DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
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.
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 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.
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 – 20 are rejected under 35 U.S.C. 103 as being unpatentable over Singh et al (US Patent Application Publication 2024/0129941, previously presented) in view of Singh et al (US Patent Application Publication 2024/0152820, previously presented), and further in view of Tehrani et al (US Patent Application Publication 2024/0291629). Hereinafter Singh941, Singh820, and Tehrani.
Regarding claim 1, Singh941 discloses a method, comprising:
receiving, by a device (network node or network entity called CPO (Computer Predictor and Orchestrator), load data identifying a load on a radio access network (RAN) (the network node (CPO) acquires a predicted future computational load of baseband processing for each selected radio cell of a plurality of radio cells, where the acquiring operation is an operation of receiving predicted future computational load of baseband processing for each selected radio cell of a plurality of radio cells, paragraphs [0097], [0108]);
selecting, by the device, one or more time series forecasting models and a classification model based on seasonality metrics associated with the load data (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; the CPO determines the radio cell based on the predicted future computational load of baseband processing and computation capabilities (i.e. selecting time series forecasting models and a classification model));
processing, by the device, the load data, with the one or more time series forecasting models, to forecast a capacity for the RAN (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; the CPO processes the load data to predict future computational load); and
selectively:
adjusting, by the device and based on the capacity exceeds the capacity threshold, the capacity to generate an adjusted capacity (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; i.e. adjust the future computational load to not exceed the computation capabilities), and
performing, by the device, one or more actions based on the adjusted capacity (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; i.e. perform the adjustment to the future computational load to not exceed the computation capabilities).
However, Singh941 does not explicitly disclose “processing, by the device, the load data and the capacity, with the classification model, to determine a binary classification outcome indicating whether the capacity exceeds a capacity threshold; and selectively: determining, by the device, that the RAN does not need an upgrade based on the binary classification outcome indicating that the capacity fails to exceed the capacity threshold; or adjusting, by the device and based on the binary classification outcome indicating that the capacity exceeds the capacity threshold.”
Singh820 discloses “processing, by the device, the load data and the capacity, with the classification model, to determine whether the capacity exceeds a capacity threshold; and selectively: determining, by the device, that the RAN does not need an upgrade based on determining that the capacity fails to exceed the capacity threshold” as the apparatus receives the characteristics data related to the at least one cell from the RAN node or the controller platform, determines whether there is a distribution shift related to the at least one temporal characteristic of the at least one cell based on the received data, selects a learning type for an update to a model in response to the distribution shift, and updates the model for the RAN to use updated model to perform at least one action to optimize the performance of the RAN node or at least one other RAN node based on the selected learning type (paragraph [0118]), where the performance metric KPI (key performance indicator) is used to identify the distribution shift along with additional attributes including (i) a threshold of the evaluated formula/function performance metric over which shift is detected and the model updated (ii) a type of characteristic based on which distribution shift is detected, including at least one of average, maximum, a given percentile for confidence interval, trends, peak to average ratio, standard deviation and higher moments (iii) a list of underlying input KPIs of the ML model, including at least one of number of connected users, number of active users, number of bearers, DL or UL PRB utilization, PDCCH utilization, PUCCH utilization, composite available capacity, total data delivered or received at a RAN node etc., (iv) a complexity constraint measure denoting how often the ML model can be updated (v) a data storage constraint measure denoting how much data can be stored for updating the ML model (vi) a preference for passive (continuous learning without drift detection) or active learning (based on a characteristic of drift detected) model update (paragraph [0087]); and the DSLM (distribution shift learning module) triggers a training update of the ML model based on a criterion (e.g. if performance/drift exceeds a threshold) (paragraph [0089]).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Singh941 and Singh820 before him or her, to incorporate the distribution shift determination, learning type selection, and model update as taught by Singh820, to improve the CPO network node of Singh941 for the motivation of optimizing the performance of the radio access network node or at least one other radio access network node (paragraph [0003] of Singh820).
While Singh941 and Singh820 teaches using forecasting model and classification model, they do not explicitly teach “a binary classification outcome indicating whether the capacity exceeds a capacity threshold.”
Tehrani discloses the system uses a variety of different ML-based approaches to dynamically control cell states, where the predictive modeling (i.e. a binary classification for determining whether a future traffic load will be above or below a threshold level) uses offline data and supervised learning to train the neural network to predict a future traffic load based on current traffic load measurements, where the data includes input parameters used directly or indirectly to generate the state vector, and output parameters used directly or indirectly to generate the action (paragraphs [0045] – [0051]).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Singh941, Singh820, and Tehrani before him or her, to incorporate the binary classification as taught by Tehrani, to improve the modified CPO network node of Singh941-Singh820 for determining the binary classification outcome of failing to exceed the threshold (below threshold) or exceeding the threshold (above threshold). The motivation for doing so would have been to reduce network capacity, improve network reliability/performance, minimize operational costs, and/or maximize operating profits (paragraph [0067] of Tehrani).
Regarding claim 2, Singh941, Singh820, and Tehrani disclose the method of claim 1, Singh941 discloses further comprising:
utilizing RAN features and a focus feature engineering technique to improve an accuracy of the classification model relative to a classification model not trained with the RAN features and the focus feature engineering technique (the CPO receives different requested data that are monitored on each cell of interest and reported to the CPO for the machine learning CP (Compute Predictor) module and the CS (Compute Scheduler) module, including cell throughput and other KPI (key performance indicators), list of key features enabled on the cell, etc., paragraphs [0077] – [0086]).
Regarding claim 3, Singh941, Singh820, and Tehrani disclose the method of claim 2, Singh941 discloses wherein the RAN features include seasonality patterns associated with the load on the RAN (the CP module contains four time series prediction/estimation model building blocks, where the foreseen time series prediction/estimation model building blocks are de-trending and de-seasonalizing to make the time series stationary and then, standard autoregressive moving average (ARMA) model can be fitted on the residual time series (“1”), additional features which the time series may depend on can be helpful for load prediction estimation (“2”), recurrent neural network (RNN)/long short term memory (LSTM) based neural network forecasting could be helpful in long term trends of load prediction/estimation (“3”), and multiple load time series from neighboring cells as features (“4”), paragraphs [0137] – [0142]).
Regarding claim 4, Singh941, Singh820, and Tehrani disclose the method of claim 1, Singh941 discloses wherein adjusting the capacity to generate the adjusted capacity comprises:
utilizing scaling or a combination of scaling and rotational stitching to adjust the capacity and generate the adjusted capacity (the CP uses a different ML/DL model based on the different prediction/estimate time length for which load estimate should be valid and also timescale, paragraph [0155]).
Regarding claim 5, Singh941, Singh820, and Tehrani disclose the method of claim 4, but Singh941 does not explicitly disclose wherein the rotational stitching limits adjustment of the capacity to a predefined rotational limit.
Singh820 discloses the apparatus receives the characteristics data related to the at least one cell from the RAN node or the controller platform, determines whether there is a distribution shift related to the at least one temporal characteristic of the at least one cell based on the received data, selects a learning type for an update to a model in response to the distribution shift, and updates the model for the RAN to use updated model to perform at least one action to optimize the performance of the RAN node or at least one other RAN node based on the selected learning type (paragraph [0118]), where the performance metric KPI (key performance indicator) is used to identify the distribution shift along with additional attributes including (i) a threshold of the evaluated formula/function performance metric over which shift is detected and the model updated (ii) a type of characteristic based on which distribution shift is detected, including at least one of average, maximum, a given percentile for confidence interval, trends, peak to average ratio, standard deviation and higher moments (iii) a list of underlying input KPIs of the ML model, including at least one of number of connected users, number of active users, number of bearers, DL or UL PRB utilization, PDCCH utilization, PUCCH utilization, composite available capacity, total data delivered or received at a RAN node etc., (iv) a complexity constraint measure denoting how often the ML model can be updated (v) a data storage constraint measure denoting how much data can be stored for updating the ML model (vi) a preference for passive (continuous learning without drift detection) or active learning (based on a characteristic of drift detected) model update (paragraph [0087]); and the DSLM (distribution shift learning module) triggers a training update of the ML model based on a criterion (e.g. if performance/drift exceeds a threshold) (paragraph [0089]).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Singh941 and Singh820 before him or her, to incorporate the distribution shift determination, learning type selection, and model update as taught by Singh820, to improve the CPO network node of Singh941 for the motivation of optimizing the performance of the radio access network node or at least one other radio access network node (paragraph [0003] of Singh820).
Regarding claim 6, Singh941, Singh820, and Tehrani disclose the method of claim 4, Singh941 discloses wherein utilizing the scaling includes applying disproportionate scaling to generate the adjusted capacity (the CP uses a different ML/DL model based on the different prediction/estimate time length for which load estimate should be valid and also timescale, paragraph [0155]).
Regarding claim 7, Singh941, Singh820, and Tehrani disclose the method of claim 1, Singh941 discloses wherein performing the one or more actions comprises one or more of:
providing the adjusted capacity for display; or
causing an upgrade of the RAN to be implemented based on the adjusted capacity (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; perform the adjustment to the future computational load to not exceed the computation capabilities (i.e. upgrade)).
Regarding claim 8, Singh941 discloses a device (network node or network entity called CPO (Computer Predictor and Orchestrator), comprising:
one or more processors (the CPO includes processor, Fig. 16, paragraphs [0180] – [0182]) configured to:
receive load data identifying a load on a radio access network (RAN) (the network node (CPO) acquires a predicted future computational load of baseband processing for each selected radio cell of a plurality of radio cells, where the acquiring operation is an operation of receiving predicted future computational load of baseband processing for each selected radio cell of a plurality of radio cells, paragraphs [0097], [0108]);
select one or more time series forecasting models and a classification model based on seasonality metrics associated with the load data (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; the CPO determines the radio cell based on the predicted future computational load of baseband processing and computation capabilities (i.e. selecting time series forecasting models and a classification model));
process the load data, with the one or more time series forecasting models, to forecast a capacity for the RAN (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; the CPO processes the load data to predict future computational load);
adjust, based on the capacity exceeds the capacity threshold, the capacity to generate an adjusted capacity (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; i.e. adjust the future computational load to not exceed the computation capabilities); and
perform one or more actions based on the adjusted capacity (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; i.e. perform the adjustment to the future computational load to not exceed the computation capabilities).
However, Singh941 does not explicitly disclose “process the load data and the capacity, with the classification model, to determine a binary classification outcome indicating whether the capacity exceeds a capacity threshold; adjust, based on the binary classification outcome indicating that the capacity exceeds the capacity threshold.”
Singh820 discloses “process the load data and the capacity, with the classification model, to determine whether the capacity exceeds a capacity threshold” as the apparatus receives the characteristics data related to the at least one cell from the RAN node or the controller platform, determines whether there is a distribution shift related to the at least one temporal characteristic of the at least one cell based on the received data, selects a learning type for an update to a model in response to the distribution shift, and updates the model for the RAN to use updated model to perform at least one action to optimize the performance of the RAN node or at least one other RAN node based on the selected learning type (paragraph [0118]), where the performance metric KPI (key performance indicator) is used to identify the distribution shift along with additional attributes including (i) a threshold of the evaluated formula/function performance metric over which shift is detected and the model updated (ii) a type of characteristic based on which distribution shift is detected, including at least one of average, maximum, a given percentile for confidence interval, trends, peak to average ratio, standard deviation and higher moments (iii) a list of underlying input KPIs of the ML model, including at least one of number of connected users, number of active users, number of bearers, DL or UL PRB utilization, PDCCH utilization, PUCCH utilization, composite available capacity, total data delivered or received at a RAN node etc., (iv) a complexity constraint measure denoting how often the ML model can be updated (v) a data storage constraint measure denoting how much data can be stored for updating the ML model (vi) a preference for passive (continuous learning without drift detection) or active learning (based on a characteristic of drift detected) model update (paragraph [0087]); and the DSLM (distribution shift learning module) triggers a training update of the ML model based on a criterion (e.g. if performance/drift exceeds a threshold) (paragraph [0089]).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Singh941 and Singh820 before him or her, to incorporate the distribution shift determination, learning type selection, and model update as taught by Singh820, to improve the CPO network node of Singh941 for the motivation of optimizing the performance of the radio access network node or at least one other radio access network node (paragraph [0003] of Singh820).
While Singh941 and Singh820 teaches using forecasting model and classification model, they do not explicitly teach “a binary classification outcome indicating whether the capacity exceeds a capacity threshold.”
Tehrani discloses the system uses a variety of different ML-based approaches to dynamically control cell states, where the predictive modeling (i.e. a binary classification for determining whether a future traffic load will be above or below a threshold level) uses offline data and supervised learning to train the neural network to predict a future traffic load based on current traffic load measurements, where the data includes input parameters used directly or indirectly to generate the state vector, and output parameters used directly or indirectly to generate the action (paragraphs [0045] – [0051]).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Singh941, Singh820, and Tehrani before him or her, to incorporate the binary classification as taught by Tehrani, to improve the modified CPO network node of Singh941-Singh820 for determining the binary classification outcome of failing to exceed the threshold (below threshold) or exceeding the threshold (above threshold). The motivation for doing so would have been to reduce network capacity, improve network reliability/performance, minimize operational costs, and/or maximize operating profits (paragraph [0067] of Tehrani).
Regarding claim 9, Singh941, Singh820, and Tehrani disclose the device of claim 8, Singh941 disclose wherein the one or more processors, to perform the one or more actions, are configured to one or more of:
cause a configuration update to be installed in the RAN based on the adjusted capacity (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; perform the adjustment to the future computational load to not exceed the computation capabilities (i.e. update)); or
cause a technician or an unmanned vehicle to be dispatched to service the RAN based on the adjusted capacity.
Regarding claim 10, Singh941, Singh820, and Tehrani disclose the device of claim 8, but Singh941 does not explicitly disclose wherein the one or more processors, to perform the one or more actions, are configured to:
retrain the classification model or the time series forecasting models based on the adjusted capacity.
Singh820 discloses the DSLM determines whether retraining is required, where the DSLM communication exchange with training server triggers retraining of the required model (paragraph [0095]).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Singh941 and Singh820 before him or her, to incorporate the retraining requirement for distribution shift determination, learning type selection, and model update as taught by Singh820, to improve the CPO network node of Singh941 for the motivation of optimizing the performance of the radio access network node or at least one other radio access network node (paragraph [0003] of Singh820).
Regarding claim 11, Singh941, Singh820, and Tehrani disclose the device of claim 8, Singh941 discloses wherein the load data includes non-linear time series data (the CP module contains four time series prediction/estimation model building blocks, where the foreseen time series prediction/estimation model building blocks are de-trending and de-seasonalizing to make the time series stationary and then, standard autoregressive moving average (ARMA) model can be fitted on the residual time series (“1”), additional features which the time series may depend on can be helpful for load prediction estimation (“2”), recurrent neural network (RNN)/long short term memory (LSTM) based neural network forecasting could be helpful in long term trends of load prediction/estimation (“3”), and multiple load time series from neighboring cells as features (“4”), paragraphs [0137] – [0142]).
Regarding claim 12, Singh941, Singh820, and Tehrani disclose the device of claim 8, Singh941 discloses wherein the load data includes a scheduler metric (the CP module contains four time series prediction/estimation model building blocks, where the foreseen time series prediction/estimation model building blocks are de-trending and de-seasonalizing to make the time series stationary and then, standard autoregressive moving average (ARMA) model can be fitted on the residual time series (“1”), additional features which the time series may depend on can be helpful for load prediction estimation (“2”), recurrent neural network (RNN)/long short term memory (LSTM) based neural network forecasting could be helpful in long term trends of load prediction/estimation (“3”), and multiple load time series from neighboring cells as features (“4”), paragraphs [0137] – [0142]).
Regarding claim 13, Singh941, Singh820, and Tehrani disclose the device of claim 8, but Singh941 does not explicitly disclose wherein the one or more processors are further configured to:
validate the adjusted capacity against historical capacity exceedance patterns of the RAN to ensure accuracy of the adjusted capacity.
Singh820 discloses the performance of different types of model update learning in the load prediction use case scenario for RAN is evaluated to predict the future load given the past load and other related features, where the distribution shift learning variations evaluation for load prediction includes the mean performance accuracy of Type A1 learning (baseline, re-training every day), Type A2 learning (fine-tune training), Type A3 learning (multi-cell fine-tune training), and Type B learning (ensemble training) is 83.4%, 84.1%, 82.7%, and 84%, respectively (paragraph [0110]).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Singh941 and Singh820 before him or her, to incorporate the load prediction for distribution shift determination, learning type selection, and model update as taught by Singh820, to improve the CPO network node of Singh941 for the motivation of optimizing the performance of the radio access network node or at least one other radio access network node (paragraph [0003] of Singh820).
Regarding claim 14, Singh941, Singh820, and Tehrani disclose the device of claim 8, Singh941 discloses wherein the one or more processors, to process the load data, with the one or more time series forecasting models, to forecast the capacity for the RAN, are configured to:
process the load data, with the one or more time series forecasting models, to generate a plurality of key performance indicators (KPIs) associated with the capacity of the RAN (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; the CPO receives different requested data that are monitored on each cell of interest and reported to the CPO for the machine learning CP (Compute Predictor) module and the CS (Compute Scheduler) module, including cell throughput and other KPI (key performance indicators), list of key features enabled on the cell, etc., paragraphs [0077] – [0086]); and
utilize the KPIs to refine the capacity forecasted for the RAN (the CPO receives different requested data that are monitored on each cell of interest and reported to the CPO for the machine learning CP (Compute Predictor) module and the CS (Compute Scheduler) module, including cell throughput and other KPI (key performance indicators), list of key features enabled on the cell, etc., paragraphs [0077] – [0086]).
Regarding claim 15, Singh941 discloses a non-transitory computer-readable medium storing a set of instructions (the CPO includes memory that stores respective programs including program instructions or computer program code, Fig. 16, paragraph [0182]), the set of instructions comprising:
one or more instructions that, when executed by one or more processors of a device (the CPO includes memory that stores respective programs including program instructions or computer program code that, when executed by the respective processor, enables the respective electronic device or apparatus to operate in accordance with the example embodiments, Fig. 16, paragraphs [0180] – [0182]), cause the device to:
receive load data identifying a load on a radio access network (RAN) (the network node (CPO) acquires a predicted future computational load of baseband processing for each selected radio cell of a plurality of radio cells, where the acquiring operation is an operation of receiving predicted future computational load of baseband processing for each selected radio cell of a plurality of radio cells, paragraphs [0097], [0108]),
wherein the load data includes non-linear time series data (the CP module contains four time series prediction/estimation model building blocks, where the foreseen time series prediction/estimation model building blocks are de-trending and de-seasonalizing to make the time series stationary and then, standard autoregressive moving average (ARMA) model can be fitted on the residual time series (“1”), additional features which the time series may depend on can be helpful for load prediction estimation (“2”), recurrent neural network (RNN)/long short term memory (LSTM) based neural network forecasting could be helpful in long term trends of load prediction/estimation (“3”), and multiple load time series from neighboring cells as features (“4”), paragraphs [0137] – [0142]);
select one or more time series forecasting models and a classification model based on seasonality metrics associated with the load data (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; the CPO determines the radio cell based on the predicted future computational load of baseband processing and computation capabilities (i.e. selecting time series forecasting models and a classification model));
process the load data, with the one or more time series forecasting models, to forecast a capacity for the RAN (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; the CPO processes the load data to predict future computational load);
adjust, based on determining that the capacity exceeds the capacity threshold, the capacity to generate an adjusted capacity (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; i.e. adjust the future computational load to not exceed the computation capabilities); and
perform one or more actions based on the adjusted capacity (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; i.e. perform the adjustment to the future computational load to not exceed the computation capabilities).
However, Singh941 does not explicitly disclose “process the load data and the capacity, with the classification model, to determine a binary classification outcome indicating whether the capacity exceeds a capacity threshold; adjust, based on the binary classification outcome indicating that the capacity exceeds the capacity threshold.”
Singh820 discloses “process the load data and the capacity, with the classification model, to determine whether the capacity exceeds a capacity threshold” as the apparatus receives the characteristics data related to the at least one cell from the RAN node or the controller platform, determines whether there is a distribution shift related to the at least one temporal characteristic of the at least one cell based on the received data, selects a learning type for an update to a model in response to the distribution shift, and updates the model for the RAN to use updated model to perform at least one action to optimize the performance of the RAN node or at least one other RAN node based on the selected learning type (paragraph [0118]), where the performance metric KPI (key performance indicator) is used to identify the distribution shift along with additional attributes including (i) a threshold of the evaluated formula/function performance metric over which shift is detected and the model updated (ii) a type of characteristic based on which distribution shift is detected, including at least one of average, maximum, a given percentile for confidence interval, trends, peak to average ratio, standard deviation and higher moments (iii) a list of underlying input KPIs of the ML model, including at least one of number of connected users, number of active users, number of bearers, DL or UL PRB utilization, PDCCH utilization, PUCCH utilization, composite available capacity, total data delivered or received at a RAN node etc., (iv) a complexity constraint measure denoting how often the ML model can be updated (v) a data storage constraint measure denoting how much data can be stored for updating the ML model (vi) a preference for passive (continuous learning without drift detection) or active learning (based on a characteristic of drift detected) model update (paragraph [0087]); and the DSLM (distribution shift learning module) triggers a training update of the ML model based on a criterion (e.g. if performance/drift exceeds a threshold) (paragraph [0089]).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Singh941 and Singh820 before him or her, to incorporate the distribution shift determination, learning type selection, and model update as taught by Singh820, to improve the CPO network node of Singh941 for the motivation of optimizing the performance of the radio access network node or at least one other radio access network node (paragraph [0003] of Singh820).
While Singh941 and Singh820 teaches using forecasting model and classification model, they do not explicitly teach “a binary classification outcome indicating whether the capacity exceeds a capacity threshold.”
Tehrani discloses the system uses a variety of different ML-based approaches to dynamically control cell states, where the predictive modeling (i.e. a binary classification for determining whether a future traffic load will be above or below a threshold level) uses offline data and supervised learning to train the neural network to predict a future traffic load based on current traffic load measurements, where the data includes input parameters used directly or indirectly to generate the state vector, and output parameters used directly or indirectly to generate the action (paragraphs [0045] – [0051]).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Singh941, Singh820, and Tehrani before him or her, to incorporate the binary classification as taught by Tehrani, to improve the modified CPO network node of Singh941-Singh820 for determining the binary classification outcome of failing to exceed the threshold (below threshold) or exceeding the threshold (above threshold). The motivation for doing so would have been to reduce network capacity, improve network reliability/performance, minimize operational costs, and/or maximize operating profits (paragraph [0067] of Tehrani).
Regarding claim 16, Singh941, Singh820, and Tehrani disclose the non-transitory computer-readable medium of claim 15, Singh941 discloses wherein the one or more instructions further cause the device to:
utilize RAN features and a focus feature engineering technique to improve an accuracy of the classification model relative to a classification model not trained with the RAN features and the focus feature engineering technique (the CPO receives different requested data that are monitored on each cell of interest and reported to the CPO for the machine learning CP (Compute Predictor) module and the CS (Compute Scheduler) module, including cell throughput and other KPI (key performance indicators), list of key features enabled on the cell, etc., paragraphs [0077] – [0086]),
wherein the RAN features include seasonality patterns associated with the load on the RAN (the CP module contains four time series prediction/estimation model building blocks, where the foreseen time series prediction/estimation model building blocks are de-trending and de-seasonalizing to make the time series stationary and then, standard autoregressive moving average (ARMA) model can be fitted on the residual time series (“1”), additional features which the time series may depend on can be helpful for load prediction estimation (“2”), recurrent neural network (RNN)/long short term memory (LSTM) based neural network forecasting could be helpful in long term trends of load prediction/estimation (“3”), and multiple load time series from neighboring cells as features (“4”), paragraphs [0137] – [0142]).
Regarding claim 17, Singh941, Singh820, and Tehrani disclose the non-transitory computer-readable medium of claim 15, Singh941 discloses wherein the one or more instructions, that cause the device to adjust the capacity to generate the adjusted capacity, cause the device to:
utilize scaling or a combination of scaling and rotational stitching to adjust the capacity and generate the adjusted capacity (the CP uses a different ML/DL model based on the different prediction/estimate time length for which load estimate should be valid and also timescale, paragraph [0155]).
However, Singh941 does not explicitly disclose “wherein the rotational stitching limits adjustment of the capacity to a predefined rotational limit.”
Singh820 discloses the apparatus receives the characteristics data related to the at least one cell from the RAN node or the controller platform, determines whether there is a distribution shift related to the at least one temporal characteristic of the at least one cell based on the received data, selects a learning type for an update to a model in response to the distribution shift, and updates the model for the RAN to use updated model to perform at least one action to optimize the performance of the RAN node or at least one other RAN node based on the selected learning type (paragraph [0118]), where the performance metric KPI (key performance indicator) is used to identify the distribution shift along with additional attributes including (i) a threshold of the evaluated formula/function performance metric over which shift is detected and the model updated (ii) a type of characteristic based on which distribution shift is detected, including at least one of average, maximum, a given percentile for confidence interval, trends, peak to average ratio, standard deviation and higher moments (iii) a list of underlying input KPIs of the ML model, including at least one of number of connected users, number of active users, number of bearers, DL or UL PRB utilization, PDCCH utilization, PUCCH utilization, composite available capacity, total data delivered or received at a RAN node etc., (iv) a complexity constraint measure denoting how often the ML model can be updated (v) a data storage constraint measure denoting how much data can be stored for updating the ML model (vi) a preference for passive (continuous learning without drift detection) or active learning (based on a characteristic of drift detected) model update (paragraph [0087]); and the DSLM (distribution shift learning module) triggers a training update of the ML model based on a criterion (e.g. if performance/drift exceeds a threshold) (paragraph [0089]).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Singh941 and Singh820 before him or her, to incorporate the distribution shift determination, learning type selection, and model update as taught by Singh820, to improve the CPO network node of Singh941 for the motivation of optimizing the performance of the radio access network node or at least one other radio access network node (paragraph [0003] of Singh820).
Regarding claim 18, Singh941, Singh820, and Tehrani disclose the non-transitory computer-readable medium of claim 15, Singh941 discloses wherein the one or more instructions, that cause the device to perform the one or more actions, cause the device to one or more of:
provide the adjusted capacity for display;
cause an upgrade of the RAN to be implemented based on the adjusted capacity (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; perform the adjustment to the future computational load to not exceed the computation capabilities (i.e. upgrade));
cause a configuration update to be installed in the RAN based on the adjusted capacity (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; perform the adjustment to the future computational load to not exceed the computation capabilities (i.e. update));
cause a technician or an unmanned vehicle to be dispatched to service the RAN based on the adjusted capacity; or
retrain the classification model or the time series forecasting models based on the adjusted capacity.
In addition, Singh820 discloses “retrain the classification model or the time series forecasting models based on the adjusted capacity” as the DSLM determines whether retraining is required, where the DSLM communication exchange with training server triggers retraining of the required model (paragraph [0095]).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Singh941 and Singh820 before him or her, to incorporate the retraining requirement for distribution shift determination, learning type selection, and model update as taught by Singh820, to improve the CPO network node of Singh941 for the motivation of optimizing the performance of the radio access network node or at least one other radio access network node (paragraph [0003] of Singh820).
Regarding claim 19, Singh941, Singh820, and Tehrani disclose the non-transitory computer-readable medium of claim 15, but Singh941 does not explicitly disclose wherein the one or more instructions further cause the device to:
validate the adjusted capacity against historical capacity exceedance patterns of the RAN to ensure accuracy in the adjusted capacity.
Singh820 discloses the performance of different types of model update learning in the load prediction use case scenario for RAN is evaluated to predict the future load given the past load and other related features, where the distribution shift learning variations evaluation for load prediction includes the mean performance accuracy of Type A1 learning (baseline, re-training every day), Type A2 learning (fine-tune training), Type A3 learning (multi-cell fine-tune training), and Type B learning (ensemble training) is 83.4%, 84.1%, 82.7%, and 84%, respectively (paragraph [0110]).
Before the effective filing date of the invention, it would have been obvious to one of ordinary skill in the art, having the teachings of Singh941 and Singh820 before him or her, to incorporate the load prediction for distribution shift determination, learning type selection, and model update as taught by Singh820, to improve the CPO network node of Singh941 for the motivation of optimizing the performance of the radio access network node or at least one other radio access network node (paragraph [0003] of Singh820).
Regarding claim 20, Singh941, Singh820, and Tehrani disclose the non-transitory computer-readable medium of claim 15, Singh941 discloses wherein the one or more instructions, that cause the device to process the load data, with the one or more time series forecasting models, to forecast the capacity for the RAN, cause the device to:
process the load data, with the one or more time series forecasting models, to generate a plurality of key performance indicators (KPIs) associated with the capacity of the RAN (the network node (CPO) determines an allocation of each selected radio cell of said plurality of radio cells to anyone of a plurality of processing resource units based on said predicted future computational load of baseband processing for each selected radio cell of said plurality of radio cells and computation capabilities of each of said plurality of processing resource units, where a first sub-set of said selected radio cells is combined for a predetermined time period such that a total of said predicted future computational load of baseband processing for each first radio cell of said first sub-set of said selected radio cells does not exceed said computation capabilities of a first processing resource unit of said plurality of processing resource units, paragraphs [0097], [0106]; the CPO receives different requested data that are monitored on each cell of interest and reported to the CPO for the machine learning CP (Compute Predictor) module and the CS (Compute Scheduler) module, including cell throughput and other KPI (key performance indicators), list of key features enabled on the cell, etc., paragraphs [0077] – [0086]); and
utilize the KPIs to refine the capacity forecasted for the RAN (the CPO receives different requested data that are monitored on each cell of interest and reported to the CPO for the machine learning CP (Compute Predictor) module and the CS (Compute Scheduler) module, including cell throughput and other KPI (key performance indicators), list of key features enabled on the cell, etc., paragraphs [0077] – [0086]).
Response to Arguments
Applicant’s arguments, see pages 9 – 11, filed August 26, 2026, with respect to claims 1 – 20 have been considered but are moot in view of the new ground(s) of rejection.
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure:
McEvilly et al (US Patent 12,167, 264) – the network performance metadata associated with a wireless network of a communications service provider is obtained and stored in a data lake, where one or more machine learning models are trained based on use of the wireless network performance metadata, and the one or more machine learning models, or predictions such as forecasts generated via use of the models, are used to service requests received at a cloud provider network seeking forward-looking performance characteristics of the wireless network
Any inquiry concerning this communication or earlier communications from the examiner should be directed to KAI J CHANG whose telephone number is (571)270-5448. The examiner can normally be reached Monday - Friday, 10AM-6PM EST.
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, Marcus Smith can be reached at (571)270-1096. 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.
/Kai Chang/Examiner, Art Unit 2468
/Thomas R Cairns/Primary Examiner, Art Unit 2468
Prosecution Insights