signaling based majority vote (MV) data at each ES in the UL (Page 2124, Fig. 2(b); Page 2125, Column 1, B. Receiver Design: “Fig. 2(b) shows the receiver design for the edge server. It has the same architecture as a conventional OFDM receiver except that the digital detector is replaced with a majority-vote based decoder for estimating the global gradient-update from the received signal” [i.e., majority vote is determined here]), broadcasting the detected MVs from the ESs (Page 2124, Fig. 2(a); Page 2124, Column 2, A. Transmitter Design, “The transmitter design for edge devices is shown in
Fig. 2(a). The design builds on a conventional OFDM transmitter. However, unlike in conventional communication systems, where coded data bits are passed to the OFDM encoder, here we feed raw quantized bits without any coding”; “we apply one-bit quantization of local gradient estimates, which simply corresponds to taking the signs of the local gradient parameters”; “Each of the binary gradient parameters is modulated into one binary phase shift keying (BPSK) symbol”), and inputting the MVs into the machine-learning model to be updated, wherein the EDs determine the sign of the gradient through over-the-air computation using orthogonal signaling based majority vote (MV) in the DL” (Page 2125, Column 2, “Finally, to attain a global gradient estimate from g for model updating, a majority-vote based decoder is adopted and enforced by simply taking the element-wise sign of g: (Majority-vote based decoder) v = sign(g) (17). The operation essentially estimates the global gradient-update by over-the-air element-wise majority vote based on the one-bit quantized local gradient estimates attained at different devices” [i.e., taking the majority votes into account to update the model]).
However, Zhu et al. does not clearly disclose the remaining limitations of the claim. To that end, Amiri et al. discloses: “without using channel state information (CSI) at a plurality of edge devices (EDs) or edges servers (ESs)” (ABSTRACT: “We assume that the channel state information (CSI) is available only at the PS [i.e., parameter server]. We instead allow the PS to employ multiple antennas to alleviate the destructive fading effect, which cannot be cancelled by the transmitters due to the lack of CSI”; Page 2, column 2: “We assume that the PS has perfect CSI, while there is no CSI at the wireless devices”). It is respectfully submitted that it would have been obvious to combine Zhu et al. with the invention of Amiri et al. to alleviate the fading effect and enable the ESs/Eds to alleviated destructive effects and align updates (e.g., see Amiri et al. @ ABSTRACT).
Regarding claim 15, Zhu et al. discloses: “a non-coherent over-the-air computation system (Page 2120, Column 2: “One promising solution is over-the-air aggregation also called over-the-air computation (AirComp) that exploits the waveform superposition property of the wireless medium to support simultaneous transmission by all the devices”) for both uplink (UL) and downlink (DL) channels in a multi-cell environment, for federated edge learning (FEEL) (Page 2120, Column 2: “A popular framework, called federated edge learning (FEEL), distributes the task of model training over many edge devices”) without using channel state information (CSI) (Page 2121, Column 2: “channel-state information (CSI) requirement for over the-air aggregation [i.e., “Air Comp’) is relaxed by exploiting multiple antennas at the edge server”) at a plurality of edge devices (EDs) or at edge servers (ESs), comprising: a machine-learning model training to process data received at a plurality of edge servers (ESs) as transmitted from a plurality of edge devices (EDs) (Page 2125, Column 2, “Finally, to attain a global gradient estimate from g for model updating, a majority-vote based decoder is adopted and enforced by simply taking the element-wise sign of g: (Majority-vote based decoder) v = sign(g) (17). The operation essentially estimates the global gradient-update by over-the-air element-wise majority vote based on the one-bit quantized local gradient estimates attained at different devices” [i.e., taking the majority votes into account to update the model]); one or more processors; and one or more non-transitory computer-readable media that store instructions that, when executed by the one or more processors, cause the one or more processors to perform operations (Page 2122, Colum1, C. Contributions and Organization: “we design an elaborate FEEL scheme, called OBDA, which features one-bit gradient quantization and QAM modulation at devices, and over-the-air majority-vote based gradient-decoding at the edge server. This novel design will allow implementing AirComp across devices that are endowed with digital modulation chips, without requiring significant changes in the hardware or the communication architecture” [i.e., “digital modulation chips” inherently include “one or more processors and non-transitory computer readable media to perform operations”]), the operations comprising: transmitting local update vectors (Page 2123, Column 2, B. Communication Model, “Local gradient estimates of edge devices are transmitted to the edge server over a broadband Multiple Access Channel (MAC)”; [i.e., updates of EDs/ESs are sent through wireless] ) as weighted votes with respective of the plurality of edge servers (ESs) functioning as aggregation nodes in the UL channel via a wireless multi-cell environment (Page 2123, Footnote 2: “the global gradient estimate is a weighted average of the local ones … the desired weighted aggregation of the local gradient estimate can also be attained by the proposed over-the-air aggregation with additional pre-processing” [i.e., the local gradients are essentially the votes and these local gradients are weighted]), independently detecting orthogonal signaling based majority vote (MV) data at each ES in the UL channel (Page 2124, Fig. 2(b); Page 2125, Column 1, B. Receiver Design: “Fig. 2(b) shows the receiver design for the edge server. It has the same architecture as a conventional OFDM receiver except that the digital detector is replaced with a majority-vote based decoder for estimating the global gradient-update from the received signal” [i.e., majority vote is determined here]), broadcasting the detected MVs from the Ess (Page 2124, Fig. 2(a); Page 2124, Column 2, A. Transmitter Design, “The transmitter design for edge devices is shown in Fig. 2(a). The design builds on a conventional OFDM transmitter. However, unlike in conventional communication systems, where coded data bits are passed to the OFDM encoder, here we feed raw quantized bits without any coding”; “we apply one-bit quantization of local gradient estimates, which simply corresponds to taking the signs of the local gradient parameters”; “Each of the binary gradient parameters is modulated into one binary phase shift keying (BPSK) symbol”), and inputting the MVs into the machine-learning model to be updated, wherein the EDs determine the sign of the gradient through over-the-air computation using orthogonal signaling based majority vote (MV) in the DL channel” (Page 2125, Column 2, “Finally, to attain a global gradient estimate from g for model updating, a majority-vote based decoder is adopted and enforced by simply taking the element-wise sign of g: (Majority-vote based decoder) v = sign(g) (17). The operation essentially estimates the global gradient-update by over-the-air element-wise majority vote based on the one-bit quantized local gradient estimates attained at different devices” [i.e., taking the majority votes into account to update the model]).
In addition, Amiri et al. discloses: “without using channel state information (CSI) at a plurality of edge devices (EDs) or edges servers (ESs)” (ABSTRACT: “We assume that the channel state information (CSI) is available only at the PS [i.e., parameter server]. We instead allow the PS to employ multiple antennas to alleviate the destructive fading effect, which cannot be cancelled by the transmitters due to the lack of CSI”; Page 2, column 2: “We assume that the PS has perfect CSI, while there is no CSI at the wireless devices”).
With respect to claims 2 and 16, Zhu et al. discloses: “the votes comprise orthogonal frequency division multiplexing (OFDM) symbols over multiple OFDM subcarriers (Page 2123, Column 2, B. Communication Model: “OFDM modulation is adopted to divide the available bandwidth B into M orthogonal sub-channels”), and aggregating operations use one-bit broadband digital aggregation (OBDA) (Page 2120, Column 1, ABSTRACT: “we propose in this work a novel digital version of broadband over-the-air aggregation, called one-bit broadband digital aggregation (OBDA)”) and frequency-shift keying (FSK)-based methodology” (Page 2124, Column 2, III ONE-BIT BROADBAND DIGITAL AGGREGATION
(OBDA): SYSTEM DESIGN: “Each of the binary gradient parameters is modulated into
one binary phase shift keying (BPSK) symbol” (i.e., FSK).
With respect to claim 4, Zhu et al. discloses: “further including exploiting interference in the multi-cell environment in both UL and DL for computations” (Page 2123, column 2, B. Communication Model: “To cope with the frequency selective fading and inter-symbol interference, OFDM modulation is adopted”; Page 2129, column 2: “we will consider the generalization of the current work to multi-cell FEEL, where the effect of inter-cell
interference should also be taken into account”).
With respect to claim 5, Zhu et al. discloses: “transmitted symbols from an ED superpose with other EDs in the cell, and with EDs in neighboring cells; and the MV calculation at the ESs in the UL exploits interference from the EDs located in the neighboring cells” (Page 2123, Column 2, B. Communication Model: “Given the simultaneous transmission of all participating devices, the server receives superimposed waveforms”).
With respect to claim 8, Amiri et al. discloses: “a fading channel, long-term channel variations are captured by regenerating the channels between the ESs and the EDs independently for each communication round” (ABSTRACT: “We instead allow the PS to employ multiple antennas to alleviate the destructive fading effect, which cannot be cancelled by the transmitters due to the lack of CSI”; Page 2, column 2: “We assume that the PS has perfect CSI, while there is no CSI at the wireless devices”).
Claims 7 and 21 are rejected under 35 U.S.C. 103 as being unpatentable over Zhu et al. in view Amir et al. and US Patent Application Publication No. 20230231690 (Li et al_3).
Claim 7 and 21 are dependent upon claims 1 and 15, respectively. As discussed above, claims 1 and 15 are disclosed by the combination of Zhu et al. in view Amir et al. Thus, those limitations of claim 1 and 15 recited in claims 7 and 21, respectively, are also disclosed by the combination of Zhu et al. and Amir et al.
However, the combination of Zhu et al. and Amir et al. does not clearly disclose the remaining limitations of the claims. To that end, Li et al_3. discloses: “the over-the-air computation methodology according to claims 1 and 15, wherein the machine learning model comprises artificial intelligence technology ([0094]: “… an over the air computation procedure. In such systems, the transmitted data may include parameters or gradients associated with updating a local data model (e.g., an artificial intelligence or machine learning model)”) over wireless or sensor networks, 5G or higher, 6G wireless standardization, or IEEE 802.11 Wi-Fi” ([0230]: “the described techniques may be applicable to various other wireless communications systems such as Ultra Mobile Broadband (UMB), Institute of Electrical and Electronics Engineers (IEEE) 802.11 (Wi-Fi)”). It is respectfully submitted that it would have been obvious to one of ordinary skill in the art at the time of the invention to modify the combination of Zhu et al. and Amir et al. in order to provide machine learning models and artificial intelligence technology (e.g., see Li et al_3. @ [0230]).
Claim 14 is rejected under 35 U.S.C. 103 as being unpatentable over Zhu et al. in view Amir et al. and US Patent Application Publication No. 20230244952 (Li et al).
Claim 14 is dependent upon claim 1. As discussed above, claim 1 is disclosed by the combination of Zhu et al. in view Amir et al. Thus, those limitations of claim 1 recited in claim 14 are also disclosed by the combination of Zhu et al. and Amir et al.
However, the combination of Zhu et al. and Amir et al. does not clearly disclose the remaining limitations of the claim. To that end, Li et al. discloses: “providing one or more processors; and providing one or more non-transitory computer-readable media that store
instructions that, when executed by the one or more processors, cause the one or more
processors to perform the methodology operations“ ([0011]: “a non-transitory computer-readable medium storing a set of instructions for wireless communication includes one or more instructions that, when executed by one or more processors of a UE, cause the UE to determine quantized parameters in an RNN, or gradients to derive the RNN, based at least in part on AI modeling at the UE as part of a federated edge learning system”). It is respectfully submitted that it would have been obvious to one of ordinary skill in the art at the time of the invention to modify the combination of Zhu et al. and Amir et al. in order to provide non-transitory computer medium to work with a processor to perform methods such as determining quantized parameters (e.g., see Li et al. @ [0011]).
Claim 23 is rejected under 35 U.S.C. 103 as being unpatentable over Zhu et al.) in view Amir et al. and US Patent Application Publication No. 20230413070 (Li et al_2).
Claim 23 is dependent upon claim 15. As discussed above, claim 15 is disclosed by the combination of Zhu et al.) in view Amir et al. Thus, those limitations of claim 15 recited in claim 23 are also disclosed by the combination of Zhu et al. and Amir et al.
However, the combination of Zhu et al. and Amir et al. does not clearly disclose the remaining limitations of the claim. To that end, Li et al_2. discloses: “the over-the-air computation system ([0053]: “an over-the-air computation of global gradients for the federated learning task”) of claim 15, wherein the machine-learning model comprises a convolution neural network with multiple convolutional layers (FIG. 4C; [0059]: “a locally connected neural network is a convolutional neural network. FIG. 4C illustrates an example of a convolutional neural network 406. The convolutional neural network 406 may be configured such that the connection strengths associated with the inputs for each neuron in the second layer are shared (e.g., 408)”; FIG. 4D; [0062]: “feature maps 420 may be further convolved via one or more subsequent convolutional layers (not shown) to generate one or more subsequent sets of feature maps (not shown)”). It is respectfully submitted that it would have been obvious to one of ordinary skill in the art at the time of the invention to modify the combination of Zhu et al. and Amir et al. in order to provide a convolutional neural network with more convolutional layers(e.g., see Li et al_2. @ [0062]).
Allowable Subject Matter
Claims 3, 6, 9-13, 17-20 and 22 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MYRON K WYCHE whose telephone number is (571)272-3390. The examiner can normally be reached 7:30 am - 3:30 pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Kathy Wang-Hurst can be reached at 571-270-5371. 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.
/Myron Wyche/ 07 March 2026
Primary Examiner AU2644