DETAILED ACTION
Applicant’s amendment and arguments filed July 2, 2026 is acknowledged.
Claims 1, 31, and 34 have been amended.
Claims 4, 7, 9, 11-13, 15, 17-19, 2, 122, 25, and 27-29 are cancelled.
Claims 1-3, 5, 6, 8, 10, 14, 16, 20, 23, 24, 26, and 30-36 are currently pending.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102 of this title, 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 text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
This application currently names joint inventors. In considering patentability of the claims the examiner presumes that the subject matter of the various claims was commonly owned as of the effective filing date of the claimed invention(s) absent any evidence to the contrary. Applicant is advised of the obligation under 37 CFR 1.56 to point out the inventor and effective filing dates of each claim that was not commonly owned as of the effective filing date of the later invention in order for the examiner to consider the applicability of 35 U.S.C. 102(b)(2)(C) for any potential 35 U.S.C. 102(a)(2) prior art against the later invention.
The factual inquiries set forth in Graham v. John Deere Co., 383 U.S. 1, 148 USPQ 459 (1966), that are applied 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, 8, 10, 14, 16, 20, 23, 24, 26, and 30 are rejected under 35 U.S.C. 103 as being unpatentable over Hu et al. (hereinafter Hu) (Non-Patent Literature – “Channel Estimation Enhancement With Generative Adversarial Networks” – IEEE Transactions on Cognitive Communications and Networking, Vol. 7, No. 1, March 2021) in view of Zirwas et al. (hereinafter Zirwas) (U.S. Patent Application Publication # 2024/0056336 A1), and further in view of O’Shea et al. (hereinafter O’Shea) (U.S. Patent Application Publication # 2023/0299862 A1).
Regarding claim 1, Hu teaches and discloses a method for estimation of a radio propagation channel realization, performed in a wireless communication system (abstract; “…Improving the accuracy of channel estimation is a significant topic in the context of wireless communication…”; page 145, column 2, lines 1-5; “…the next generation wireless communication systems are going to encounter with lower latency, higher data rate, higher mobility and more stringent constraints on wireless resources. It is urgent to devote effort in improving the accuracy of channel estimation with limited temporal or spatial resources…”; figure 3), the method comprising:
obtaining a generative adversarial network, GAN, structure, wherein the GAN structure comprises a generative part and a discriminative part (page 145, column 2, lines 41-45; “…the training algorithm of GANs essentially plays a two-player minimax game over two adversarial nets, including a generator network and a discriminator network…”; page 146, column 1, lines 1-4; page 148, column 2, lines 34-37; “…The proposed CWGAN-GP framework is shown in Fig. 2, where the input of the generator and discriminator networks are both conditioned on information…”; Figure 2; teaches obtaining a generative adversarial network, GAN, comprising a generator part and discriminator part),
configuring the GAN structure as a conditioned GAN structure, where the generative part is arranged to be conditioned by pilot symbol data (training sequence) comprising radio propagation channel data obtained from pilot symbol transmissions over the radio propagation channel (page 146, column 2, lines 43-47; “…a conditional WGAN-GP (CWGAN-GP) framework for the problem of improving the training stability and the learning ability of GANs. The CWGAN-GP can utilize the training sequence and the corresponding receive sequence as a conditional information…”; page 147, column 1, lines 42-49; page 149, column 1, lines 12-28; “…Given the training sequence pk and the receive sequence yk , the trained network G can generate samples that can be considered as mimic receive sequences. In other words, the CWGAN-GP frame-work captures the conditional distribution…during the online training stage…”; teaches the GAN is configured as a conditional GAN, such as GWGAN-GP, where the generator is configured to be conditioned by a training sequence comprising propagation data from the sequence),
training the GAN structure by conditioning the generative part on the pilot symbol data and feeding a corresponding output from the generative part to the discriminative part together with reference channel realization data corresponding to the pilot symbol data (page 145, column 2, lines 41-45; “…the training algorithm of GANs essentially plays a two-player minimax game over two adversarial nets, including a generator network and a discriminator network…”; page 146, column 1, lines 44-51; “…our proposed GAN-based channel estimation enhancement, GANs are trained online with the actual receive sequence. Therefore, the actual receive sequence (corresponding to a given training sequence) is a training sample in our study. Remarkably, the actual samples used to train GANs are referred to as training samples, while those used to evaluate the performance of the trained GANs are referred to as testing samples…”; page 146, column 2, lines 43-47; “…a conditional WGAN-GP (CWGAN-GP) framework for the problem of improving the training stability and the learning ability of GANs. The CWGAN-GP can utilize the training sequence and the corresponding receive sequence as a conditional information…”; teaches training the conditional GAN , where the generator is configured to be conditioned by a training sequence comprising propagation data from the sequence and the corresponding output to the discriminator with reference data),
extracting a channel estimator from the GAN structure, the channel estimator being the generative part of the GAN structure, and estimating a radio propagation channel realization by feeding pilot symbol data to the channel estimator (abstract; page 146, column 2, lines 27-42; “…an online GAN-based channel estimation enhancement algorithm, where the enhanced estimated channel gain is obtained by the LS estimation method with the aforementioned combined sequence…The LS estimation method with the original receive sequence is set as a baseline. This baseline is com-pared with the proposed GAN-based algorithm…GANs are trained online until a stopping criterion is satisfied. Afterwards, the enhancement is implemented by the trained CWGAN-GP to generating the mimic receive sequences, and then channel estimation is performed with these new sequences…”; Channel Estimation Enhancement Stage; Figure 3; page 149, column 1, lines 12-28 and 40-42; “…the channel estimation enhancement stage, the enhancement is implemented by generating the mimic receive sequences according to the actual receive sequence yk . With these new sequences, channel gain can be estimated by applying the LS estimation method… the channel gain is estimated by directly using the LS estimation…”; page 149, column 2, lines 1-10; “…otherwise the channel gain is estimated following the online training stage and the channel estimation enhancement stage…”; teaches a channel estimator from the GAN for performing channel estimation of the propagation channel be feeding the sequence to the channel estimator).
However, Hu may not explicitly disclose configuring the GAN structure as a conditioned GAN structure, where the generative part is arranged to be conditioned by pilot symbol data comprising radio propagation channel data obtained from pilot symbol transmissions over the radio propagation channel (although Hu does teach training of the GAN is characterized by training symbols, pk,…Given the training sequence pk and the receive sequence yk , the trained network G can generate samples that can be considered as mimic receive sequences. In other words, the CWGAN-GP frame-work captures the conditional distribution…during the online training stage…”; page 148, column 2; page 149, column 1).
Nonetheless, in the same field of endeavor, Zirwas teaches and suggests configuring the GAN structure as a conditioned GAN structure ([0030]; “…This information is combined to produce a channel estimate. In some embodiments, conditional generative adversarial networks (cGANs) are used as the first machine-learning model…”), where the generative part is arranged to be conditioned by pilot symbol data comprising radio propagation channel data obtained from pilot symbol transmissions over the radio propagation channel ([0041]; [0047]; “…each of the M antenna elements can perform 1-bit measurements. Moreover, for those 1-bit measurements, it is assumed the user equipment (UE) sends the pilots mixed with a random sequence…”; [0074]; teaches configuring the cGAN with the GAN conditioned by pilot information for radio propagation channel).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to incorporate configuring the cGAN with the GAN conditioned by pilot information for radio propagation channel as taught by Zirwas with the method and apparatus for channel estimation enhancement with GAN as disclosed by Hu for the purpose of provide accurate channel knowledge for improving channel estimation, as suggested by Zirwas.
However, Hu, as modified by Zirwas, may not explicitly disclose a wireless communication system comprising one or more access points and one or more wireless devices for channel estimation (although Hu does teach a wireless communication network, which would inherently include related network devices, such as an access point and wireless devices and Zirwas teaches channel estimation related to a base station and UE).
Nonetheless, in the same field of endeavor, O’Shea teaches a wireless communication system comprising one or more access points and one or more wireless devices for channel estimation ([0045]; “…disclosed techniques are described primarily with respect to cellular communications networks, e.g., 3GPP 5G-NR cellular systems. However, the disclosed techniques are also applicable to other systems as noted above, including, for example, 4G LTE, 6G, or Wi-Fi networks…”; [0046]; [0061]; “…machine-learning network 212 is trained or deployed, or both, over one or more communications channels 207, or approximations of a communications channel, which can be, for example, a 5G-NR wireless communications channel for transmitting or receiving data in a cellular network…the device 201 or the device 208 is a mobile device, such as a cellular phone, a tablet or a notebook, while the other device is a network base station…”; figures 2A, 6, and 7; teaches a method for channel estimation in a wireless communication network including one or more base station and one or more mobile devices).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to incorporate a method for channel estimation in a wireless communication network including one or more base station and one or more mobile devices as taught by O’Shea with the method and apparatus for channel estimation enhancement with GAN as disclosed by Hu, as modified by Zirwas, for the purpose of using machine learning to improve network performance and optimize network parameters, as suggested by O’Shea.
Regarding claim 2, Hu discloses a method for channel estimation enhancement with GAN for use in wireless communication network, but may not explicitly disclose performed in part or entirely in one of the access points, in one of the wireless devices, and/or in a remote server.
Nonetheless, in the same field of endeavor, O’Shea further teaches and suggests performed in part or entirely in one of the access points, in one of the wireless devices, and/or in a remote server ([0045]; [0046]; [0061]; “…machine-learning network 212 is trained or deployed, or both, over one or more communications channels 207, or approximations of a communications channel, which can be, for example, a 5G-NR wireless communications channel for transmitting or receiving data in a cellular network…the device 201 or the device 208 is a mobile device, such as a cellular phone, a tablet or a notebook, while the other device is a network base station…”; figures 2A, 6, and 7; teaches a method for channel estimation in a wireless communication network including one or more base station and one or more mobile devices).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to incorporate a method for channel estimation in a wireless communication network including one or more base station and one or more mobile devices as taught by O’Shea with the method and apparatus for channel estimation enhancement with GAN as disclosed by Hu, as modified by Zirwas and O’Shea, for the purpose of using machine learning to improve network performance and optimize network parameters, as suggested by O’Shea.
Regarding claim 3, Hu discloses a method for channel estimation enhancement with GAN for use in wireless communication network, but may not explicitly disclose where the wireless communication system is an orthogonal frequency division multiplexed, OFDM, based system, and the pilot symbol transmissions comprise transmission of Demodulation Reference Signal, DMRS, resource elements, RE, in the OFDM based system.
Nonetheless, in the same field of endeavor, O’Shea further teaches and suggests where the wireless communication system is an orthogonal frequency division multiplexed, OFDM, based system, and the pilot symbol transmissions comprise transmission of Demodulation Reference Signal, DMRS, resource elements, RE, in the OFDM based system ([0010]; [0037]; [0048]; [0098]; teaches an OFDM system and pilot symbols comprising DMRS resource elements in the OFDM system).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to incorporate an OFDM system and pilot symbols comprising DMRS resource elements in the OFDM system as taught by O’Shea with the method and apparatus for channel estimation enhancement with GAN as disclosed by Hu, as modified by Zirwas and O’Shea, for the purpose of using machine learning to improve network performance and optimize network parameters, as suggested by O’Shea.
Regarding claim 5, Hu, as modified by Zirwas and O’Shea, further teaches and suggests wherein the training is performed until a termination criterion associated with a capability of the generative part to generate an output classified as a true channel realization by the discriminative part (page 146, column 2, lines 37-39; “…the proposed GAN-based algorithm, GANs are trained online until a stopping criterion is satisfied. Afterwards, the enhancement is implemented by the trained CWGAN-GP to generating the mimic receive sequences, and then channel estimation is performed…”; page 150, column 1, lines 18-25; “…the CWGAN-GP can be alternately and iteratively trained until this stopping criterion is satisfied…”; teaches training until a stopping criterion is satisfied).
Regarding claim 6, Hu discloses a method for channel estimation enhancement with GAN for use in wireless communication network, but may not explicitly disclose performing an offline training procedure comprising generating the pilot symbol data and the reference channel realization data by computer simulation of a radio propagation channel model, the radio propagation channel model comprising any of a 3GPP tapped delay line, TDL, model, a 3GPP clustered delay line, CDL, model, and a 3GPP spatial channel model, SCM, model.
Nonetheless, in the same field of endeavor, O’Shea further teaches and suggests performing an offline training procedure comprising generating the pilot symbol data and the reference channel realization data by computer simulation of a radio propagation channel model, the radio propagation channel model comprising any of a 3GPP tapped delay line, TDL, model, a 3GPP clustered delay line, CDL, model, and a 3GPP spatial channel model, SCM, model ([0008]; [0080]; [0081]; teaches pre-training by computer simulation of channel models).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to incorporate pre-training by computer simulation of channel models as taught by O’Shea with the method and apparatus for channel estimation enhancement with GAN as disclosed by Hu, as modified by O’Shea, for the purpose of using machine learning to improve network performance and optimize network parameters, as suggested by O’Shea.
Regarding claim 8, Hu, as modified by Zirwas and O’Shea, further teaches and suggests performing an additional online training procedure involving pilot symbol transmissions over a radio propagation channel between an access point and a wireless device in the wireless communication system (page 146, column 1, lines 44-51; “…our proposed GAN-based channel estimation enhancement, GANs are trained online with the actual receive sequence. Therefore, the actual receive sequence (corresponding to a given training sequence) is a training sample in our study. Remarkably, the actual samples used to train GANs are referred to as training samples, while those used to evaluate the performance of the trained GANs are referred to as testing samples…”; page 146, column 2, lines 27-42; “…an online GAN-based channel estimation enhancement algorithm, where the enhanced estimated channel gain is obtained by the LS estimation method with the aforementioned combined sequence…The LS estimation method with the original receive sequence is set as a baseline. This baseline is compared with the proposed GAN-based algorithm…GANs are trained online until a stopping criterion is satisfied. Afterwards, the enhancement is implemented by the trained CWGAN-GP to generating the mimic receive sequences, and then channel estimation is performed with these new sequences…”; page 149, column 2, “Online Training Stage”; teaches online training procedure over a network channel).
Regarding claim 10, Hu, as modified by Zirwas and O’Shea, further teaches and suggests performing an online training procedure comprising extracting the pilot symbol data and the reference channel realization data from an ongoing communication in the wireless communication system, the ongoing communication in the wireless communication system comprising either: a transmission of pilot-weaved frames comprising only known information symbols where the online training procedure comprises training of one or more GAN structures by extracting one or more pre-determined pilot symbol patterns from the received pilot-weaved frames; and/or transmission of pilot-weaved frames comprising predetermined pilot symbol patterns and pseudo-pilot symbols ((page 146, column 1, lines 44-51; “…our proposed GAN-based channel estimation enhancement, GANs are trained online with the actual receive sequence. Therefore, the actual receive sequence (corresponding to a given training sequence) is a training sample in our study. Remarkably, the actual samples used to train GANs are referred to as training samples, while those used to evaluate the performance of the trained GANs are referred to as testing samples…”; page 146, column 2, lines 27-42; “…an online GAN-based channel estimation enhancement algorithm, where the enhanced estimated channel gain is obtained by the LS estimation method with the aforementioned combined sequence…The LS estimation method with the original receive sequence is set as a baseline. This baseline is compared with the proposed GAN-based algorithm…GANs are trained online until a stopping criterion is satisfied. Afterwards, the enhancement is implemented by the trained CWGAN-GP to generating the mimic receive sequences, and then channel estimation is performed with these new sequences…”; page 149, column 2, “Online Training Stage”; teaches an online training procedure related to training sequences).
Regarding claim 14, Hu discloses a method for channel estimation enhancement with GAN for use in wireless communication network, but may not explicitly disclose either or both of the following: training the GAN structure using a respective cross entropy loss function for each of the generative part and the discriminative part; and training the GAN structure using a loss function comprising an adversarial loss and a Euclidean distance between the output from the generative part and the corresponding reference channel realization data.
Nonetheless, in the same field of endeavor, O’Shea further teaches and suggests either or both of the following: training the GAN structure using a respective cross entropy loss function for each of the generative part and the discriminative part; and training the GAN structure using a loss function comprising an adversarial loss and a Euclidean distance between the output from the generative part and the corresponding reference channel realization data ([0068]; [0069]; [0075]; [0081]; teaches training of the GAN using loss function).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to incorporate training of the GAN using loss function as taught by O’Shea with the method and apparatus for channel estimation enhancement with GAN as disclosed by Hu, as modified by Zirwas and O’Shea, for the purpose of using machine learning to improve network performance and optimize network parameters, as suggested by O’Shea.
Regarding claim 16, Hu discloses a method for channel estimation enhancement with GAN for use in wireless communication network, but may not explicitly disclose training the GAN structure at an access point of the wireless communication system based on communication over an uplink, UL, from a first wireless device to the access point, and transmitting the channel estimator to the first wireless device upon the generative part of the GAN structure reaching a predetermined convergence criterion; and downloading the discriminative part to the first wireless device and using the discriminative part in a fault detection structure at the wireless device.
Nonetheless, in the same field of endeavor, O’Shea further teaches and suggests training the GAN structure at an access point of the wireless communication system based on communication over an uplink, UL, from a first wireless device to the access point, and transmitting the channel estimator to the first wireless device upon the generative part of the GAN structure reaching a predetermined convergence criterion; and downloading the discriminative part to the first wireless device and using the discriminative part in a fault detection structure at the wireless device ([0061]; [0068]; [0081]; [0124]; [0125]; teaches training of the GAN at a base station based on an uplink from a UE and transmitting a channel estimation to the UE).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to incorporate training of the GAN at a base station based on an uplink from a UE and transmitting a channel estimation to the UE as taught by O’Shea with the method and apparatus for channel estimation enhancement with GAN as disclosed by Hu, as modified by Zirwas and O’Shea, for the purpose of using machine learning to improve network performance and optimize network parameters, as suggested by O’Shea.
Regarding claim 20, Hu discloses a method for channel estimation enhancement with GAN for use in wireless communication network, but may not explicitly disclose transmitting a channel estimator trained at a first access point to the wireless device in response to either: the wireless device performing a handover procedure for service by the first access point, wherein the channel estimator is transmitted by the first access point; or the wireless device entering a geographical area, wherein the GAN channel estimator is trained based on communication in the geographical area.
Nonetheless, in the same field of endeavor, O’Shea further teaches and suggests transmitting a channel estimator trained at a first access point to the wireless device in response to either: the wireless device performing a handover procedure for service by the first access point, wherein the channel estimator is transmitted by the first access point; or the wireless device entering a geographical area, wherein the GAN channel estimator is trained based on communication in the geographical area ([0012]; [0046]; [0052]; [0061]; teaches transmitting a channel estimation and training of the GAN at a base station in response to the location of the UE).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to incorporate transmitting a channel estimation and training of the GAN at a base station in response to the location of the UE as taught by O’Shea with the method and apparatus for channel estimation enhancement with GAN as disclosed by Hu, as modified by Zirwas and O’Shea, for the purpose of using machine learning to improve network performance and optimize network parameters, as suggested by O’Shea.
Regarding claim 23, Hu discloses a method for channel estimation enhancement with GAN for use in wireless communication network, but may not explicitly disclose transmitting a channel estimator to the wireless device, where the channel estimator has been trained based on communication involving a specific type of wireless device, where the wireless device is associated with the specific type of wireless device.
Nonetheless, in the same field of endeavor, O’Shea further teaches and suggests transmitting a channel estimator to the wireless device, where the channel estimator has been trained based on communication involving a specific type of wireless device, where the wireless device is associated with the specific type of wireless device ([0012]; [0046]; [0052]; [0061]; teaches transmitting a channel estimation and training of the GAN at a base station based on the type of wireless device).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to incorporate transmitting a channel estimation and training of the GAN at a base station based on the type of wireless device as taught by O’Shea with the method and apparatus for channel estimation enhancement with GAN as disclosed by Hu, as modified by Zirwas and O’Shea, for the purpose of using machine learning to improve network performance and optimize network parameters, as suggested by O’Shea.
Regarding claim 24, Hu, as modified by Zirwas and O’Shea, further teaches and suggests training the GAN structure by conditioning the generative part on the pilot symbol data and feeding a corresponding output from the generative part to the discriminative part together with reference channel realization data concurrent with the pilot symbol data, and extracting a channel estimator for estimating a current channel realization from the GAN structure; and processing a radio transmission received by a network node from an access point or from a wireless device based on the estimated current radio propagation channel realization (page 145, column 2, lines 41-45; “…the training algorithm of GANs essentially plays a two-player minimax game over two adversarial nets, including a generator network and a discriminator network…”; page 146, column 1, lines 44-51; “…our proposed GAN-based channel estimation enhancement, GANs are trained online with the actual receive sequence. Therefore, the actual receive sequence (corresponding to a given training sequence) is a training sample in our study. Remarkably, the actual samples used to train GANs are referred to as training samples, while those used to evaluate the performance of the trained GANs are referred to as testing samples…”; page 146, column 2, lines 43-47; “…a conditional WGAN-GP (CWGAN-GP) framework for the problem of improving the training stability and the learning ability of GANs. The CWGAN-GP can utilize the training sequence and the corresponding receive sequence as a conditional information…”; teaches training the conditional GAN , where the generator is configured to be conditioned by a training sequence comprising propagation data from the sequence and the corresponding output to the discriminator with reference data).
Regarding claim 26, Hu, as modified by Zirwas and O’Shea, further teaches and suggests training the GAN structure by conditioning the generative part on the pilot symbol data and feeding a corresponding output from the generative part to the discriminative part together with reference channel realization data contiguous in time to the pilot symbol data and extracting a channel predictor for estimating a future channel realization from the GAN structure (page 145, column 2, lines 41-45; “…the training algorithm of GANs essentially plays a two-player minimax game over two adversarial nets, including a generator network and a discriminator network…”; page 146, column 1, lines 44-51; “…our proposed GAN-based channel estimation enhancement, GANs are trained online with the actual receive sequence. Therefore, the actual receive sequence (corresponding to a given training sequence) is a training sample in our study. Remarkably, the actual samples used to train GANs are referred to as training samples, while those used to evaluate the performance of the trained GANs are referred to as testing samples…”; page 146, column 2, lines 43-47; “…a conditional WGAN-GP (CWGAN-GP) framework for the problem of improving the training stability and the learning ability of GANs. The CWGAN-GP can utilize the training sequence and the corresponding receive sequence as a conditional information…”; teaches training the conditional GAN , where the generator is configured to be conditioned by a training sequence comprising propagation data from the sequence and the corresponding output to the discriminator with reference data).
Regarding claim 30, Hu discloses a method for channel estimation enhancement with GAN for use in wireless communication network, but may not explicitly disclose training the GAN structure to estimate and/or predict a radio propagation channel realization as a channel response comprising complex elements in a MIMO channel matrix, and/or vectors spanning a MIMO channel matrix eigen-vector space and/or a MIMO channel precoding matrix index, PMI.
Nonetheless, in the same field of endeavor, O’Shea further teaches and suggests training the GAN structure to estimate and/or predict a radio propagation channel realization as a channel response comprising complex elements in a MIMO channel matrix, and/or vectors spanning a MIMO channel matrix eigen-vector space and/or a MIMO channel precoding matrix index, PMI ([0055]; [0081]; [0089]; teaches training the GAN for channel estimation comprising a MIMO channel matrix).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to incorporate training the GAN for channel estimation comprising a MIMO channel matrix as taught by O’Shea with the method and apparatus for channel estimation enhancement with GAN as disclosed by Hu, as modified by Zirwas and O’Shea, for the purpose of using machine learning to improve network performance and optimize network parameters, as suggested by O’Shea.
Claims 31-36 are rejected under 35 U.S.C. 103 as being unpatentable over O’Shea et al. (hereinafter O’Shea) (U.S. Patent Application Publication # 2023/0299862 A1) in view of Hu et al. (hereinafter Hu) (Non-Patent Literature – “Channel Estimation Enhancement With Generative Adversarial Networks” – IEEE Transactions on Cognitive Communications and Networking, Vol. 7, No. 1, March 2021), and further in view of Zirwas et al. (hereinafter Zirwas) (U.S. Patent Application Publication # 2024/0056336 A1).
Regarding claim 31, O’Shea teaches and discloses a network node (base station; figures 2A, 6 and 7) comprised in a wireless communication system (wireless communication network; figures 2A, 6, and 7), wherein the network node is configured to facilitate estimation of a radio propagation channel realization between one or more access points and one or more wireless devices ([0045]; [0046]; [0061]; “…machine-learning network 212 is trained or deployed, or both, over one or more communications channels 207, or approximations of a communications channel, which can be, for example, a 5G-NR wireless communications channel for transmitting or receiving data in a cellular network…the device 201 or the device 208 is a mobile device, such as a cellular phone, a tablet or a notebook, while the other device is a network base station…”; figures 2A, 6, and 7; teaches a method for channel estimation in a wireless communication), the network node comprising processing circuitry (inherent component of the base station) arranged to: obtain a generative adversarial network, GAN, structure and the GAN conditioned by pilot symbol data comprising radio propagation channel data obtained from pilot symbol transmissions over the radio propagation channel ([0037]; “…Some of the time-frequency subcarriers within an OFDM grid can be allocated as reference tones or pilot signals. Pilot signals can be resource elements with known values; these can be referred to as pilot resource elements...”; [0048]; [0061]; [0081]; teaches training of the GAN at a base station based on an uplink from a UE and transmitting a channel estimation to the UE).
However, O’Shea does not explicitly disclose wherein the GAN structure comprises a generative part and a discriminative part, train the GAN structure by conditioning the generative part on the pilot symbol data and feeding a corresponding output from the generative part to the discriminative part together with reference channel realization data corresponding to the pilot symbol data, and extract a channel estimator from the GAN structure, the channel estimator being the generative part of the GAN structure.
Nonetheless, in the same field of endeavor, Hu teaches and discloses wherein the GAN structure comprises a generative part and a discriminative part (page 145, column 2, lines 41-45; “…the training algorithm of GANs essentially plays a two-player minimax game over two adversarial nets, including a generator network and a discriminator network…”; page 146, column 1, lines 1-4; page 148, column 2, lines 34-37; “…The proposed CWGAN-GP framework is shown in Fig. 2, where the input of the generator and discriminator networks are both conditioned on information…”; Figure 2; teaches obtaining a generative adversarial network, GAN, comprising a generator part and discriminator part), train the GAN structure by conditioning the generative part on the pilot symbol data and feeding a corresponding output from the generative part to the discriminative part together with reference channel realization data corresponding to the pilot symbol data (page 145, column 2, lines 41-45; “…the training algorithm of GANs essentially plays a two-player minimax game over two adversarial nets, including a generator network and a discriminator network…”; page 146, column 1, lines 44-51; “…our proposed GAN-based channel estimation enhancement, GANs are trained online with the actual receive sequence. Therefore, the actual receive sequence (corresponding to a given training sequence) is a training sample in our study. Remarkably, the actual samples used to train GANs are referred to as training samples, while those used to evaluate the performance of the trained GANs are referred to as testing samples…”; page 146, column 2, lines 43-47; “…a conditional WGAN-GP (CWGAN-GP) framework for the problem of improving the training stability and the learning ability of GANs. The CWGAN-GP can utilize the training sequence and the corresponding receive sequence as a conditional information…”; teaches training the conditional GAN , where the generator is configured to be conditioned by a training sequence comprising propagation data from the sequence and the corresponding output to the discriminator with reference data), and extract a channel estimator from the GAN structure, the channel estimator being the generative part of the GAN structure (abstract; page 146, column 2, lines 27-42; “…an online GAN-based channel estimation enhancement algorithm, where the enhanced estimated channel gain is obtained by the LS estimation method with the aforementioned combined sequence…The LS estimation method with the original receive sequence is set as a baseline. This baseline is com-pared with the proposed GAN-based algorithm…GANs are trained online until a stopping criterion is satisfied. Afterwards, the enhancement is implemented by the trained CWGAN-GP to generating the mimic receive sequences, and then channel estimation is performed with these new sequences…”; Channel Estimation Enhancement Stage; Figure 3; page 149, column 1, lines 12-28 and 40-42; “…the channel estimation enhancement stage, the enhancement is implemented by generating the mimic receive sequences according to the actual receive sequence yk . With these new sequences, channel gain can be estimated by applying the LS estimation method… the channel gain is estimated by directly using the LS estimation…”; page 149, column 2, lines 1-10; “…otherwise the channel gain is estimated following the online training stage and the channel estimation enhancement stage…”; teaches a channel estimator from the GAN for performing channel estimation of the propagation channel be feeding the sequence to the channel estimator).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to incorporate the method and apparatus for channel estimation enhancement with GAN and training the GAN for channel estimation as taught by Hu with the method and apparatus for channel estimation with GAN as disclosed by O’Shea, for the purpose of improving the accuracy of channel estimation, as suggested by Hu.
However, O’Shea, as modified by Hu, may not explicitly disclose configure the GAN structure as a conditioned GAN structure, where the generative part is arranged to be conditioned by pilot symbol data comprising radio propagation channel data obtained from pilot symbol transmissions over the radio propagation channel (although Hu does teach training of the GAN is characterized by training symbols, pk,…Given the training sequence pk and the receive sequence yk , the trained network G can generate samples that can be considered as mimic receive sequences. In other words, the CWGAN-GP frame-work captures the conditional distribution…during the online training stage…”; page 148, column 2; page 149, column 1).
Nonetheless, in the same field of endeavor, Zirwas teaches and suggests configuring the GAN structure as a conditioned GAN structure ([0030]; “…This information is combined to produce a channel estimate. In some embodiments, conditional generative adversarial networks (cGANs) are used as the first machine-learning model…”), where the generative part is arranged to be conditioned by pilot symbol data comprising radio propagation channel data obtained from pilot symbol transmissions over the radio propagation channel ([0041]; [0047]; “…each of the M antenna elements can perform 1-bit measurements. Moreover, for those 1-bit measurements, it is assumed the user equipment (UE) sends the pilots mixed with a random sequence…”; [0074]; teaches configuring the cGAN with the GAN conditioned by pilot information for radio propagation channel).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to incorporate configuring the cGAN with the GAN conditioned by pilot information for radio propagation channel as taught by Zirwas with the method and apparatus for channel estimation enhancement with GAN as disclosed by O’Shea, as modified by Hu, for the purpose of provide accurate channel knowledge for improving channel estimation, as suggested by Zirwas.
Regarding claims 32 and 35, O’Shea, as modified by Hu and Zirwas, further teaches and suggests wherein the processing circuitry is further arranged to estimate a radio propagation channel realization by feeding pilot symbol data to the channel estimator ([0037]; [0048]; [0061]; [0081]; teaches training of the GAN at a base station and determining channel estimation by feeding pilot symbols for the channel estimation).
Regarding claims 33 and 36, O’Shea, as modified by Hu and Zirwas, further teaches and suggests comprising a network interface, wherein the processing circuitry is further arranged to transmit the channel estimator to an access point and/or to a wireless device comprised in the wireless communication system ([0045]; [0046]; [0061]; “…machine-learning network 212 is trained or deployed, or both, over one or more communications channels 207, or approximations of a communications channel, which can be, for example, a 5G-NR wireless communications channel for transmitting or receiving data in a cellular network…the device 201 or the device 208 is a mobile device, such as a cellular phone, a tablet or a notebook, while the other device is a network base station…”; figures 2A, 6, and 7; teaches a method for channel estimation in a wireless communication network including one or more base station and one or more mobile devices).
Regarding claim 34, O’Shea teaches and discloses a wireless device (UE/wireless device; figures 2A, 6 and 7) comprised in a wireless communication system (wireless communication network; figures 2A, 6, and 7), wherein the wireless device is configured to facilitate estimation of a radio propagation channel realization between one or more access points and the wireless device ([0045]; [0046]; [0061]; “…machine-learning network 212 is trained or deployed, or both, over one or more communications channels 207, or approximations of a communications channel, which can be, for example, a 5G-NR wireless communications channel for transmitting or receiving data in a cellular network…the device 201 or the device 208 is a mobile device, such as a cellular phone, a tablet or a notebook, while the other device is a network base station…”; figures 2A, 6, and 7; teaches a method for channel estimation in a wireless communication), the wireless device comprising processing circuitry (inherent component of the UE/wireless device) arranged to: obtain a generative adversarial network, GAN, structure and the GAN conditioned by pilot symbol data comprising radio propagation channel data obtained from pilot symbol transmissions over the radio propagation channel ([0037]; [0048]; [0061]; [0081]; teaches training of the GAN at a base station based on an uplink from a UE and transmitting a channel estimation to the UE).
However, O’Shea does not explicitly disclose wherein the GAN structure comprises a generative part and a discriminative part, train the GAN structure by conditioning the generative part on the pilot symbol data and feeding a corresponding output from the generative part to the discriminative part together with reference channel realization data corresponding to the pilot symbol data, and extract a channel estimator from the GAN structure, the channel estimator being the generative part of the GAN structure.
Nonetheless, in the same field of endeavor, Hu teaches and discloses wherein the GAN structure comprises a generative part and a discriminative part (page 145, column 2, lines 41-45; “…the training algorithm of GANs essentially plays a two-player minimax game over two adversarial nets, including a generator network and a discriminator network…”; page 146, column 1, lines 1-4; page 148, column 2, lines 34-37; “…The proposed CWGAN-GP framework is shown in Fig. 2, where the input of the generator and discriminator networks are both conditioned on information…”; Figure 2; teaches obtaining a generative adversarial network, GAN, comprising a generator part and discriminator part), train the GAN structure by conditioning the generative part on the pilot symbol data and feeding a corresponding output from the generative part to the discriminative part together with reference channel realization data corresponding to the pilot symbol data (page 145, column 2, lines 41-45; “…the training algorithm of GANs essentially plays a two-player minimax game over two adversarial nets, including a generator network and a discriminator network…”; page 146, column 1, lines 44-51; “…our proposed GAN-based channel estimation enhancement, GANs are trained online with the actual receive sequence. Therefore, the actual receive sequence (corresponding to a given training sequence) is a training sample in our study. Remarkably, the actual samples used to train GANs are referred to as training samples, while those used to evaluate the performance of the trained GANs are referred to as testing samples…”; page 146, column 2, lines 43-47; “…a conditional WGAN-GP (CWGAN-GP) framework for the problem of improving the training stability and the learning ability of GANs. The CWGAN-GP can utilize the training sequence and the corresponding receive sequence as a conditional information…”; teaches training the conditional GAN , where the generator is configured to be conditioned by a training sequence comprising propagation data from the sequence and the corresponding output to the discriminator with reference data), and extract a channel estimator from the GAN structure, the channel estimator being the generative part of the GAN structure (abstract; page 146, column 2, lines 27-42; “…an online GAN-based channel estimation enhancement algorithm, where the enhanced estimated channel gain is obtained by the LS estimation method with the aforementioned combined sequence…The LS estimation method with the original receive sequence is set as a baseline. This baseline is com-pared with the proposed GAN-based algorithm…GANs are trained online until a stopping criterion is satisfied. Afterwards, the enhancement is implemented by the trained CWGAN-GP to generating the mimic receive sequences, and then channel estimation is performed with these new sequences…”; Channel Estimation Enhancement Stage; Figure 3; page 149, column 1, lines 12-28 and 40-42; “…the channel estimation enhancement stage, the enhancement is implemented by generating the mimic receive sequences according to the actual receive sequence yk . With these new sequences, channel gain can be estimated by applying the LS estimation method… the channel gain is estimated by directly using the LS estimation…”; page 149, column 2, lines 1-10; “…otherwise the channel gain is estimated following the online training stage and the channel estimation enhancement stage…”; teaches a channel estimator from the GAN for performing channel estimation of the propagation channel be feeding the sequence to the channel estimator).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to incorporate the method and apparatus for channel estimation enhancement with GAN and training the GAN for channel estimation as taught by Hu with the method and apparatus for channel estimation with GAN as disclosed by O’Shea, for the purpose of improving the accuracy of channel estimation, as suggested by Hu.
However, O’Shea, as modified by Hu, may not explicitly disclose configure the GAN structure as a conditioned GAN structure, where the generative part is arranged to be conditioned by pilot symbol data comprising radio propagation channel data obtained from pilot symbol transmissions over the radio propagation channel (although Hu does teach training of the GAN is characterized by training symbols, pk,…Given the training sequence pk and the receive sequence yk , the trained network G can generate samples that can be considered as mimic receive sequences. In other words, the CWGAN-GP frame-work captures the conditional distribution…during the online training stage…”; page 148, column 2; page 149, column 1).
Nonetheless, in the same field of endeavor, Zirwas teaches and suggests configuring the GAN structure as a conditioned GAN structure ([0030]; “…This information is combined to produce a channel estimate. In some embodiments, conditional generative adversarial networks (cGANs) are used as the first machine-learning model…”), where the generative part is arranged to be conditioned by pilot symbol data comprising radio propagation channel data obtained from pilot symbol transmissions over the radio propagation channel ([0041]; [0047]; “…each of the M antenna elements can perform 1-bit measurements. Moreover, for those 1-bit measurements, it is assumed the user equipment (UE) sends the pilots mixed with a random sequence…”; [0074]; teaches configuring the cGAN with the GAN conditioned by pilot information for radio propagation channel).
Therefore, it would have been obvious to a person of ordinary skill in the art before the effective filing date of the claimed invention to incorporate configuring the cGAN with the GAN conditioned by pilot information for radio propagation channel as taught by Zirwas with the method and apparatus for channel estimation enhancement with GAN as disclosed by O’Shea, as modified by Hu, for the purpose of provide accurate channel knowledge for improving channel estimation, as suggested by Zirwas.
Response to Arguments
Applicant’s arguments, filed July 2, 2026, with respect to the rejection(s) of claim(s) 1-3, 5, 6, 8, 10, 14, 16, 20, 23, 24, 26, and 30-36 have been fully considered and are persuasive. Therefore, the rejection has been withdrawn. However, upon further consideration, a new ground(s) of rejection is made in view of Zirwas et al. (U.S. Patent Application Publication # 2024/0056336 A1).
Conclusion
The prior art made of record and not relied upon is considered pertinent to Applicant’s disclosure.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to SUK JIN KANG whose telephone number is (571) 270-1771. The examiner can normally be reached on Monday-Friday 8am-5pm.
Examiner interviews are available via telephone, in-person, and video conferencing using a USPTO supplied web-based collaboration tool. To schedule an interview, applicant is encouraged to use the USPTO Automated Interview Request (AIR) at http://www.uspto.gov/interviewpractice.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, Chirag Shah can be reached on (571) 272-3144. The fax phone number for the organization where this application or proceeding is assigned is 571-273-8300.
Information regarding the status of an application may be obtained from the Patent Application Information Retrieval (PAIR) system. Status information for published applications may be obtained from either Private PAIR or Public PAIR. Status information for unpublished applications is available through Private PAIR only. For more information about the PAIR system, see http://pair-direct.uspto.gov. Should you have questions on access to the Private PAIR system, contact the Electronic Business Center (EBC) at 866-217-9197 (toll-free). If you would like assistance from a USPTO Customer Service Representative or access to the automated information system, call 800-786-9199 (IN USA OR CANADA) or 571-272-1000.
Any inquiry of a general nature or relating to the status of this application or proceeding should be directed to the receptionist/customer service whose telephone number is (571) 272-2600.
/Suk Jin Kang/
Examiner, Art Unit 2477
August 28, 2026
Prosecution Insights