Detailed Action
This Office Action is in response to the remarks entered on 06/12/2026. Claims 6-11 and 16-19 have been canceled. Claims 1-5 and 12-15 are presently pending.
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 .
Claim Rejections - 35 USC § 112
Amended claims were received on 06/12/2026. 35 U.S.C. 112 rejections have been withdrawn.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
Claims 1, 3-5 and 12-15 are rejected under 35 U.S.C. 103 as being unpatentable over Zeng et al. (US 20210273706 A1, hereinafter ‘Zeng’) in view of Aragao et al. (Aragao et al, “A Mechanism to Control the Congestion in Machine-to-Machine Communication in LTE-A Networks”, 2017, hereinafter ‘Aragao’) in view of Sattiraju et al. (Sattiraju et al, “Performance Analysis of Deep Learning based on Recurrent Neural Networks for Channel Coding”, 2018, hereinafter ‘Sattiraju’) and further in view of Liva et al. (“Short Turbo Codes over High Order Fields”, 2013, hereinafter ‘Liva’).
Regarding Claim 1, Zeng teaches:
A method comprising: ([Zeng, 0092; Fig. 8] discloses the UE comprises CSI Encoder 804 and CSI Decoder 802, Encoder/Decoder parameters 806, and a parameter generation 805. [Zeng, 0059 and 0088] collectively discloses that the CSI encoder and the CSI decoder are implemented using neural networks)
transmitting, by a user equipment (UE) including a first neural network encoder and a first neural network decoder[Zeng, 0003-0005; 0044-0045; 0050] collectively disclose the UE transmitting CSI to the Base Station (uplink) and the Base Station communicating with the UE on the downlink)
receiving, by the UE, a random access response from the base station; ([Zeng, 0064] discloses that the CSI encoder 311 is implemented by base stations 105 and UEs 115. [Zeng, 0072] discloses the payload portion may be encoded by the CSI encoder 311 and includes random access procedure messages. [Zeng, 0067-0068] discloses the first network node (base station 105) providing new CSIRS-based CSI encoder parameters (first parameter) to the second neural network node (UE 115) for use in channel compression, and the first network node sending the decoder parameter (second parameter) to the second network node)
performing, by the UE, an uplink transmission to the base station [Zeng, 0003-0005; 0044-0045; 0050] collectively disclose the UE transmitting CSI to the Base Station (uplink) and the Base Station communicating with the UE on the downlink)
receiving, by the UE, information from a base station, wherein the information includes a first parameter related to the first neural network encoder and a second parameter related to the first neural network decoder; and ([Zeng, 0003-0005; 0044-0045; 0050] collectively disclose the UE transmitting CSI to the Base Station (uplink) and the Base Station communicating with the UE on the downlink. [Zeng, 0067-0068] discloses the first network node (base station 105) providing new CSIRS-based CSI encoder parameters (first parameter) to the second neural network node (UE 115) for use in channel compression, and the first network node sending the decoder parameter (second parameter) to the second network node)
communicating, by the UE, with the base station based on the information, ([Zeng, 0067-0068] discloses the first network node (base station 105) providing new CSIRS-based CSI encoder parameters (neural network parameter) to the second neural network node (UE 115) for use in channel compression, and the first network node sending the decoder parameter to the second network node. [Zeng, 0062] discloses the UE encoding the CSI with a CSI encoder using neural-network based channel compression and sending the CSI to the base station. The UE also may send additional information regarding which reference signal the CSI encoded payload is based upon)
wherein the UE transmits uplink data to the base station based on the first parameter, ([Zeng, 0062] discloses the UE encoding the CSI with a CSI encoder using neural-network based channel compression and sending the CSI to the base station. The UE also may send additional information regarding which reference signal the CSI encoded payload is based upon)
wherein the UE receives downlink data from the base station based on the second parameter, ([Zeng, 0065-0066] The UE contains a CSI encoder and a CSI decoder where the CSI decoder receives CSI feedback from a communication link. [Zeng, 0003-0005; 0044-0045; 0050] collectively disclose the UE transmitting CSI to the Base Station (uplink) and the Base Station communicating with the UE on the downlink)
However, Zeng does not specifically disclose:
transmitting, by a user equipment (UE) including a first neural network encoder and a first neural network decoder, a preamble to a base station through a physical random access channel (PRACH);
performing, by the UE, an uplink transmission to the base station based on an uplink grant scheduled in the random access response;
receiving, by the UE, a contention resolution message from the base station;
wherein the first neural network encoder includes an interleaver, a recursive systematic convolutional (RSC) code, an accumulator, and a neural networks arranged in parallel,
wherein an input data sequence input to the first neural network encoder is branched into a plurality of paths to provide different input data respectively to the plurality of neural networks, and
wherein the plurality of paths comprise at least a first path passing through the accumulator to provide first input data to a first neural network among the plurality of neural networks, and a second path passing through the interleaver to provide second input data ot a second neural network among the plurality of neural networks.
Aragao teaches:
transmitting, [Aragao, page 2, left col, A. Random-Access Channel Procedure, line 1-16] (Msg1) discloses transmitting a preamble code to eNodeB (the base station) on the PRACH)
performing, by the UE, an uplink transmission to the base station based on an uplink grant scheduled in the random access response; ([Aragao, page 2, left col, A. Random-Access Channel Procedure, line 1-16] Random Access Response – RAR (Msg2) and (Msg3) disclose eNodeB receiving the access request, assigns a identifier to devices and grants resources on the uplink channel for subsequent messages exchanges (i.e., uplink grant scheduled in the random access response), and the device (UE) send the identifier assigned in the previous message)
receiving, by the UE, a contention resolution message from the base station; ([Aragao, page 2, left col, A. Random-Access Channel Procedure, line 1-16] (Msg4) discloses waiting for and receiving the contention resolution message from the eNodeB (the base station). If the device identifier is present in the contention message, an ACK message is sent to the eNodeB)
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention, having both the teachings of Zeng and Aragao to use the method of transmitting a preamble to the base station, performing an uplink transmission to the base station, and receiving a contention resolution message from the base station of Aragao to implement the machine learning based communication method of Zeng. The suggestion and/or motivation for doing so is to improve the accuracy and security of the communication system by confirming that a correct UE has been identified after transmitting signal.
However, Zeng in view of Aragao do not specifically disclose:
wherein the first neural network encoder includes an interleaver, a recursive systematic convolutional (RSC) code, an accumulator, and a neural networks arranged in parallel,
wherein the plurality of paths comprise at least a first path passing through the accumulator to provide first input data to a first neural network among the plurality of neural networks, and a second path passing through the interleaver to provide second input data or a second neural network among the plurality of neural networks.
Sattiraju teaches:
wherein the first neural network encoder includes an interleaver, a recursive systematic convolutional (RSC) code, a neural networks arranged in parallel, ([Sattiraju, page 1, right col, line 31 - page 2, left col, line 2; Fig. 1] and [Sattiraju, page 2, left col, line 19-26] collectively discloses a LTE turbo encoder which includes Deep Learning encoders connected in parallel and separated by an interleaver. LTE uses two 8-state identical Recursive Systematic Convolutional (RSC) encoders that are concatenated in parallel. The interleaver is used to scramble the bits and to provide different input data to each neural network. [Sattiraju, page 4, left col, lines 1-9] further supports that the RNN (recurrent neural network) is used to encode the binary bits and the LTE turbo encoder structure disclosed in the paragraph [Sattiraju, page 2, left col, line 19-26] is used to encoder the data)
wherein the plurality of paths comprise at least a first path ([Sattiraju, page 1, right col, line 31 - page 2, left col, line 2; Fig. 1] and [Sattiraju, page 2, left col, line 19-26] collectively discloses a LTE turbo encoder which includes Deep Learning encoders connected in parallel and separated by an interleaver. The interleaver is used to scramble the bits and to provide different input data to each neural network. [Sattiraju, page 4, left col, lines 1-9] further supports that the RNN (recurrent neural network) is used to encode the binary bits and the LTE turbo encoder structure disclosed in the paragraph [Sattiraju, page 2, left col, line 19-26] is used to encoder the data)
It would have been obvious to one of ordinary skill in the art before the effective filing date of
the claimed invention, having both the teachings of Zeng, Aragao and Sattiraju to use the method of generating separate input data for each encoder using an interleaver of Sattiraju to implement the machine learning based communication method of Zeng. The suggestion and/or motivation for doing so is to improve the accuracy of the machine learning based communication system by spreading burst errors into single bit errors which can be easily corrected through error correction codes.
However, Zeng in view of Aragao and further in view of Sattiraju do not specifically disclose:
wherein the first neural network encoder includes an interleaver, a recursive systematic convolutional (RSC) code, an accumulator, and a neural networks arranged in parallel
wherein the plurality of paths comprise at least a first path passing through the accumulator to provide first input data to a first neural network among the plurality of neural networks
Liva teaches:
wherein the first neural network encoder includes an interleaver, a recursive systematic convolutional (RSC) code, an accumulator, and a ([Liva, page 2203, left col, A. Parallel Concatenation, lines 1 – 36] and [page 2203, right col, Fig. 2, a) PCCC encoder structure] The parallel connection of PCCC encoder structure is called ‘tail-biting structure’ because the output of the encoder
p
i
-
1
(
1
)
is fed into the new input as shown in equation (3) to generate accumulated input
p
i
(
1
)
=
g
i
(
1
)
u
i
+
f
i
(
1
)
p
i
-
1
(
1
)
. An interleaver
π
is used to divide the input data for the first encoder
p
i
(
1
)
and the second encoder
p
i
(
2
)
. The information word u is input to a rate-1, memory-1 time-variant recursive systematic convolutional (RSC) tail-biting encoder)
wherein the plurality of paths comprise at least a first path passing through the accumulator to provide first input data to a first ([Liva, page 2203, left col, A. Parallel Concatenation, lines 1 – 36] and [page 2203, right col, Fig. 2, a) PCCC encoder structure] The parallel connection of PCCC encoder structure is called ‘tail-biting structure’ because the output of the encoder
p
i
-
1
(
1
)
is fed into the new input as shown in equation (3) to generate accumulated input
p
i
(
1
)
=
g
i
(
1
)
u
i
+
f
i
(
1
)
p
i
-
1
(
1
)
. An interleaver
π
is used to divide the input data for the first encoder
p
i
(
1
)
and the second encoder
p
i
(
2
)
)
It would have been obvious to one of ordinary skill in the art before the effective filing date of
the claimed invention, having both the teachings of Zeng, Aragao, Sattiraju and Liva to use the method of passing a first input to the accumulator to provide first input data to a first encoder of Liva to implement the machine learning based communication method of Zeng. The suggestion and/or motivation for doing so is to improve the efficiency of the signal transmission and encoding system by reducing redundant memory accesses by keeping partial results in a register rather than storing and retrieving
p
i
-
1
(
1
)
from a memory device.
Regarding Claim 3, Zeng teaches:
The method of claim 1, wherein the information informs of at least one of a type of a neural network, a number of layers of the neural network, an activation function for each of the layers, an optimization method for the neural network, or a weight for each of the layers. ([Zeng, 0003-0005; 0044-0045; 0050] collectively disclose the UE transmitting CSI to the Base Station (uplink) and the Base Station communicating with the UE on the downlink. [Zeng, 0067-0068] discloses the first network node (base station 105) providing new CSIRS-based CSI encoder parameters (first parameter) to the second neural network node (UE 115) for use in channel compression, and the first network node sending the decoder parameter (second parameter) to the second network node)
Regarding Claim 4, Zeng teaches:
The method of claim 3, wherein the weight is defined in advance. ([Zeng, 0003-0005; 0044-0045; 0050] collectively disclose the UE transmitting CSI to the Base Station (uplink) and the Base Station communicating with the UE on the downlink. [Zeng, 0067-0068] discloses the first network node (base station 105) providing new (defined in advance in the first network node) CSIRS-based CSI encoder parameters (first parameter) to the second neural network node (UE 115) for use in channel compression, and the first network node sending the decoder parameter (second parameter) to the second network node)
Regarding Claim 5, Zeng teaches:
The method of claim 1, wherein the base station includes a second neural network encoder and a second neural network decoder composed of a neural network. ([Zeng, 0091-0092; Fig. 7; Fig. 8] discloses the Base Station 105 comprising CSI Encoder 704 (second neural network encoder), CSI decoder 702 (second neural network decoder), Encoder/Decoder parameter generation 705, Encoder/Decoder parameters 706, and the UE 115 including CSI Encoder 804 and CSI Decoder 802, Encoder/Decoder parameters 806, and a parameter generation 805. [Zeng, 0059 and 0088] collectively discloses that the CSI encoder and the CSI decoder are implemented using neural networks)
Regarding Claim 12, Zeng teaches:
The method of claim 1, wherein the first parameter and the second parameter are generated based on training performed by the base station. ([Zeng, 0067-0068] discloses the first network node (base station 105) having RS observations which is used to train instances of CSI encoder and CSI decoder. The trained CSIRS-based encoder parameters and the CSI decoder parameters are sent to the second network node)
Regarding Claim 13, Zeng teaches:
The method of claim 1, wherein the first parameter and the second parameter are generated by a training device, and wherein the UE receives the first parameter and the second parameter transmitted to the base station by the training device from the base station. ([Zeng, 0067-0068] discloses the first network node (base station 105) having RS observations which is used to train instances of CSI encoder and CSI decoder. The trained CSIRS-based encoder parameters and the CSI decoder parameters are sent to the second network node)
Regarding Claim 14, Zeng teaches:
The method of claim 1, wherein the information comprises at least one of a transmission-related weight and a reception-related weight. ([Zeng, 0067-0068] discloses the first network node (base station 105) providing new CSIRS-based CSI encoder parameters (transmission related weight) to the second neural network node (UE 115) for use in channel compression, and the first network node sending the decoder parameter (reception related weight) to the second network node. [Zeng, 0059] The CSI encoder provides channel compression (transmission) and the CSI decoder provides channel decompression (reception) )
Regarding Claim 15, Zeng teaches:
A user equipment (UE) comprising: at least one memory; at least one transceiver; and at least one processor operably connectable to the at least one memory and the at least one transceiver, wherein the at least one memory stores instructions that, based on being executed by the at least one processor, cause the at least one processor to perform operations comprising: ([Zeng, 0064] The UE and the Base Station comprises a network node transmit processor, storage memories, and/or other circuitry such as controller/processor that provides one or more functions of CSI feedback. [Zeng, 0092; Fig. 8] discloses the UE comprises CSI Encoder 804 and CSI Decoder 802, Encoder/Decoder parameters 806, and a parameter generation 805. [Zeng, 0059 and 0088] collectively discloses that the CSI encoder and the CSI decoder are implemented using neural networks)
Claim 15 is an apparatus claim which implements the same features as the method claim 1, and is rejected for at least the same reasons.
Claim 2 is rejected under 35 U.S.C. 103 as being unpatentable over Zeng in view of Aragao in view of Sattiraju in view of Liva and further in view of Chen et al. (US 20230084164 A1, hereinafter ‘Chen’)
Regarding Claim 2, Zeng in view of Aragao in view of Sattiraju and further in view of Liva teaches:
The method of claim 1.
However, Zeng in view of Aragao in view of Sattiraju and further in view of Liva do not specifically disclose:
wherein the information is transmitted based on radio resource control (RRC) signaling, medium access control (MAC) signaling or layer 1 (L1) signaling.
Chen teaches:
wherein the information is transmitted based on radio resource control (RRC) signaling, medium access control (MAC) signaling or layer 1 (L1) signaling. ([Chen, 0086] discloses the UE utilizes media access control-control element MAC-CE or radio resource control (RRC) signaling)
It would have been obvious to one of ordinary skill in the art before the effective filing date of
the claimed invention, having both the teachings of Zeng, Aragao, Sattiraju, Liva and Chen to use the method of transmitting information based on RRC signaling or MAC signaling of Chen to implement the machine learning based communication method of Zeng. The suggestion and/or motivation for doing so is to improve the security of the machine learning based communication system by verifying the sender's frame check sequences.
Response to Arguments
Response to Arguments under 35 U.S.C. 112
Amended claims were received on 06/12/2026. 35 U.S.C. 112 rejections have been withdrawn.
Response to Arguments under 35 U.S.C. 103
Arguments: Applicant asserts that (a) Zeng fails to teach the highly specific, internal structural topology of the neural network encoder as recited in the amended claim because the neural network in Zeng is treated as a black box [Remarks, pages 8-9], and (b) Sattiraju fails to remedy the deficiency of Zeng because the accumulator is a pre-encoding component and Sattiraju’s shift register is strictly a post-encoding component [Remarks, page 10].
Examiner’s Response: First, the examiner agrees that Zeng fails to teach the internal structural topology of the neural network encoder recited in the amended claim 1 “wherein an input data sequence input to the first neural network encoder is branched into a plurality of paths to provide different input data respectively to the plurality of neural networks, and wherein the plurality of paths comprise at least a first path passing through the accumulator to provide first input data to a first neural network among the plurality of neural networks, and a second path passing through the interleaver to provide second input data or a second neural network among the plurality of neural networks.” However, secondary references Sattiraju and Liva remedy the deficiency of Zeng and it is obvious for someone who knows the art to introduce the parallel neural network encoder of Sattiraju and accumulator of Liva into Zeng to implement the present invention. Regarding the argument (b), the examiner introduced Liva. Liva teaches accumulating encoder output prior to the encoder.
Therefore, 35 U.S.C. 103 rejections for claims 1-5 and 12-15 are maintained.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Jiang et al., “Turbo Autoencoder: Deep learning based channel codes for point-to-point communication channels” (This prior art is pertinent because it discloses using deep learning models to implement a Turbo Encoder)
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JUN KWON whose telephone number is (571)272-2072. The examiner can normally be reached Monday – Friday 8:00AM – 5:00PM ET.
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, Abdullah Kawsar can be reached at (571)270-3169. 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.
/JUN KWON/Examiner, Art Unit 2127
/ABDULLAH AL KAWSAR/Supervisory Patent Examiner, Art Unit 2127
Prosecution Insights