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 statements (IDS) submitted on 8/27/2024 and 5/5/2026 are in compliance with the provisions of 37 CFR 1.97. Accordingly, the information disclosure statements
are being considered by the examiner.
Claim Objections
Claim 1, 8, 18 and 23 are objected to because of the following informalities:
In claim 1, lines 9-10, “one or more encoders” should read “the one or more encoders”.
In claim 8, line 10, “one or more encoders” should read “the one or more encoders”.
In claim 18, line 9, “one or more decoders” should read “the one or more decoders”.
In claim 23, line 9, “one or more decoders” should read “the one or more decoders”.
Appropriate correction is required.
Claim Rejections - 35 USC § 102
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claim(s) 1-3, 6-10, 16-20, 23-24 and 28 is/are rejected under 35 U.S.C. 102 (a)(2) as being anticipated by WO 2023113677 A1 (hereinafter Timo) (priority document US 63265417 20211215, hereinafter prov5417).
Regarding claim 1, Timo teaches A apparatus of wireless communication for a user
equipment (UE) vendor, comprising (Timo First node 601 in Fig. 6; page 17 line 33 to page 18 line 3, a first node comprising an NN-based AE-encoder, for training the AE- encoder in a training phase of the AE-encoder. The AE-encoder is trained to provide encoded CSI, e.g., from a first communications node, such as a UE, to a second communications node, such as a radio access node, over a communications channel in a communications network. The communications channel may be a wireless communications channel (prov5417 page 17 lines 28-33; Fig. 6).):
a transceiver (Timo page 33 lines 17-20, The first node 601 and the second node 602 may comprise a respective input and output interface, IF, 1006, 1106 configured to communicate with each other, see Figures 10-11. The input and output interface may comprise a wireless receiver (not shown) and a wireless transmitter (not shown) (prov5417 page 31 line 33 to page 32 line 2; Fig. 10-11).);
a memory configured to store instructions (Timo memory 1002 in Fig. 10; page 35 lines 23-25,
The first node 601 may further comprise a memory 1002. The memory comprises instructions executable by the processor in the first node 601 (prov5417 page 32 lines 30-32; Fig. 10).); and
one or more processors communicatively coupled with the transceiver and the memory, wherein the one or more processors are configured to execute the instructions to (Timo processor 1004 in Fig. 10; page 35, lines 30-34, a respective computer program 1003 and 1103 comprises instructions, which when executed by the processor 1004, 1104, cause the processor 1004, 1104 of the respective first node 601 and second node 602 to perform the actions above (prov5417 page 33 lines 1-5; Fig. 10).):
train one or more encoder-decoder pairs based on a raw dataset of the UE vendor (Timo page
11, lines 18-27, the AE training process is a highly iterative process that may be expensive - consuming significant time, compute, memory, and power resources. Therefore, it may be expected that AE architecture design and training will largely be performed offline, e.g., in a development environment, using appropriate compute infrastructure, training data, validation data, and test data. Data for training, validation, and testing may be collected from one or more of the following examples:
real measurements recorded in live networks,
synthetic radio channel data from, e.g., 3GPP channel models or ray tracing models and/or
digital twins, and
mobile drive tests (prov5417 page 11, lines 12-21)
Fig. 7; page 23, lines 18-34, The flow chart illustrates a computer-implemented method,
performed by the first node 601 for training the AE-encoder 601-1 in a training phase of the AE-encoder 601-1. As a first optional action 700 of Figure 7 the first node 601 provides the second node 602 with meta data associated with the AE. The meta data may comprise an indication of any one or more of: an indication to a reference AE-decoder architecture that the AE-encoder 601-1 has been pre-trained with (prov5417 page 22, lines 9-24; Fig. 7)
Note: AE-encoder 601-1 and a reference AE-decoder architecture corresponds to encoder-decoder pair.);
generate one or more training sets based on outputs of one or more encoders running the raw
dataset of the UE vendor (Timo page 24, lines 12-13, In action 702 the first node 601 computes, with the AE-encoder 601-1, encoder output data based on channel data e.g., training channel data (prov5417 page 23 lines 1-2).
page 24, lines 19-23, In action 703 the first node 601 provides AE-encoder data to the second node 602 comprising the NN-based AE-decoder 602-1. The AE-encoder data includes the encoder output data computed with the AE-encoder 601-1 (prov5417 page 23 lines 9-11).); and
communicate the one or more training sets to a network entity vendor (Timo page 24, lines 19-23, In action 703 the first node 601 provides AE-encoder data to the second node 602 comprising the NN-based AE-decoder 602-1. The AE-encoder data includes the encoder output data computed with the AE-encoder 601-1 (prov5417 page 23 lines 9-11).).
Regarding claim 8, Timo teaches A apparatus of wireless communication for a user equipment (UE) vendor, comprising (Timo First node 601 in Fig. 6; page 17 line 33 to page 18 line 3, a first node comprising an NN-based AE-encoder, for training the AE- encoder in a training phase of the AE-encoder. The AE-encoder is trained to provide encoded CSI, e.g., from a first communications node, such as a UE, to a second communications node, such as a radio access node, over a communications channel in a communications network. The communications channel may be a wireless communications channel (prov5417 page 17 lines 28-33; Fig. 6).):
a transceiver (Timo page 33 lines 17-20, The first node 601 and the second node 602 may comprise a respective input and output interface, IF, 1006, 1106 configured to communicate with each other, see Figures 10-11. The input and output interface may comprise a wireless receiver (not shown) and a wireless transmitter (not shown) (prov5417 page 31 line 33 to page 32 line 2; Fig. 10-11).);
a memory configured to store instructions (Timo memory 1002 in Fig. 10; page 35 lines 23-25,
The first node 601 may further comprise a memory 1002. The memory comprises instructions executable by the processor in the first node 601 (prov5417 page 32 lines 30-32; Fig. 10).); and
one or more processors communicatively coupled with the transceiver and the memory, wherein the one or more processors are configured to execute the instructions to (Timo processor 1004 in Fig. 10; page 35, lines 30-34, a respective computer program 1003 and 1103 comprises instructions, which when executed by the processor 1004, 1104, cause the processor 1004, 1104 of the respective first node 601 and second node 602 to perform the actions above (prov5417 page 33 lines 1-5; Fig. 10).):
train one or more encoder-decoder pairs based on a raw dataset of the UE vendor (Timo page
11, lines 18-27, the AE training process is a highly iterative process that may be expensive - consuming significant time, compute, memory, and power resources. Therefore, it may be expected that AE architecture design and training will largely be performed offline, e.g., in a development environment, using appropriate compute infrastructure, training data, validation data, and test data. Data for training, validation, and testing may be collected from one or more of the following examples:
real measurements recorded in live networks,
synthetic radio channel data from, e.g., 3GPP channel models or ray tracing models and/or
digital twins, and
mobile drive tests (prov5417 page 11, lines 12-21)
Fig. 7; page 23, lines 18-34, The flow chart illustrates a computer-implemented method,
performed by the first node 601 for training the AE-encoder 601-1 in a training phase of the AE-encoder 601-1. As a first optional action 700 of Figure 7 the first node 601 provides the second node 602 with meta data associated with the AE. The meta data may comprise an indication of any one or more of: an indication to a reference AE-decoder architecture that the AE-encoder 601-1 has been pre-trained with (prov5417 page 22, lines 9-24; Fig. 7)
Note: AE-encoder 601-1 and a reference AE-decoder architecture corresponds to encoder-decoder pair.);
generate two training sets for each of the one or more encoder-decoder pairs based on outputs of one or more encoders running the raw dataset of the UE vendor (Timo page 24, lines 12-13, In action 702 the first node 601 computes, with the AE-encoder 601-1, encoder output data based on channel data e.g., training channel data (prov5417 page 23 lines 1-2).
page 24, lines 19-23, In action 703 the first node 601 provides AE-encoder data to the second node 602 comprising the NN-based AE-decoder 602-1. The AE-encoder data includes the encoder output data computed with the AE-encoder 601-1 (prov5417 page 23 lines 9-11).
page 26 line 35 to page 27 line 10, The AE-encoder training may include the following:
The first node 601 sends a batch of AE-encoder data to the second node 602, where the batch of AE-encoder data includes a batch of output data from the AE-encoder 601-1.
The first node 601 receives AE-encoder update assistance information from the second node 602.
The first node 601 updates trainable AE-encoder parameters and repeats the two above steps until a certain pass/fail criterion is fulfilled (prov5417 page 15, lines 15-25).
Note: iterative training corresponds to generating and communicating two training sets); and
communicate the two training sets to a network entity vendor (Timo page 24, lines 19-23, In action 703 the first node 601 provides AE-encoder data to the second node 602 comprising the NN-based AE-decoder 602-1. The AE-encoder data includes the encoder output data computed with the AE-encoder 601-1 (prov5417 page 23 lines 9-11).
page 26 line 35 to page 27 line 10, The AE-encoder training may include the following:
The first node 601 sends a batch of AE-encoder data to the second node 602, where the batch of AE-encoder data includes a batch of output data from the AE-encoder 601-1.
The first node 601 receives AE-encoder update assistance information from the second node 602.
The first node 601 updates trainable AE-encoder parameters and repeats the two above steps until a certain pass/fail criterion is fulfilled (prov5417 page 15, lines 15-25).
Note: iterative training corresponds to generating and communicating two training sets.).
Regarding claim 18, Timo teaches A apparatus of wireless communication for a network entity
vendor, comprising (Timo Fig. 6, Second node 602; page 18 lines 6-10, providing AE-encoder data to a second node comprising a NN-based AE-decoder and having access to the channel data representing a communications channel between a first communications node and a second communications node, wherein the AEencoder data includes encoder output data computed with the AE-encoder based on the channel data (prov5417 page 18 lines 1-6).):
a transceiver (Timo page 33 lines 17-20, The first node 601 and the second node 602 may comprise a respective input and output interface, IF, 1006, 1106 configured to communicate with each other, see Figures 10-11. The input and output interface may comprise a wireless receiver (not shown) and a wireless transmitter (not shown) (prov5417 page 31 line 33 to page 32 line 2; Fig. 10-11).);
a memory configured to store instructions (Timo memory 1102 in Fig. 11; page 35 lines 23-25, The second node 602 may further comprise a memory 1102. The memory comprises instructions executable by the processor in the second node 602 (prov5417 page 32 lines 30-32; Fig. 11).); and
one or more processors communicatively coupled with the transceiver and the memory, wherein the one or more processors are configured to execute the instructions to (Timo processor 1104 in Fig. 11; page 35, lines 30-34, a respective computer program 1003 and 1103 comprises instructions, which when executed by the processor 1004, 1104, cause the processor 1004, 1104 of the respective first node 601 and second node 602 to perform the actions above (prov5417 page 33 lines 1-5, Fig. 11). ):
receive one or more training sets from a UE vendor, the one or more training sets corresponding to one or more encoder-decoder pairs (Timo Fig. 9; page 31, lines 23-27, The flow chart illustrates a computer-implemented method, performed by the second node 602 comprising the NN-based AE-decoder 602-1, for assisting in training the NN-based AEencoder 601-1 comprised in the first node 601, in the training phase of the AE-encoder 601-1 (prov5417 page 30 lines 5-9; Fig. 9)
page 31, lines 34-35, As a first optional action 900 of Figure 9 the second node 602 receives, from the first node 601, meta data associated with the AE (prov5417 page 30 lines 16-17).
page 23, lines 24-34, The meta data may comprise an indication of any one or more of: an indication to a reference AE-decoder architecture that the AE-encoder 601-has been pre-trained with (prov5417 page 22 lines 13-24).
Note: AE-encoder 601-1 and a reference AE-decoder architecture corresponds to encoder-decoder pair.
page 32, lines 7-10, In an optional action 902 the second node 602 selects an appropriate AE-decoder to be used for the training of the AE-encoder, based on the indication of the type of
encoder and/or decoder type and/or architecture. For example, the second node 602 may select an AE-decoder 602-1 that is compatible with the AE-encoder 601-1 (prov5417 page 30 lines 24-27)
page 32, lines 17-18, In action 903 the second node 602 receives AE-encoder data from the first node 601, wherein the AE-encoder data includes encoder output data (prov5417 page 30 lines 34-35).); and
train one or more decoders associated with the one or more training sets (Timo page 32, lines 20-23, In action 904 the second node 602 computes training assistance information based on the encoder output data and based on channel data used by the AE-encoder 601-1 in the first node 601 to compute the encoder output data (prov5417 page 31 lines 1-3).
page 24, lines 30-35, The training assistance information may comprise one or more of:
a gradient vector of a loss function computed by the second node 602 with respect to a respective encoder parameter of the AE-encoder 601-1, a loss value of the loss function, an indication of the loss, an indication of whether or not the AE-encoder 601-1 has achieved sufficient training performance on the shared channel data when used with the 35 AE-decoder 602-1 such that a pass criterion is fulfilled. The loss may quantify a reconstruction error of the shared channel data (prov5417 page 23 lines 18-24).
Note: computing training assistance information corresponds to training AE-decoder 602-1.).
Regarding claim 23, Timo teaches A apparatus of wireless communication for a network entity vendor, comprising (Timo Fig. 6, Second node 602; page 18 lines 6-10, providing AE-encoder data to a second node comprising a NN-based AE-decoder and having access to the channel data representing a communications channel between a first communications node and a second communications node, wherein the AEencoder data includes encoder output data computed with the AE-encoder based on the channel data (prov5417 page 18 lines 1-6).):
a transceiver (Timo page 33 lines 17-20, The first node 601 and the second node 602 may comprise a respective input and output interface, IF, 1006, 1106 configured to communicate with each other, see Figures 10-11. The input and output interface may comprise a wireless receiver (not shown) and a wireless transmitter (not shown) (prov5417 page 31 line 33 to page 32 line 2; Fig. 10-11).);
a memory configured to store instructions (Timo memory 1102 in Fig. 11; page 35 lines 23-25, The second node 602 may further comprise a memory 1102. The memory comprises instructions executable by the processor in the second node 602 (prov5417 page 32 lines 30-32; Fig. 11).); and
one or more processors communicatively coupled with the transceiver and the memory, wherein the one or more processors are configured to execute the instructions to (Timo processor 1104 in Fig. 11; page 35, lines 30-34, a respective computer program 1003 and 1103 comprises instructions, which when executed by the processor 1004, 1104, cause the processor 1004, 1104 of the respective first node 601 and second node 602 to perform the actions above (prov5417 page 33 lines 1-5, Fig. 11).):
receive two training sets from a UE vendor, the two training sets corresponding to one or more encoder-decoder pairs (Timo Fig. 9; page 31, lines 23-27, The flow chart illustrates a computer-implemented method, performed by the second node 602 comprising the NN-based AE-decoder 602-1, for assisting in training the NN-based AEencoder 601-1 comprised in the first node 601, in the training phase of the AE-encoder 601-1 (prov5417 page 30 lines 5-9; Fig. 9)
page 31, lines 34-35, As a first optional action 900 of Figure 9 the second node 602 receives, from the first node 601, meta data associated with the AE (prov5417 page 30 lines 16-17).
page 23, lines 24-34, The meta data may comprise an indication of any one or more of: an indication to a reference AE-decoder architecture that the AE-encoder 601-has been pre-trained with (prov5417 page 22 lines 13-24).
Note: AE-encoder 601-1 and a reference AE-decoder architecture corresponds to encoder-decoder pair.
page 32, lines 7-10, In an optional action 902 the second node 602 selects an appropriate AE-decoder to be used for the training of the AE-encoder, based on the indication of the type of
encoder and/or decoder type and/or architecture. For example, the second node 602 may select an AE-decoder 602-1 that is compatible with the AE-encoder 601-1 (prov5417 page 30 lines 24-27)
page 32, lines 17-18, In action 903 the second node 602 receives AE-encoder data from the first node 601, wherein the AE-encoder data includes encoder output data (prov5417 page 30 lines 34-35).
page 26 line 35 to page 27 line 10, The AE-encoder training may include the following:
The first node 601 sends a batch of AE-encoder data to the second node 602, where the batch of AE-encoder data includes a batch of output data from the AE-encoder 601-1.
The first node 601 receives AE-encoder update assistance information from the second node 602. This information is used to update the AEencoder trainable parameters, e.g., the training information may be a gradient vector, a loss value, or other useful state information about the AE-decoder.
The first node 601 updates trainable AE-encoder parameters and repeats the two above steps until a certain pass/fail criterion is fulfilled (prov5417 page 15, lines 15-25).
Note: iterative training corresponds to receiving two training sets.); and
train one or more decoders associated with the two training sets (Timo page 32, lines 20-23, In action 904 the second node 602 computes training assistance information based on the encoder output data and based on channel data used by the AE-encoder 601-1 in the first node 601 to compute the encoder output data (prov5417 page 31 lines 1-3).
page 24, lines 30-35, The training assistance information may comprise one or more of:
a gradient vector of a loss function computed by the second node 602 with respect to a respective encoder parameter of the AE-encoder 601-1, a loss value of the loss function, an indication of the loss, an indication of whether or not the AE-encoder 601-1 has achieved sufficient training performance on the shared channel data when used with the 35 AE-decoder 602-1 such that a pass criterion is fulfilled. The loss may quantify a reconstruction error of the shared channel data (prov5417 page 23 lines 18-24).
Note: computing training assistance information corresponds to training AE-decoder 602-1.
page 26 line 35 to page 27 line 10, The AE-encoder training may include the following:
The first node 601 sends a batch of AE-encoder data to the second node 602, where the batch of AE-encoder data includes a batch of output data from the AE-encoder 601-1.
The first node 601 receives AE-encoder update assistance information from the second node 602. This information is used to update the AEencoder trainable parameters, e.g., the training information may be a gradient vector, a loss value, or other useful state information about the AE-decoder.
The first node 601 updates trainable AE-encoder parameters and repeats the two above steps until a certain pass/fail criterion is fulfilled (prov5417 page 15, lines 15-25).
Note: iterative training corresponds to two training sets.).
Regarding claim 2, Timo teaches The apparatus of claim 1.
Timo teaches wherein each of the one or more training sets include an encoder identification
(ID), encoder output, and desired decoder output (Timo page 23, lines 18-34, The flow chart illustrates a computer-implemented method, performed by the first node 601 for training the AE-encoder 601-1 in a training phase of the AE-encoder 601-1.
As a first optional action 700 of Figure 7 the first node 601 provides the second node 602 with meta data associated with the AE. The meta data may comprise an indication of any one or more of: an AE-encoder type for the AE-encoder 601-1. The AE-encoder type for the AE-encoder 601-1
may refer to an architecture (prov5417 page 22, lines 9-24).
page 24, lines 19-24, In action 703 the first node 601 provides AE-encoder data to the second node 602 comprising the NN-based AE-decoder 602-1 ...The AE-encoder data includes the
encoder output data computed with the AE-encoder 601-1. The AE-encoder data may further include the channel data (prov5417 page 23, lines 9-12).
Note: the channel data is the desired decoder output.).
Claim 10, 19 and 24 recites similar limitations of claim 2 respectively, are thus rejected under similar rational.
Regarding claim 3, Timo teaches The apparatus of claim 1.
Timo teaches wherein each of the one or more training sets is associated with one or more
metadata to enable switching between one or more models during inference (Timo page 23, lines 18-34, The flow chart illustrates a computer-implemented method, performed by the first node 601 for training the AE-encoder 601-1 in a training phase of the AE-encoder 601-1.
As a first optional action 700 of Figure 7 the first node 601 provides the second node 602 with meta data associated with the AE. The meta data may comprise an indication of any one or more of:
a number of AE nodes in the output layer Y of the AE-encoder (prov5417 page 22, lines 9-24)
page 11, lines 2-8, The training process, however, typically involves optimizing many other parameters, e.g., higher-level hyperparameters that define the model or the training process. Some example hyperparameters are as follows:
• The architecture of the AE, e.g., dense, convolutional, transformer.
• Architecture-specific parameters, e.g., the number of nodes per layer in a dense network, or the kernel sizes of a convolutional network.
Note: meta data (number of nodes in the encoder etc) defines the training model.).
Claim 20 and 28 recites similar limitations of claim 3 respectively, are thus rejected under similar rational.
Regarding claim 6, Timo teaches The apparatus of claim 1.
Timo teaches wherein an encoder output of the one or more encoders corresponds to a
compressed channel status information (CSI) feedback (CSF) message (Timo page 8, lines 19-24, AE encoder to compress the estimated channel …down to a binary codeword. The binary codeword is
reported to the network over an uplink control channel and/or data channel. In practice, this codeword will likely form one part of a channel state information (CS/) report (prov5417 page 8, lines 9-14)).
Claim 16 recites similar limitations of claim 6, is thus rejected under similar rational.
Regarding claim 7, Timo teaches The apparatus of claim 1.
Timo teaches wherein a decoder output of one or more decoders of the one or more encoder-
decoder pairs includes a reconstructed CSF corresponding to one or more precoding vectors (Timo page 8, lines 25-28; AE decoder to reconstruct the estimated channel. The decompressed output of the AE decoder is used by the network in, for example, MIMO preceding (prov5417 page 8 lines 15-17)).
Claim 17 recites similar limitations of claim 7, is thus rejected under similar rational.
Regarding claim 9, Timo teaches The apparatus of claim 8.
Timo teaches wherein the one or more processors configured to generate the two training sets
(Timo page 24, lines 12-13; page 24, lines 19-23; page 26 line 35 to page 27 line 10; cited above in rejection of claim 8.).
Timo teaches configured to generate a first training set based on outputs of the encoder and
decoder using the raw dataset of the UE vendor (Timo Timo page 11, lines 18-27, the AE training process is a highly iterative process that may be expensive - consuming significant time, compute, memory, and power resources. Therefore, it may be expected that AE architecture design and training will largely be performed offline, e.g., in a development environment, using appropriate compute infrastructure, training data, validation data, and test data. Data for training, validation, and testing may be collected from one or more of the following examples:
real measurements recorded in live networks,
synthetic radio channel data from, e.g., 3GPP channel models or ray tracing models and/or
digital twins, and
mobile drive tests (prov5417 page 11, lines 12-21)
Fig. 7; page 23, lines 18-34, The flow chart illustrates a computer-implemented method,
performed by the first node 601 for training the AE-encoder 601-1 in a training phase of the AE-encoder 601-1. As a first optional action 700 of Figure 7 the first node 601 provides the second node 602 with meta data associated with the AE. The meta data may comprise an indication of any one or more of: an indication to a reference AE-decoder architecture that the AE-encoder 601-1 has been pre-trained with (prov5417 page 22, lines 9-24; Fig. 7)
Note: AE-encoder 601-1 and a reference AE-decoder architecture corresponds to encoder-decoder pair.
page 24, lines 12-13, In action 702 the first node 601 computes, with the AE-encoder 601-1, encoder output data based on channel data e.g., training channel data (prov5417 page 23 lines 1-2).
Page 24, lines 19-23, In action 703 the first node 601 provides AE-encoder data to the second node 602 comprising the NN-based AE-decoder 602-1. The AE-encoder data includes the encoder output data computed with the AE-encoder 601-1 (prov5417 page 23 lines 9-11).).
Claim Rejections - 35 USC § 103
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
The factual inquiries for establishing a background for determining obviousness under 35 U.S.C. 103 are summarized as follows:
1. Determining the scope and contents of the prior art.
2. Ascertaining the differences between the prior art and the claims at issue.
3. Resolving the level of ordinary skill in the pertinent art.
4. Considering objective evidence present in the application indicating obviousness or nonobviousness.
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.
Claim(s) 4, 21 and 29 is/are rejected under 35 U.S.C. 103 as being unpatentable over Timo as applied to claims 3, 20 and 28 above, and further in view of WO2023151989A1 (hereinafter Alabbasi) (priority document US 63308755 2022-02-10, hereinafter prov8755).
Regarding claim 4, Timo teaches The apparatus of claim 3.
Timo teaches the one or more metadata (See rejection of claim 3 above), Timo does not
explicitly teach metadata includes at least one of UE antenna configuration, signal-to-noise ratio (SNR), Reference Signal Receive Power (RSRP), delay speed, average delay, and time stamp.
Alabbasi in the same or similar field of endeavor teaches metadata includes at least one of UE antenna configuration, signal-to-noise ratio (SNR), Reference Signal Receive Power (RSRP), delay speed, average delay, and time stamp (Alabbasi [0151] the type of metadata required, e.g. cell resource utilization, UE RSRP cell measurements, etc. … Specific conditions to be used for the generation of metadata, e.g. specific radio conditions, e.g. assumed pathloss, or specific load conditions, e.g. PRB utilization (prov 8755 [0150]).).
By modifying Timo’s teachings of the one or more metadata
with Alabbasi’s teachings of metadata includes at least one of UE antenna configuration, signal-to-noise ratio (SNR), Reference Signal Receive Power (RSRP), delay speed, average delay, and time stamp,
the modification results in
wherein the one or more metadata includes at least one of UE antenna configuration, signal-to-noise ratio (SNR), Reference Signal Receive Power (RSRP), delay speed, average delay, and time stamp.
It would have been prima facie obvious to one of ordinary skill in the art before the effective
filing date of the claimed invention to have modified Timo with Alabbasi’s above teachings. The motivation is improving network performance (Alabbasi [0081-0082] (prov8755 [0075-0076])).
Claim 21 and 29 recites similar limitations of claim 4 respectively, are thus rejected under similar rational.
Claim(s) 11-12 and 25 is/are rejected under 35 U.S.C. 103 as being unpatentable over Timo as applied to claims 9 and 23 above, and further in view of WO 2022090461 A1 (hereinafter Gilli).
Regarding claim 11, Timo teaches The apparatus of claim 9.
Although Timo teaches wherein the one or more processors configured to generate the two
training sets are further configured to generate a second training set (Timo page 24, lines 12-13; page 24, lines 19-23; page 26 line 35 to page 27 line 10; cited above in rejection of claim 8.), Timo does not explicitly teach based on perturbing the encoder output in the first training set by a vector and computing a corresponding decoder output.
Gilli in the same or similar field of endeavor teaches based on perturbing the encoder output in the first training set by a vector and computing a corresponding decoder output (Gilli Fig. 2; page 6, lines 1-7, A variational autoencoder is associated to the machine learning model and includes an encoder module 20 coupled to the processing means 12 to receive in input both the input query data 14 and the corresponding output prediction data 16. The encoder module 20 is configured to generate latent space data z 22 that depends on both the input query data and the output prediction data. The latent space data z are provided as input to a decoder module 24, a further input of the decoder module being coupled to the output prediction data 16. The decoder module 24 is configured to generate synthetic input query data X’ 26.
Note: z the encoder output in the first training set.
Fig. 5; page 8, lines 23-26, Starting from the encoding data z, a number of neighboring points (synthetic perturbations) Z’ = z+dZ are generated by adopting a predetermined sampling strategy. d. The set of points (latent space synthetic perturbation data) Z’ is used to generate synthetic input query data X' through the decoder module 24.
Note: dZ is the perturbing vector. X’ is the corresponding decoder output.).
By modifying Timo’s teachings of wherein the one or more processors configured to generate the two training sets are further configured to generate a second training set
with Gilli’s teachings of based on perturbing the encoder output in the first training set by a vector and computing a corresponding decoder output, the modification results in
wherein the one or more processors configured to generate the two training sets are further configured to generate a second training set based on perturbing the encoder output in the first training set by a vector and computing a corresponding decoder output.
It would have been prima facie obvious to one of ordinary skill in the art before the effective
filing date of the claimed invention to have modified Timo with Gilli’s above teachings. The motivation is improving the deployment of machine learning models (Gilli page 3, lines 7-9).
Regarding claim 12, Timo in view of Gilli teaches The apparatus of claim 11.
Timo teaches wherein the second training set includes the encoder ID (Timo page 24, lines 12-
13; page 24, lines 19-23; page 26 line 35 to page 27 line 10; cited above in rejection of claim 8.
page 23, lines 18-34, The flow chart illustrates a computer-implemented method, performed by the first node 601 for training the AE-encoder 601-1 in a training phase of the AE-encoder 601-1.
As a first optional action 700 of Figure 7 the first node 601 provides the second node 602 with meta data associated with the AE. The meta data may comprise an indication of any one or more of: an AE-encoder type for the AE-encoder 601-1. The AE-encoder type for the AE-encoder 601-1
may refer to an architecture (prov5417 page 22, lines 9-24).), Timo does not explicitly
teach a combination of the encoder output and the vector, and the corresponding decoder output.
Gilli teaches a combination of the encoder output and the vector, and the corresponding decoder output (Gilli Fig. 5; page 8, lines 23-26, Starting from the encoding data z, a number of neighboring points (synthetic perturbations) Z’ = z+dZ are generated by adopting a predetermined sampling strategy. d. The set of points (latent space synthetic perturbation data) Z’ is used to generate synthetic input query data X' through the decoder module 24.
Note: Z’ = z+dZ is the combination of the encoder output and the vector. X’ is the corresponding decoder output).
By modifying Timo’s teachings of wherein the second training set includes the encoder ID
with Gilli’s teachings of a combination of the encoder output and the vector, and the
corresponding decoder output,
the modification results in
wherein the second training set includes the encoder ID, a combination of the encoder output
and the vector, and the corresponding decoder output.
It would have been prima facie obvious to one of ordinary skill in the art before the effective
filing date of the claimed invention to have modified Timo as modified by Gilli with Gilli’s above teachings. The motivation is improving the deployment of machine learning models (Gilli page 3, lines 7-9).
Claim 25 recites similar limitations of claim 12, is thus rejected under similar rational.
Claim(s) 13-15 and 26-27 is/are rejected under 35 U.S.C. 103 as being unpatentable over Timo as applied to claims 8 and 23 above, and further in view of US 20220245449 A1 (hereinafter Freitas).
Regarding claim 13, Timo teaches The apparatus of claim 8.
Although Timo teaches wherein the one or more processors are configured to: communicate the
identified decoder to the network entity vendor (Timo Fig. 7; page 23, lines 18-34; The flow chart illustrates a computer-implemented method, performed by the first node 601 for training the AE-encoder 601-1 in a training phase of the AE-encoder 601-1.
As a first optional action 700 of Figure 7 the first node 601 provides the second node 602 with meta data associated with the AE. The meta data may comprise an indication of any one or more of: an indication to a reference AE-decoder architecture that the AE-encoder 601-
has been pre-trained with (prov5417 page 22, lines 9-24)), Timo does not explicitly teach identify a substitute decoder that approximates the found decoder within one of the one or more encoder-decoder pairs.
Freitas in the same or similar field of endeavor teaches identify a substitute decoder that
approximates the found decoder within one of the one or more encoder-decoder pairs (Freitas [0045] This single non-symmetric decoder is constructed using a two-step training. In the first step, a set of symmetric encoder-decoder pairs is created using the state of art approach illustrated in Picture 2. In order to achieve various bitrates and quality levels, different Lagrangian multipliers values are set. These different values are chosen to cover the desired range of rate-distortion trade-offs. After training the symmetric models with the needed configurations, the trained neural network parameters (weights, biases, etc.) are saved to be used in the second step.
[0046] The second step is illustrated in Picture 5 for a single training iteration 501. The procedure illustrated in FIG. 5 is repeated for all media in the training dataset. The set of encoders 502 previously trained in the first step is frozen during the second training step. In this manner, the neural network parameters learned in the first step are not updated in the second step in order to make the decoder able to decode the latent representation produced by any of the previously trained encoders.... After several iterations, the decoder learns how to decode the latent representation generated from different encoders. After trained, the decoder (508) can reconstruct (509) the input independently of the latent representation, regardless of the encoded bitrate, and, therefore, can substitute the n symmetric decoders.
Note: the non-symmetric decoder (508) is the identified substitute decoder.).
By modifying Timo’s teachings of wherein the one or more processors are configured to: communicate the identified decoder to the network entity vendor
with Freitas’s teachings of identify a substitute decoder that approximates the found decoder within one of the one or more encoder-decoder pairs,
the modification results in
wherein the one or more processors are configured to: identify a substitute decoder that
approximates the found decoder within one of the one or more encoder-decoder pairs; and communicate the substitute decoder to the network entity vendor.
It would have been prima facie obvious to one of ordinary skill in the art before the effective
filing date of the claimed invention to have modified Timo with Freitas’s above teachings. The motivation is reducing number of model parameters (Freitas [0003]).
Claim 26 recites similar limitations of claim 13, is thus rejected under similar rational.
Regarding claim 14, Timo in view of Freitas teaches The apparatus of claim 13.
Timo does not explicitly teach wherein the substitute decoder produces at least a similar output
as the found decoder in response to a first training set and a second training set.
Freitas teaches wherein the substitute decoder produces at least a similar output as the found decoder in response to a first training set and a second training set (Timo [0046] The second step is illustrated in Picture 5 for a single training iteration 501. The procedure illustrated in FIG. 5 is repeated for all media in the training dataset. The set of encoders 502 previously trained in the first step is frozen during the second training step. In this manner, the neural network parameters learned in the first step are not updated in the second step in order to make the decoder able to decode the latent representation produced by any of the previously trained encoders.... After several iterations, the decoder learns how to decode the latent representation generated from different encoders. After trained, the decoder (508) can reconstruct (509) the input independently of the latent representation, regardless of the encoded bitrate, and, therefore, can substitute the n symmetric decoders.).
It would have been prima facie obvious to one of ordinary skill in the art before the effective
filing date of the claimed invention to have modified Timo as modified by Freitas with Freitas’s above teachings. The motivation is reducing number of model parameters (Freitas [0003]).
Claim 27 recites similar limitations of claim 14, is thus rejected under similar rational.
Regarding claim 15, Timo in view of Freitas teaches The apparatus of claim 13, wherein the one
or more processors configured to identify the substitute decoder (See rejection of claim 13 above)
Timo does not explicitly teach are further configured to identify a respective substitute decoder for each encoder ID corresponding to a respective encoder of the one or more encoder-decoder pairs.
Freitas teaches are further configured to identify a respective substitute decoder for each encoder ID corresponding to a respective encoder of the one or more encoder-decoder pairs (Freitas [0047] FIG. 6 is a flow diagram of an example process for training the non-symmetric codec. For convenience, the process will be assumed as being performed by a system of one or more computers located in one or more locations. For instance, a system implementation described with reference to Picture 2 and Picture 5, appropriately programmed in accordance with the present application. The system receives a training data (601), which can be any appropriate form of data, e.g., image, or video data. The system processes the data using multiple symmetric encoder-decoders pairs to generate multiple latent representations of data, e.g., at low (602), medium (603), and high (604) bitrates. After trained, the system discards the trained symmetric decoders (605, 606, 607) and makes the trained encoder weights constants (608, 609, 610). These fixed weights are used in a second training stage that performs the steps depicted in Picture 5. The system instantiates a new decoder (611). The system uses the training data as input in the fixed encoders to (605, 606, 607) to update the weights of the new instantiated decoder (612). After trained, the non-symmetric decoder can reconstruct different latent representations (613).
Note: three encoder-decoder pairs are shown in the Fig. 6.).
It would have been prima facie obvious to one of ordinary skill in the art before the effective
filing date of the claimed invention to have modified Timo as modified by Freitas with Freitas’s above teachings. The motivation is reducing number of model parameters (Freitas [0003]).
Allowable Subject Matter
Claims 5, 22 and 30 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
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
Timo et al. US20250047346 Communications nodes and methods for proprietary machine learning-based csi reporting.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to David Z Sun whose telephone number is (571)270-0750. The examiner can normally be reached Monday-Friday 0800am-0500pm.
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, Moo Jeong can be reached at 571-272-9617. 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.
/D.Z.S./Examiner, Art Unit 2418
/Moo Jeong/Supervisory Patent Examiner, Art Unit 2418
Prosecution Insights