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 .
This Office Action is in response to communications filed on 12/17/2025.
Claims 1, 4-8, 11-15 & 18-20 are pending and presented for examination.
Response to Amendment
Claims 2, 3, 9, 10, 16 & 17 have been cancelled.
Claims 1, 4, 8, 11, 15 & 18 have been amended.
Rejections to claims 1, 6, 8, 11 13, 15, 18 & 20 under 35 USC 112(b) have been introduced based on amendments to these claims.
Response to Arguments
Applicant's arguments filed 12/17/2025 have been fully considered but they are not persuasive.
Regarding claim 1, applicant submits that this claim is patentable because Yousefi and Previti are silent as to some features of amended claim 1. Examiner respectfully disagrees, noting that per 35 U.S.C. 103, a patent for a claimed invention may not be obtained 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 (see §MPEP 2141).
Applicant argues that Previti is directed to “UL/DL bandwidth allocation Optimization and bears no relation whatsoever to traffic load management of a sector of cells. Examiner respectfully disagrees noting that [0006] of Previti discloses a reinforcement learning (RL) method for traffic load management, and fig 1 & [0013] of Previti discloses that while only two access points are shown as part of communications network 100, this was done solely for the sake of brevity and communications network 100 can include virtually any number of access points. Thus, a broadest reasonable interpretation of Previti is that Previti is directed to traffic load management of an access point (i.e. a sector) of any number of access points (i.e. of cells).
Applicant argues that Previti never mentions “sector” and that the bandwidth allocation solution and access point traffic offloading would not motivate one of ordinary skill to arrive at recitations of claim 1 such that “wherein the recommended load distribution parameters for the sector are configured to maximize an aggregated user throughput of the sector.”. Examiner respectfully disagrees, noting that to someone having skill in the art, a broadest reasonable interpretation is that an access point can represent a sector of a cell. Thus, it is moot that Previti does not mention “sector” because an access point can represent a sector. Under the broadest reasonable interpretation that an access point may represent a sector, the teachings of Previti (see [0006]) to perform an RL method for traffic load management from one access point to another access point based on maximizing a throughput capacity teach “wherein the recommended load distribution parameters for the sector are configured to maximize an aggregated user throughput of the sector”.
Based on the above discussion, examiner maintains rejection of claim 1 under 35 USC 103 based on Yousefi in view of Previti.
Regarding all pending claims, applicant submits that all pending claims are in condition for immediate allowance in view of the amendments and arguments made to all pending claims. Examiner respectfully disagrees, and for the same reasons as discussed above maintains rejection of all pending claims based on 35 USC 103.
Claim Rejections - 35 USC § 112
The following is a quotation of 35 U.S.C. 112(b):
(b) CONCLUSION.—The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the inventor or a joint inventor regards as the invention.
The following is a quotation of 35 U.S.C. 112 (pre-AIA ), second paragraph:
The specification shall conclude with one or more claims particularly pointing out and distinctly claiming the subject matter which the applicant regards as his invention.
Claims 1, 6, 8, 11 13, 15, 18 & 20 rejected under 35 U.S.C. 112(b) or 35 U.S.C. 112 (pre-AIA ), second paragraph, as being indefinite for failing to particularly point out and distinctly claim the subject matter which the inventor or a joint inventor (or for applications subject to pre-AIA 35 U.S.C. 112, the applicant), regards as the invention.
Claims 1, 8 & 15 recite the limitation “train the machine learning model based on the training data”. There is insufficient antecedent basis for these limitations in the claim. For the purpose of this review, examiner is interpreting this claim as “train the machine learning system based on the training data”.
Claim 1 recites the limitation “processing the input data at the machine learning system includes to, based on the optimization model, directly predict the recommended load distribution parameters for the sector”. There is insufficient antecedent basis for these limitations in the claim. For the purpose of this review, examiner is interpreting this claim as “processing the input data at the machine learning system includes to, based on an optimization model, directly predict the recommended load distribution parameters for the sector”.
Claim 6 recites the limitation “The system of claim 2”. There is insufficient antecedent basis for this limitation in the claim since claim 2 has been cancelled. For the purpose of this review, examiner is interpreting this claim as “The system of claim 1”.
Claim 13 recites the limitation “The system of claim 9”. There is insufficient antecedent basis for this limitation in the claim since claim 9 has been cancelled. For the purpose of this review, examiner is interpreting this claim as “The system of claim 8”.
Claim 20 recites the limitation “The system of claim 16”. There is insufficient antecedent basis for this limitation in the claim since claim 16 has been cancelled. For the purpose of this review, examiner is interpreting this claim as “The system of claim 15”.
Claims 11 & 18 recite the limitation “training the machine learning model comprises training a yield prediction model based on the training data with information on the training sectors as input”. There is insufficient antecedent basis for this limitation in these claims. For the purpose of this review, examiner is interpreting these claims as “training the machine learning system comprises training a yield prediction model based on the training data with information on the training sectors as input”.
Claim Rejections - 35 USC § 103
In the event the determination of the status of the application as subject to AIA 35 U.S.C. 102 and 103 (or as subject to pre-AIA 35 U.S.C. 102 and 103) is incorrect, any correction of the statutory basis (i.e., changing from AIA to pre-AIA ) for the rejection will not be considered a new ground of rejection if the prior art relied upon, and the rationale supporting the rejection, would be the same under either status.
The following is a quotation of 35 U.S.C. 103 which forms the basis for all obviousness rejections set forth in this Office action:
A patent for a claimed invention may not be obtained, notwithstanding that the claimed invention is not identically disclosed as set forth in section 102, if the differences between the claimed invention and the prior art are such that the claimed invention as a whole would have been obvious before the effective filing date of the claimed invention to a person having ordinary skill in the art to which the claimed invention pertains. Patentability shall not be negated by the manner in which the invention was made.
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.
Claims 1, 8 & 15 are rejected under 35 U.S.C. 103 as being unpatentable over Yousefi’zadeh et al. (US 10841853)(herein after “Yousefi”) in view of Previti et al. (US 2023/0262521)(herein after “Previti”).
Regarding claim 1, Yousefi discloses a system that provides load management in a Radio Access Network (RAN) (Fig 11 & col 5, lines 65-67 disclose a system for load balancing in a 5G cellular network.), the system comprising:
at least one processor and memory (Col 6, lines 1-4 disclose that the system may comprise a processor and a memory.); the at least one processor causes the system to:
receive, at a machine learning system, input data for a sector of the RAN having a plurality of cells overlapping at the sector (Fig 11 & col 6, lines 17-20 discloses the processor causing the system to import periodical measurements of cell towers (i.e. input data for a plurality of cell towers, cell towers herein after referred to as “cells”). Col 6, lines 21-67 and col 7, lines 1-10 disclose that the imported periodical measurements are used as input to a machine learning algorithm such as LLSR, ARIMA or MLPDL coupled with an optimization algorithm such as CSA, BCDSA or GA (i.e. the machine learning algorithm coupled with the optimization algorithm represent a machine learning system). Fig 10 & col 2, lines 30-38 disclose that the plurality of cells represent sectors of 3-sector sites (thus, the imported periodic measurements for one of the cells of the plurality of cells represents input data for a sector of the RAN). Table 1 and col 8, lines 38-39 disclose an overlap percentage between cells ɳi,j indicating that the cells are overlapping.), wherein prior to the receiving, the at least one processor causes the system to collect training data for a plurality of training sectors (Fig 11 & col 6, lines 17-67 and col 7, lines 1-14 disclose an iterative process for importing periodical measurements for a plurality of cells after waiting for the expiration of a timer, then choosing a learning algorithm to provide congestion thresholds for optimization algorithms that adjust tilt and handover thresholds to redistribute traffic over the plurality of cells, and then waiting for the expiration of the timer to repeat the process. Col 2, lines 39-59 disclose that the machine learning techniques are iteratively trained by the iterative process. Thus, the periodical measurements imported to the machine learning system at each cycle of the iterative process represents training data for the plurality of cells (i.e. training cells) that is collected and used to train the machine learning system prior to receiving the next round of periodical measurements, that represent input data for each cell of the plurality of cells.), and train the machine learning model based on the training data with information on the training sectors as input, and recommended load distribution parameters as output (Fig 11, col 2, lines 39-59, col 6, lines 1-67 & col 7, lines 1-14 disclose that the periodical measurements imported (i.e. as input) to the machine learning system at each cycle of the iterative process represents training data with information on the training cells that is collected and used to train the machine learning techniques for the next round of periodical measurements. Col 6, lines 23-67 & col 7, lines 1-12 disclose that the machine learning system produces predicted congestion thresholds fed to optimization algorithms that provide recommended power, antenna tilt and handover threshold changes to redistribute traffic (i.e. recommended load distribution parameters as output that is disseminated to the cells.);
process the input data at the machine learning system to determine recommended load distribution parameters for the sector based on a machine learning model, wherein processing the input data at the machine learning system includes to, based on the optimization model, directly predict the recommended load distribution parameters for the sector (Fig 11 & col 6, lines 21-67 & col 7, lines 1-10 disclose that periodical measurements (i.e. input data) are imported to the machine learning system at each cycle following a previous cycle (i.e. that trained the machine learning system), and one of three machine learning algorithms of LLSR, ARIMA or MLPDL (i.e. a machine learning model) coupled with one of three optimization algorithms of CSA, BCDSA or GA (i.e. an optimization model) processes the periodical measurements to directly disseminate recommended changes in power, antenna tilt and handover thresholds to the operating parameters for the cells (i.e. directly predict load distribution parameters for the sector).); and
apply the recommended load distribution parameters in the sector to distribute users among the cells (Fig 11 & col 7, lines 11-12 disclose that the determined operating parameter changes are disseminated to the cells. Col 6, lines 57-67 and col 7, lines 1-10 disclose that the recommended changes of power, antenna tilt and handover thresholds effectively redistribute traffic from congested cells to non-congested cells (i.e. distribute users among the cells).).
Yousefi fails to disclose wherein the recommended load distribution parameters, for the sector, are configured to maximize an aggregated user throughput of the sector.
However, Previti teaches wherein the recommended load distribution parameters are configured to maximize an aggregated user throughput of the sector ([0006] discloses a reinforcement learning method for traffic load management based on a reward value that may consist of maximizing UL or DL throughput.).
Therefore, it would have been obvious to someone having ordinary skill in the ort prior to the effective filing date of the claimed invention to have a system for providing load balancing in a 5G RAN including a machine learning system configured to minimize congestion in a sector through recommended load distribution parameters, as disclosed by Yousefi, and substitute the machine learning system configured to minimize congestion with the reinforcement learning method for traffic load management that maximizes an aggregate user throughput of the sector, as taught by Previti. The motivation to do so would be to have a system using machine learning to determine operating parameter changes for a cell in a wireless RAN that redistributes traffic between the cell and adjacent cells to maximize UL or DL throughput in the cell during hot spot periods of time for the cell (e.g. for a cell covering a stadium during an event at a stadium with a large number of people using their devices).
Regarding claim 8, Yousefi discloses a method of load management in a Radio Access Network (RAN) (Fig 11 & col 3, lines 34-36 disclose a method to balance load of a 5G cellular network.), the method comprising:
receiving, at a machine learning system, input data for a sector of the RAN having a plurality of cells overlapping at the sector (Fig 11 & col 6, lines 17-20 discloses the processor causing the system to import periodical measurements of cell towers (i.e. input data for a plurality of cell towers, cell towers herein after referred to as “cells”). Col 6, lines 21-67 and col 7, lines 1-10 disclose that the imported periodical measurements are used as input to a machine learning algorithm such as LLSR, ARIMA or MLPDL coupled with an optimization algorithm such as CSA, BCDSA or GA (i.e. the machine learning algorithm coupled with the optimization algorithm represent a machine learning system). Fig 10 & col 2, lines 30-38 disclose that the plurality of cells represent sectors of 3-sector sites (thus, the imported periodic measurements for one of the cells of the plurality of cells represents input data for a sector of the RAN). Table 1 and col 8, lines 38-39 disclose an overlap percentage between cells ɳi,j indicating that the cells are overlapping.), wherein, the method includes, prior to the receiving, collecting training data for a plurality of training sectors (Fig 11 & col 6, lines 17-67 and col 7, lines 1-14 disclose an iterative process for importing periodical measurements for a plurality of cells after waiting for the expiration of a timer, then choosing a learning algorithm to provide congestion thresholds for optimization algorithms that adjust tilt and handover thresholds to redistribute traffic over the plurality of cells, and then waiting for the expiration of the timer to repeat the process. Col 2, lines 39-59 disclose that the machine learning techniques are iteratively trained by the iterative process. Thus, the periodical measurements imported to the machine learning system at each cycle of the iterative process represents training data for the plurality of cells (i.e. training cells) that is collected and used to train the machine learning system prior to receiving the next round of periodical measurements, that represent input data for each cell of the plurality of cells.), and train the machine learning model based on the training data by training an optimization model based on the training data with information on the training sectors as input, and recommended load distribution parameters as output (Fig 11, col 2, lines 39-59, col 6, lines 1-67 & col 7, lines 1-14 disclose that the periodical measurements imported to the machine learning system at each cycle of the iterative process represents training data with information on the training cells that is collected (i.e. as input) and used to train the machine learning system for the next round of periodical measurements. Col 6, lines 23-67 & col 7, lines 1-12 disclose that the machine learning system produces predicted congestion thresholds fed to optimization algorithms (i.e. trains an optimization model), based on the periodical measurements imported, that provide recommended power, antenna tilt and handover threshold changes to redistribute traffic (i.e. recommended load distribution parameters as output that is disseminated to the cells.).);
processing the input data at the machine learning system to determine recommended load distribution parameters for the sector based on a machine learning model, wherein processing the input data at the machine learning system comprises processing the input data at the machine learning system to directly predict the recommended load distribution parameters for the sector based on the optimization model (Fig 11 & col 6, lines 21-67 & col 7, lines 1-10 disclose that periodical measurements (i.e. input data) are imported to the machine learning system at each cycle following a previous cycle (i.e. that trained the machine learning system), and one of three machine learning algorithms of LLSR, ARIMA or MLPDL (i.e. a machine learning model) coupled with one of three optimization algorithms of CSA, BCDSA or GA (i.e. the optimization model) processes the periodical measurements to directly disseminate recommended changes in power, antenna tilt and handover thresholds to the operating parameters for the cells (i.e. directly predict load distribution parameters for the sector) based on the CSA, BCDSA or GA optimization model.).); and
applying the recommended load distribution parameters in the sector to distribute users among the cells (Fig 11 & col 7, lines 11-12 disclose that the determined operating parameter changes are disseminated to the cells. Col 6, lines 57-67 and col 7, lines 1-10 disclose that the recommended changes of power, antenna tilt and handover thresholds effectively redistribute traffic from congested cells to non-congested cells (i.e. distribute users among the cells).).
Yousefi fails to disclose wherein the recommended load distribution parameters, for the sector, are configured to maximize an aggregated user throughput of the sector.
However, Previti teaches wherein the recommended load distribution parameters are configured to maximize an aggregated user throughput of the sector ([0006] discloses a reinforcement learning method for traffic load management based on a reward value that may consist of maximizing UL or DL throughput.).
Therefore, it would have been obvious to someone having ordinary skill in the ort prior to the effective filing date of the claimed invention to have a method for providing load balancing in a 5G RAN including a machine learning system configured to minimize congestion in a sector through recommended load distribution parameters, as disclosed by Yousefi, and substitute the machine learning system configured to minimize congestion with the reinforcement learning method for traffic load management that maximizes an aggregate user throughput of the sector, as taught by Previti. The motivation to do so would be to have a method that uses machine learning to determine operating parameter changes for a cell in a wireless RAN that redistributes traffic between the cell and adjacent cells to maximize UL or DL throughput in the cell during hot spot periods of time for the cell (e.g. for a cell covering a stadium during an event at a stadium with a large number of people using their devices).
Regarding claim 15, Yousefi discloses a non-transitory computer readable medium embodying programmed instructions executed by a processor, wherein the instructions when executed by the processor cause the processor to perform load management in a Radio Access Network (RAN) (Fig 11 & col 6, lines 1-4 disclose a memory storing instructions readable by a computing device that can be executed by a processor. Col 5, lines 65-67 disclose the memory and instructions may cause the processor to perform load balancing in a 5G cellular network.), which comprises to:
receive, at a machine learning system, input data for a sector of the RAN having a plurality of cells overlapping at the sector (Fig 11 & col 6, lines 17-20 discloses the processor causing the system to import periodical measurements of cell towers (i.e. input data for a plurality of cell towers, cell towers herein after referred to as “cells”). Col 6, lines 21-67 and col 7, lines 1-10 disclose that the imported periodical measurements are used as input to a machine learning algorithm such as LLSR, ARIMA or MLPDL coupled with an optimization algorithm such as CSA, BCDSA or GA (i.e. the machine learning algorithm coupled with the optimization algorithm represent a machine learning system). Fig 10 & col 2, lines 30-38 disclose that the plurality of cells represent sectors of 3-sector sites (thus, the imported periodic measurements for one of the cells of the plurality of cells represents input data for a sector of the RAN). Table 1 and col 8, lines 38-39 disclose an overlap percentage between cells ɳi,j indicating that the cells are overlapping.), wherein prior to the receiving, the instructions cause the processor to collect training data for a plurality of training sectors (Fig 11 & col 6, lines 17-67 and col 7, lines 1-14 disclose an iterative process for importing periodical measurements for a plurality of cells after waiting for the expiration of a timer, then choosing a learning algorithm to provide congestion thresholds for optimization algorithms that adjust tilt and handover thresholds to redistribute traffic over the plurality of cells, and then waiting for the expiration of the timer to repeat the process. Col 2, lines 39-59 disclose that the machine learning techniques are iteratively trained by the iterative process. Thus, the periodical measurements imported to the machine learning system at each cycle of the iterative process represents training data for the plurality of cells (i.e. training cells) that is collected and used to train the machine learning system prior to receiving the next round of periodical measurements, that represent input data for each cell of the plurality of cells.), and train the machine learning model based on the training data, wherein training the machine learning model comprises training an optimization model based on the training data with information on the training sectors as input, and recommended load distribution parameters as output (Fig 11, col 2, lines 39-59, col 6, lines 1-67 & col 7, lines 1-14 disclose that the periodical measurements imported to the machine learning system at each cycle of the iterative process represents training data with information on the training cells that is collected (i.e. as input) and used to train the machine learning system for the next round of periodical measurements. Col 6, lines 23-67 & col 7, lines 1-12 disclose that the machine learning system produces predicted congestion thresholds fed to optimization algorithms (i.e. trains an optimization model), based on the periodical measurements imported, that provide recommended power, antenna tilt and handover threshold changes to redistribute traffic (i.e. recommended load distribution parameters as output that is disseminated to the cells.).);
process the input data at the machine learning system to determine recommended load distribution parameters for the sector based on a machine learning model, wherein processing the input data at the machine learning system comprises processing the input data at the machine learning system to directly predict the recommended load distribution parameters for the sector based on the optimization model (Fig 11 & col 6, lines 21-67 & col 7, lines 1-10 disclose that periodical measurements (i.e. input data) are imported to the machine learning system at each cycle following a previous cycle (i.e. that trained the machine learning system), and one of three machine learning algorithms of LLSR, ARIMA or MLPDL (i.e. a machine learning model) coupled with one of three optimization algorithms of CSA, BCDSA or GA (i.e. the optimization model) processes the periodical measurements to directly disseminate recommended changes in power, antenna tilt and handover thresholds to the operating parameters for the cells (i.e. directly predict load distribution parameters for the sector) based on the CSA, BCDSA or GA optimization model.); and
apply the recommended load distribution parameters in the sector to distribute users among the cells (Fig 11 & col 7, lines 11-12 disclose that the determined operating parameter changes are disseminated to the cells. Col 6, lines 57-67 and col 7, lines 1-10 disclose that the recommended changes of power, antenna tilt and handover thresholds effectively redistribute traffic from congested cells to non-congested cells (i.e. distribute users among the cells).).
Yousefi fails to disclose wherein the recommended load distribution parameters, for the sector, are configured to maximize an aggregated user throughput of the sector.
However, Previti teaches wherein the recommended load distribution parameters are configured to maximize an aggregated user throughput of the sector ([0006] discloses a reinforcement learning method for traffic load management based on a reward value that may consist of maximizing UL or DL throughput.).
Therefore, it would have been obvious to someone having ordinary skill in the ort prior to the effective filing date of the claimed invention to have a non-transitory computer readable medium embodying programmed instructions executed by a processor, wherein the instructions when executed by the processor cause the processor to perform load balancing in a 5G RAN including a machine learning system configured to minimize congestion in a sector through recommended load distribution parameters, as disclosed by Yousefi, and substitute the machine learning system configured to minimize congestion with the reinforcement learning method for traffic load management that maximizes an aggregate user throughput of the sector, as taught by Previti. The motivation to do so would be to have a non-transitory computer readable medium embodying programmed instructions executed by a processor, wherein the instructions when executed by the processor cause the processor to use machine learning to determine operating parameter changes for a cell in a wireless RAN that redistributes traffic between the cell and adjacent cells to maximize UL or DL throughput in the cell during hot spot periods of time for the cell (e.g. for a cell covering a stadium during an event at a stadium with a large number of people using their devices).
Claim 4 is rejected under 35 U.S.C. 103 as being unpatentable over Yousefi’zadeh et al. (US 10841853)(herein after “Yousefi”) in view of Previti et al. (US 2023/0262521)(herein after “Previti”), as applied to claim 1, and further in view of Kumar et al. (US 2019/0205386)(herein after “Kumar”).
Regarding claim 4, Yousefi in view of Previti discloses the system of claim 1.
Yousefi discloses wherein the at least one processor causes the system to: train a yield prediction model based on the training data with information on the training sectors as input, and a positive yield classification as output (Col 12, lines 40-61 disclose a formulated optimization problem for predicting congestion (i.e. a congestion yield prediction model), per equation (1), for determining a minimal total cluster congestion ɅY over a set of combinations of changing power ƿi, changing antenna tilt tI and changing handover threshold ħi where each predicted congestion value is either x (i.e. a positive yield classification), when x is greater than 0, or 0 (a negative yield classification) when x is less than or equal to 0. The optimization problem for predicting congestion is used in the system of Fig 11, col 6, lines 1-67 & col 7, lines 1-14. Thus, the training of the machine learning system based on training with periodic measurements from the training cells as input, as disclosed in Fig 11, col 6, lines 1-67 & col 7, lines 1-14, comprises training of a congestion prediction model (i.e. a congestion yield prediction model) based on training with periodic measurements from the training cells as input, and a congestion value (i.e. positive yield classification) as output..);
receive different combinations of load distribution parameters for the sector (Col 12, lines 40-67 & col 13, lines 1-13 disclose that predicted congestion values be calculated for different combinations of changing power ƿi, changing antenna tilt tI and changing handover threshold ħi,j (i.e. different combinations of load distribution parameters) for i,j [Symbol font/0xCE] {1…,N} that would be received by the CSA, BCDSA and GA optimization algorithms.);
for each combination of the different combinations, the at least one processor causes the system to process the input data and the combination of load distribution parameters to determine whether the combination of load distribution parameters provides a positive yield in congestion for the sector based on the yield prediction model, and to determine that the combination of load distribution parameters provides a positive yield based the yield prediction model. (Fig 11, col 12, lines 40-67 & col 13, lines 1-13 disclose that for different combinations of changing power ƿi, changing antenna tilt tI and changing handover threshold ħi (i.e. for each combination of the different combinations of load distribution parameters) the machine learning system processes the periodic measurements (i.e. the input data) and the combination of changing power ƿi, changing antenna tilt tI and changing handover threshold ħi (i.e. load distribution parameters) to determine whether each combination of changing power ƿi, changing antenna tilt tI and changing handover threshold ħi provides a positive yield in congestion for sector i based on the congestion prediction model of equation 1.);
the at least one processor causes the system to select the combination of load distribution parameters that provides the highest probability of a positive yield in congestion as the recommended load distribution parameters (Col 12, lines 40-67 & Col 13, lines 1-13 disclose that the combination of changing power ƿi, changing antenna tilt tI and changing handover threshold ħi is selected (i.e. is recommended for load distribution parameter changes) that minimizes the predicted congestion (i.e. provides the highest probability of a positive yield by minimizing the difference between the average number of UEs in cell i and the predicted average number of UEs in cell i to meet a congestion threshold.).).
Yousefi fails to disclose wherein the positive yield is a positive yield in aggregated user throughput.
However, Previti teaches wherein the positive yield is a positive yield in aggregated user throughput ([0006] discloses a reinforcement learning (RL) agent that provides reward values that may be an UL or DL throughput status (i.e. a positive yield in aggregated user throughput) for each action of a group of actions related to traffic load management.).
Therefore, it would have been obvious to someone having ordinary skill in the ort prior to the effective filing date of the claimed invention to have the system of claim 1 comprising a congestion yield prediction model, as disclosed by Yousefi in view of Previti, and substitute the congestion yield prediction model with the RL UL or DL throughput yield prediction model, as taught by Previti. The motivation to do so would be to have a system using RL provide an estimated throughput yield for a cell based on traffic load management actions such as changing cell transmit power, tilt or handover parameters to maximize usage of resources for a cluster of cells without reducing aggregate throughput.
Yousefi fails to disclose wherein the positive yield classification is a Boolean value; and wherein the determining that the combination of load distribution parameters provides a positive yield is based on determining a probability.
However, Kumar further teaches wherein the positive yield classification is a Boolean value ([0052] discloses an intent arbitration model trained by machine learning that outputs Boolean predictions of whether the outcomes are positive.); and
wherein the determining that the combination of load distribution parameters provides a positive yield is based on determining a probability ([0052] discloses the intent arbitration model trained by machine learning can provide scalar values representing the probabilities that the outcomes were positive.).
Therefore, it would have been obvious to someone having ordinary skill in the art prior to the effective filing date of the claimed invention to have the system of claim 1 with a throughput yield prediction model based on a combination of load distribution parameters where a selection of load distribution parameters is made based on a highest yield that maximizes throughput, as disclosed by Yousefi in view of Previti, where the positive yield classification is a Boolean value; and wherein the determining that the combination of load distribution parameters provides a positive yield is based on determining a probability, as taught by Kumar. The motivation to do so would be to have a system using machine learning provide an estimate of whether a predicted cell throughput is greater than a current actual cell throughput based on a combination of changing cell transmission powers, tilts and handover parameters, and select the set cell transmission powers, tilts and handover parameters that has the highest probability of providing a cell throughput increase in order to perform load balancing in a manner which minimizes the risk of reducing cell throughput.
Claim 5 is rejected under 35 U.S.C. 103 as being unpatentable over Yousefi’zadeh et al. (US 10841853)(herein after “Yousefi”) in view of Previti et al. (US 2023/0262521)(herein after “Previti”) and further in view of Kumar et al. (US 2019/0205386)(herein after “Kumar”), as applied to claim 4.
Regarding claim 5, Yousefi in view of Previti and further in view of Kumar disclose the system of claim 4.
Yousefi discloses wherein: the training data used to train the yield prediction model is generated by iterative sector optimization performed on the training sectors (Fig 11, col 6, lines 1-67 & col 7, lines 1-14 disclose at step 103 importing periodical measurements that provide training data to learning algorithms (i.e. yield prediction models), wherein the imported periodical measurements are generated through iterations of changing cell power, tilt and handover parameters for a group of cells (i.e. sector optimization) on the group of cells after each expiration of a refresh timer.); and
for the iterative sector optimization for each of the training sectors, the at least one processor causes the system to: identify a plurality of cells overlapping at a training sector (Fig 10 & col 2, lines 30-38 discloses identifying a cellular network consisting of a plurality of cells that represent sectors of 3-sector sites, where a plurality of adjacent sites represent a cluster. Table 1 and col 8, lines 38-39 disclose an overlap percentage between cells ɳi,j indicating that the cells are overlapping.); and
perform multiple optimization iterations of: determining a total number of active users in the training sector (Fig 11, col 7, lines 11-14 discloses after step 115 going back to step 102 and repeating the process shown in fig 11 after each time a refresh timer expires (i.e. multiple optimization iterations). Table 1, col 6, lines 56-67, col 7, lines 1-10 & col 12, lines 40-54 disclose determining a total number of active users in a training cell i represented by [Symbol font/0x6C]i);
determining a target number of users per cell associated with a threshold of congestion of the training sector, wherein a sum of the target number of users per cell is equal to the total number of active users in the training sector (Table 1 & col 12, lines 40-60 discloses an average (i.e. target) number of UEs connected to cell i associated with a congestion threshold Ʌi, with the constraint that a sum of the average number of users per cell across the cells must be equal to the total load of the cluster of cells (i.e. total number of users in the cluster of cells.).); and
determining recommended load distribution parameters for the training sector based on the target number of users per cell that minimizes congestion of the training sector (Fig 11, col 6, lines 56-67, col 7, lines 1-10 & table 1, col 12, lines 40-54 disclose disseminating recommended changes in cell transmit power, tilt and handover parameters to the cells in the cluster based on the average number of users per cell Ʌi used in the CSA, BCDSA or GA optimization algorithms.); and
the at least one processor causes the system to associate a positive yield or negative yield with the recommended load distribution parameters (Fig 11, col 7, lines 11-12 & table 1, col 12, lines 40-54 disclose that the disseminated (i.e. recommended) changes to cell transmit power, tilt and handover changes are associated with the minimum cluster congestion ɅY positive yield.).
Yousefi fails to disclose wherein the yield prediction model determining is based on maximizing an aggregated user throughput.
However, Previti teaches wherein the yield prediction model determining is based on a maximizing an aggregated user throughput ([0006] discloses a reinforcement learning (RL) method for traffic load management based on a reward value that may consist of maximizing UL or DL throughput.).
Therefore, it would have been obvious to someone having ordinary skill in the art prior to the effective filing date of the claimed invention to have the system of claim 4 with a congestion yield prediction model that determines a target number of users per of the training sector associated with a threshold of congestion, and determines recommended load distribution parameters for the training sector based on the target number of users per cell that minimizes congestion of the training sector, as disclosed by Yousefi in view of Previti and further in view of Kumar, and substitute the congestion yield prediction model with an RL method that maximizes UL or DL throughput, as taught by Previti. The motivation to do so would be to have a system using machine learning that provides a recommendation for changes in cell transmit power, tilt and handover parameters based on a target number of users per cell that maximizes aggregate throughput while keeping the total number of users across the cluster of cells the same in order to provide load balancing that improves user experience through increased user throughput.
Claim 11 is rejected under 35 U.S.C. 103 as being unpatentable over Yousefi’zadeh et al. (US 10841853)(herein after “Yousefi”) in view of Previti et al. (US 2023/0262521)(herein after “Previti”), as applied to claim 8, and further in view of Kumar et al. (US 2019/0205386)(herein after “Kumar”).
Regarding claim 11, Yousefi in view of Previti disclose the method of claim 8.
Yousefi discloses wherein: training the machine learning model comprises training a yield prediction model based on the training data with information on the training sectors as input, and a positive yield classification as output (Col 12, lines 40-61 disclose a formulated optimization problem for predicting congestion (i.e. a congestion yield prediction model), per equation (1), for determining a minimal total cluster congestion ɅY over a set of combinations of changing power ƿi, changing antenna tilt tI and changing handover threshold ħi where each predicted congestion value is either x (i.e. a positive yield classification), when x is greater than 0, or 0 (a negative yield classification) when x is less than or equal to 0. The optimization problem for predicting congestion is used in the method of Fig 11, col 3, lines 34-67 & col 4, lines 1-47. Thus, training of the machine learning system is based on training with periodic measurements from the training cells as input, as disclosed in Fig 11, col 3, lines 34-67 & col 4, lines 1-47, comprises training of a yield prediction model based on training with periodic measurements from the training cells as input, and a congestion value (i.e. positive yield classification) as output.);
the method further comprises receiving different combinations of load distribution parameters for the sector (Col 12, lines 40-67 & Col 13, lines 1-13 disclose that predicted congestion values be calculated for different combinations of changing power ƿi, changing antenna tilt tI and changing handover threshold ħi,j (i.e. load distribution parameters) for i,j [Symbol font/0xCE] {1…,N} that would be received by the CSA, BCDSA and GA optimization algorithms.); and
processing the input data at the machine learning system to determine the recommended load distribution parameters for the sector comprises: for each combination of the different combinations, processing the input data and the combination of load distribution parameters to determine whether the combination of load distribution parameters provides a positive yield in congestion for the sector based on the yield prediction model, to determine that the combination of load distribution parameters provides a positive yield in aggregated user throughput based the yield prediction model (Col 12, lines 40-67 & Col 13, lines 1-13 disclose that predicted congestion values, based on the input data to the LLSR, ARIMA or MLPDL learning algorithms that apply optimization inputs to the predicted congestion model, be calculated for different combinations of changing power ƿi, changing antenna tilt tI and changing handover threshold ħi (i.e. load distribution parameters) to determine whether each combination of changing power ƿi, changing antenna tilt tI and changing handover threshold ħi provides a positive yield in congestion for sector i based on the congestion prediction model of equation 1.); and
selecting the combination of load distribution parameters that provides the highest probability of a positive yield as the recommended load distribution parameters (Fig 11, Col 12, lines 40-67 & Col 13, lines 1-13 disclose that the combination of changing power ƿi, changing antenna tilt tI and changing handover threshold ħi is selected that minimizes the predicted congestion (i.e. provides the highest probability of a positive yield by minimizing the difference between the average number of UEs in cell i and the predicted average number of UEs in cell i to meet a congestion threshold).).
Yousefi fails to disclose wherein the positive yield is a positive yield in aggregated user throughput;
However, Previti teaches wherein the positive yield is a positive yield in aggregated user throughput ([0006] discloses a reinforcement learning (RL) agent that provides reward values that may be an UL or DL throughput status (i.e. a positive yield in aggregated user throughput) for each action of a group of actions related to traffic load management.).
Therefore, it would have been obvious to someone having ordinary skill in the ort prior to the effective filing date of the claimed invention to have the method of claim 8 comprising a congestion yield prediction model, as disclosed by Yousefi in view of Previti, and substitute the congestion yield prediction model with the RL UL or DL throughput yield prediction model, as taught by Previti. The motivation to do so would be to have a method using RL provide an estimated throughput yield for a cell based on traffic load management actions such as changing cell transmit power, tilt or handover parameters to maximize usage of resources for a cluster of cells without reducing aggregate throughput.
Yousefi fails to disclose wherein the positive yield classification is a Boolean value; and wherein the determining that the combination of load distribution parameters provides a positive yield is based on determining a probability.
However, Kumar further teaches wherein the positive yield classification is a Boolean value ([0052] discloses an intent arbitration model trained by machine learning that outputs Boolean predictions of whether the outcomes are positive.); and
wherein the determining that the combination of load distribution parameters provides a positive yield is based on determining a probability ([0052] discloses the intent arbitration model trained by machine learning can provide scalar values representing the probabilities that the outcomes were positive.).
Therefore, it would have been obvious to someone having ordinary skill in the art prior to the effective filing date of the claimed invention to have a method of claim 8, with a throughput yield prediction model based on a combination of load distribution parameters where a selection of load distribution parameters is made based on a highest yield that maximizes throughput, as disclosed by Yousefi in view of Previti, wherein the positive yield classification is a Boolean value; and wherein the determining that the combination of load distribution parameters provides a positive yield is based on determining a probability, as taught by Kumar. The motivation to do so would be to have a method where a machine learning system provides an estimate of whether a predicted cell throughput is greater than a current actual cell throughput based on a combination of changing cell transmission powers, tilts and handover parameters, and selects the set cell transmission powers, tilts and handover parameters that has the highest probability of providing a cell throughput increase in order to perform load balancing in a manner which minimizes the risk of reducing cell throughput.
Claim 12 is rejected under 35 U.S.C. 103 as being unpatentable over Yousefi’zadeh et al. (US 10841853)(herein after “Yousefi”) in view of Previti et al. (US 2023/0262521)(herein after “Previti”) and further in view of Kumar et al. (US 2019/0205386)(herein after “Kumar”), as applied to claim 11.
Regarding claim 12, Yousefi in view of Previti and further in view of Kumar disclose the method of claim 11.
Yousefi discloses wherein: the training data used to train the yield prediction model is generated by iterative sector optimization performed on the training sectors (Fig 11, col 3, lines 34-67 & col 4, lines 1-47 disclose at step 103 importing periodical measurements that provide training data to learning algorithms (i.e. yield prediction models), wherein the imported periodical measurements are generated through iterations of changing cell power, tilt and handover parameters for a group of cells (i.e. sector optimization) on the group of cells after each expiration of a refresh timer.); and
for the iterative sector optimization for each of the training sectors, the method further comprises: identifying a plurality of cells overlapping at a training sector (Fig 10 & col 2, lines 30-38 discloses identifying a cellular network consisting of a plurality of cells that represent sectors of 3-sector sites, where a plurality of adjacent sites represent a cluster. Table 1 and col 8, lines 38-39 disclose an overlap percentage between cells ɳi,j indicating that the cells are overlapping.); and
performing multiple optimization iterations of: determining a total number of active users in the training sector (Fig 11, col 4, lines 44-47 discloses after step 115 going back to step 102 and repeating the process shown in fig 11 after each time a refresh timer expires (i.e. multiple optimization iterations). Table 1, col 4, lines 22-43 & col 12, lines 40-54 disclose determining a total number of active users in a training cell i represented by [Symbol font/0x6C]i);
determining a target number of users per cell associated with a threshold of congestion of the training sector, wherein a sum of the target number of users per cell is equal to the total number of active users in the training sector (Table 1 & col 12, lines 40-60 discloses an average (i.e. target) number of UEs connected to cell i associated with a congestion threshold Ʌi, with the constraint that a sum of the average number of users per cell across the cells must be equal to the total load of the cluster of cells (i.e. total number of users in the cluster of cells.).); and
determining recommended load distribution parameters for the training sector based on the target number of users per cell that minimizes congestion of the training sector (Fig 11, col 4, lines 22-45 & table 1, col 12, lines 40-54 disclose disseminating recommended changes in cell transmit power, tilt and handover parameters to the cells in the cluster based on the average number of users per cell Ʌi used in the CSA, BCDSA or GA optimization algorithms.); and
wherein a positive yield or negative yield is associated with the recommended load distribution parameters (Fig 11, col 4, lines 44-45 & table 1, col 12, lines 40-54 disclose that the disseminated (i.e. recommended) changes to cell transmit power, tilt and handover changes are associated with the minimum cluster congestion ɅY positive yield.).
Yousefi fails to disclose wherein the yield prediction model determining is based on maximizing an aggregated user throughput.
However, Previti teaches wherein the yield prediction model determining is based on a maximizing an aggregated user throughput ([0006] discloses a reinforcement learning (RL) method for traffic load management based on a reward value that may consist of maximizing UL or DL throughput.).
Therefore, it would have been obvious to someone having ordinary skill in the art prior to the effective filing date of the claimed invention to have the method of claim 11 with a congestion yield prediction model that determines a target number of users per of the training sector associated with a threshold of congestion, and determines recommended load distribution parameters for the training sector based on the target number of users per cell that minimizes congestion of the training sector, as disclosed by Yousefi in view of Previti and further in view of Kumar, and substitute the congestion yield prediction model with an RL method that maximizes UL or DL throughput, as taught by Previti. The motivation to do so would be to have a method using machine learning that provides a recommendation for changes in cell transmit power, tilt and handover parameters based on a target number of users per cell that maximizes aggregate throughput while keeping the total number of users across the cluster of cells the same in order to provide load balancing that improves user experience through increased user throughput.
Claim 18 is rejected under 35 U.S.C. 103 as being unpatentable over Yousefi’zadeh et al. (US 10841853)(herein after “Yousefi”) in view of Previti et al. (US 2023/0262521)(herein after “Previti”), as applied to claim 15, and further in view of Kumar et al. (US 2019/0205386)(herein after “Kumar”).
Regarding claim 18, Yousefi in view of Previti discloses the computer readable medium of claim 15.
Yousefi discloses wherein: training the machine learning model comprises training a yield prediction model based on the training data with information on the training sectors as input, and a positive yield classification as output (Col 12, lines 40-61 disclose a formulated optimization problem for predicting congestion (i.e. a congestion yield prediction model), per equation (1), for determining a minimal total cluster congestion ɅY over a set of combinations of changing power ƿi, changing antenna tilt tI and changing handover threshold ħi where each predicted congestion value is either x (i.e. a positive yield classification), when x is greater than 0, or 0 (a negative yield classification) when x is less than or equal to 0. The optimization problem for predicting congestion is used in the system of Fig 11, col 6, lines 1-67 & col 7, lines 1-14. Thus, the training of the machine learning system based on training with periodic measurements from the training cells as input, as disclosed in Fig 11 col 6, lines 1-67 & col 7, lines 1-14, comprises training of a yield prediction model based on training with periodic measurements from the training cells as input, and a congestion value (i.e. positive yield classification) as output.);
the method further comprises receiving different combinations of load distribution parameters for the sector (Col 12, lines 40-67 & Col 13, lines 1-13 disclose that predicted congestion values be calculated for different combinations of changing power ƿi, changing antenna tilt tI and changing handover threshold ħi,j (i.e. load distribution parameters) for i,j [Symbol font/0xCE] {1…,N} that would be received by the CSA, BCDSA and GA optimization algorithms.); and
processing the input data at the machine learning system to determine the recommended load distribution parameters for the sector comprises: for each combination of the different combinations, processing the input data and the combination of load distribution parameters to determine whether the combination of load distribution parameters provides a positive yield in congestion for the sector based on the yield prediction model, to determine that the combination of load distribution parameters provides a positive yield in aggregated user throughput based the yield prediction model (Fig 11, Col 12, lines 40-67 & Col 13, lines 1-13 disclose that predicted congestion values, based on the input data to the LLSR, ARIMA or MLPDL learning algorithms that apply optimization inputs to the predicted congestion model, be calculated for different combinations of changing power ƿi, changing antenna tilt tI and changing handover threshold ħi (i.e. load distribution parameters) to determine whether each combination of changing power ƿi, changing antenna tilt tI and changing handover threshold ħi provides a positive yield in congestion for sector i based on the congestion prediction model of equation 1.); and
selecting the combination of load distribution parameters that provides the highest positive yield as the recommended load distribution parameters (Col 12, lines 40-67 & Col 13, lines 1-13 disclose that the combination of changing power ƿi, changing antenna tilt tI and changing handover threshold ħi is selected that minimizes the predicted congestion (i.e. provides the highest positive yield by minimizing the difference between the average number of UEs in cell i and the predicted average number of UEs in cell i to meet a congestion threshold.).).
Yousefi fails to disclose wherein the positive yield is a positive yield in aggregated user throughput.
However, Previti teaches wherein the positive yield is a positive yield in aggregated user throughput ([0006] discloses a reinforcement learning (RL) agent that provides reward values that may be an UL or DL throughput status (i.e. a positive yield in aggregated user throughput) for each action of a group of actions related to traffic load management.).
Therefore, it would have been obvious to someone having ordinary skill in the ort prior to the effective filing date of the claimed invention to have the computer readable medium of claim 15 comprising a congestion yield prediction model, as disclosed by Yousefi in view of Previti, and substitute the congestion yield prediction model with the RL UL or DL throughput yield prediction model, as taught by Previti. The motivation to do so would be to have a non-transitory computer readable medium comprising memory with instructions that cause a processor to use RL to provide an estimated throughput yield for a cell based on traffic load management actions such as changing cell transmit power, tilt or handover parameters to maximize usage of resources for a cluster of cells without reducing aggregate throughput.
Yousefi fails to disclose wherein the positive yield classification is a Boolean value; and wherein the determining that the combination of load distribution parameters provides a positive yield is based on determining a probability.
Kumar further teaches wherein the positive yield classification is a Boolean value ([0052] discloses an intent arbitration model trained by machine learning that outputs Boolean predictions of whether the outcomes are positive.); and
wherein the determining that the combination of load distribution parameters provides a positive yield is based on determining a probability ([0052] discloses the intent arbitration model trained by machine learning can provide scalar values representing the probabilities that the outcomes were positive.).
Therefore, it would have been obvious to someone having ordinary skill in the art prior to the effective filing date of the claimed invention to have the computer readable medium of claim 15 with a throughput yield prediction model based on a combination of load distribution parameters where a selection of load distribution parameters is made based on a highest yield that maximizes throughput, as disclosed by Yousefi in view of Previti, wherein the positive yield classification is a Boolean value; and wherein the determining that the combination of load distribution parameters provides a positive yield is based on determining a probability, as taught by Kumar. The motivation to do so would be to have a non-transitory computer readable medium comprising memory with instructions that cause a processor to use machine learning to provide an estimate of whether a predicted cell throughput is greater than a current actual cell throughput based on a combination of changing cell transmission powers, tilts and handover parameters, and select the set cell transmission powers, tilts and handover parameters that has the highest probability of providing a cell throughput increase in order to perform load balancing in a manner which minimizes the risk of reducing cell throughput.
Claim 19 is rejected under 35 U.S.C. 103 as being unpatentable over Yousefi’zadeh et al. (US 10841853)(herein after “Yousefi”) in view of Previti et al. (US 2023/0262521)(herein after “Previti”) and further in view of Kumar et al. (US 2019/0205386)(herein after “Kumar”), as applied to claim 18.
Regarding claim 19, Yousefi in view of Previti and further in view of Kumar disclose the computer readable medium of claim 18.
Yousefi discloses wherein: the training data used to train the yield prediction model is generated by iterative sector optimization performed on the training sectors (Fig 11, col 6, lines 1-67 & col 7, lines 1-14 disclose at step 103 importing periodical measurements that provide training data to learning algorithms (i.e. yield prediction models), wherein the imported periodical measurements are generated through iterations of changing cell power, tilt and handover parameters for a group of cells (i.e. sector optimization) on the group of cells after each expiration of a refresh timer.); and
for the iterative sector optimization for each of the training sectors, the method further comprises: identifying a plurality of cells overlapping at a training sector (Fig 10 & col 2, lines 30-38 discloses identifying a cellular network consisting of a plurality of cells that represent sectors of 3-sector sites, where a plurality of adjacent sites represent a cluster. Table 1 and col 8, lines 38-39 disclose an overlap percentage between cells ɳi,j indicating that the cells are overlapping.); and
performing multiple optimization iterations of: determining a total number of active users in the training sector (Fig 11, col 7, lines 11-14 discloses after step 115 going back to step 102 and repeating the process shown in fig 11 after each time a refresh timer expires (i.e. multiple optimization iterations). Table 1, col 6, lines 56-67, col 7, lines 1-10 & col 12, lines 40-54 disclose determining a total number of active users in a training cell i represented by [Symbol font/0x6C]i);
determining a target number of users per cell associated with a threshold of congestion of the training sector, wherein a sum of the target number of users per cell is equal to the total number of active users in the training sector (Table 1 & col 12, lines 40-60 discloses an average (i.e. target) number of UEs connected to cell i associated with a congestion threshold Ʌi, with the constraint that a sum of the average number of users per cell across the cells must be equal to the total load of the cluster of cells (i.e. total number of users in the cluster of cells.).); and
determining recommended load distribution parameters for the training sector based on the target number of users per cell that minimizes congestion of the training sector (Fig 11, col 6, lines 56-67, col 7, lines 1-10 & table 1, col 12, lines 40-54 disclose disseminating recommended changes in cell transmit power, tilt and handover parameters to the cells in the cluster based on the average number of users per cell Ʌi used in the CSA, BCDSA or GA optimization algorithms.); and
wherein a positive yield or negative yield is associated with the recommended load distribution parameters (Fig 11, col 7, lines 11-12 & table 1, col 12, lines 40-54 disclose that the disseminated (i.e. recommended) changes to cell transmit power, tilt and handover changes are associated with the minimum cluster congestion ɅY positive yield.).
Yousefi fails to disclose wherein the yield prediction model determining is based on maximizing an aggregated user throughput.
However, Previti teaches wherein the yield prediction model determining is based on a maximizing an aggregated user throughput ([0006] discloses a reinforcement learning (RL) method for traffic load management based on a reward value that may consist of maximizing UL or DL throughput.).
Therefore, it would have been obvious to someone having ordinary skill in the art prior to the effective filing date of the claimed invention to have the computer readable medium of claim 18 with a congestion yield prediction model that determines a target number of users per of the training sector associated with a threshold of congestion, and determines recommended load distribution parameters for the training sector based on the target number of users per cell that minimizes congestion of the training sector, as disclosed by Yousefi in view of Previti and further in view of Kumar, and substitute the congestion yield prediction model with an RL method that maximizes UL or DL throughput, as taught by Previti. The motivation to do so would be to have a non-transitory computer readable medium comprising memory with instructions that cause a processor to use machine learning to provide a recommendation for changes in cell transmit power, tilt and handover parameters based on a target number of users per cell that maximizes aggregate throughput while keeping the total number of users across the cluster of cells the same in order to provide load balancing that improves user experience through increased user throughput.
Claim 6 is rejected under 35 U.S.C. 103 as being unpatentable over Yousefi’zadeh et al. (US 10841853)(herein after “Yousefi”) in view of Previti et al. (US 2023/0262521)(herein after “Previti”), as applied to claim 2, and further in view of Figura et al. (US 2013/0182700)(herein after “Figura”).
Regarding claim 6, Yousefi in view of Previti disclose the system of claim 2.
Yousefi discloses wherein the at least one processor causes the system to: identify a plurality of raw values collected over an observation period (Fig 11 & col 6, lines 18-20 disclose identifying and importing collected periodical cell measurements over a refresh timer interval (i.e. observation period.); and
store the raw values as the training data (Fig 11 & col 6, lines 18-20 disclose importing (i.e. storing and then importing) the collected periodical cell measurements as training data for LLSR, ARIMA or MPPDL learning algorithms.).
Yousefi fails to disclose wherein the raw values represent raw Key Performance indicator (KPI) values; and determining a representative KPI value for the KPI over the observation period based on the raw KPI values; and storing the representative KPI value as the training data.
However, Figura teaches wherein the raw values represent raw Key Performance indicator (KPI) values ([0025] discloses a plurality of KPIs such as minimum voice quality score per user, average call duration per network node, number of calls per region, etc.); and
determining a representative KPI value for the KPI over the observation period based on the raw KPI values ([0092] discloses a normal KPI (i.e. representative KPI) based on discording outlier KPI values (i.e. raw outlier KPI values), such as the minimum or maximum KPI value, and averaging the remaining KPI values (i.e. remaining raw KPI values) within a learning period.); and
storing the representative KPI value as the training data ([0092] discloses a normal KPI (i.e. representative KPI) that could be stored as training data in place of raw KPI values.).
Therefore, it would have been obvious to someone having ordinary skill in the art prior to the effective filing date of the claimed invention to have a system that identifies and stores raw values collected over an observation period as training data, as disclosed by Yousefi, wherein the raw values represent raw Key Performance indicator (KPI) values; and determining a representative KPI value for the KPI over the observation period based on the raw KPI values (i.e. that can be stored in place of storing the raw values), as taught by Figura. The motivation to do so would be to have a system using machine learning that can be trained through being provided periodical cell measured average KPIs from a cluster of cells based on standardized KPIs so that the machine learning system can be identically used in different networks using different vendor equipment without requiring modify the learning system to account for vendor specific defined KPIs.
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Yousefi’zadeh et al. (US 10841853)(herein after “Yousefi”) in view of Previti et al. (US 2023/0262521)(herein after “Previti”) and Figura et al. (US 2013/0182700)(herein after “Figura”), as applied to claim 6, and further in view of Maheshwari et al. (US 2016/0104076)(herein after “Maheshwari”).
Regarding claim 7, Yousefi in view of Previti and further in view of Figura disclose the system of claim 6.
Yousefi discloses wherein: the raw values are collected at collection times over the observation period (Fig 11 & col 3, lines 50-52 disclose identifying and importing collected periodical cell measurements over a refresh timer interval (i.e. observation period.).).
Yousefi fails to disclose wherein: the raw values are raw KPIs; the raw KPI values are arranged into bins for the KPI; and the at least one processor causes the system to: determine a cumulative for each bin of raw KPI values at each of the collection times over the observation period; normalize the cumulative for each of the bins at each of the collection times; and calculate a median for each of the bins over the observation period to determine the representative KPI value for the KPI.
However, Mehashwari further teaches wherein: the raw values are raw KPIs ([0240] discloses raw KPIs such as CPU usage, memory usage or request response time.);
the raw KPI values are arranged into bins for the KPI ([0627] discloses bucketing of calculated values (i.e. raw KPIs) grouped together for determining a KPI.); and
the at least one processor causes the system to: determine a cumulative for each bin of raw KPI values at each of the collection times over the observation period ([0500] discloses an aggregate KPI (i.e. cumulative KPI) that is an aggregate of the individual KPIs (i.e. raw KPIs) for a period of time (e.g. every second).);
normalize the cumulative for each of the bins at each of the collection times ([0628] discloses normalizing each KPI by multiplying by a weighting value.); and
calculate a median for each of the bins over the observation period to determine the representative KPI value for the KPI ([1201] discloses that the KPI values (i.e. representative KPI values) may be transformed into the statistical median value.).
Therefore, it would have been obvious to someone having ordinary skill in the art prior to the effective filing date of the claimed invention to have the system of claim 6 that collects raw values over a refresh time interval, as disclosed by Yousefi in view of Previti and further in view of Figura, wherein the raw values are individual KPIs; the individual KPI values are arranged into buckets (i.e. bins) for the KPI; and the at least one processor causes the system to: determine an aggregate (i.e. cumulative) for each bucket of individual KPI values at each of the collection times over a period of time; normalize the cumulative for each of the buckets at each of the collection times; and transform (i.e. calculate) a median for each of the buckets over the period of time to represent the KPI, as taught by Mehashwari. The motivation to do so would be to have a system using machine learning that can be trained through being provided periodical representative KPIs based on median calculations of the individual KPIs determined through aggregate, normalized buckets of individual KPIs, from a cluster of cells to improve the training data by removing outlier individual KPIs from skewing the representative KPI calculations.
Claim 13 is rejected under 35 U.S.C. 103 as being unpatentable over Yousefi’zadeh et al. (US 10841853)(herein after “Yousefi”) in view of Previti et al. (US 2023/0262521)(herein after “Previti”), as applied to claim 9, and further in view of Figura et al. (US 2013/0182700)(herein after “Figura”).
Regarding claim 13, Yousefi in view of Previti disclose the method of claim 9
Yousefi discloses further comprising: identifying a plurality of raw values collected over an observation period (Fig 11 & col 3, lines 50-52 disclose identifying and importing collected periodical cell measurements over a refresh timer interval (i.e. observation period.); and
storing the raw values as the training data (Fig 11 & col 3, lines 50-52 disclose importing (i.e. storing and then importing) the collected periodical cell measurements as training data for LLSR, ARIMA or MPPDL learning algorithms.).
Yousefi fails to disclose wherein the raw values represent raw Key Performance indicator (KPI) values; and determining a representative KPI value for the KPI over the observation period based on the raw KPI values; and storing the representative KPI value as the training data.
However, Figura teaches wherein the raw values represent raw Key Performance indicator (KPI) values ([0025] discloses a plurality of KPIs such as minimum voice quality score per user, average call duration per network node, number of calls per region, etc.); and
determining a representative KPI value for the KPI over the observation period based on the raw KPI values ([0092] discloses a normal KPI (i.e. representative KPI) based on discording outlier KPI values (i.e. raw outlier KPI values), such as the minimum or maximum KPI value, and averaging the remaining KPI values (i.e. remaining raw KPI values) within a learning period.); and
storing the representative KPI value as the training data ([0092] discloses a normal KPI (i.e. representative KPI) that could be stored as training data in place of raw KPI values.).
Therefore, it would have been obvious to someone having ordinary skill in the art prior to the effective filing date of the claimed invention to have a method that identifies and stores raw values collected over an observation period as training data, as disclosed by Yousefi, wherein the raw values represent raw Key Performance indicator (KPI) values; and determining a representative KPI value for the KPI over the observation period based on the raw KPI values (i.e. that can be stored in place of storing the raw values), as taught by Figura. The motivation to do so would be to have a method using machine learning that can be trained through being provided periodical cell measured average KPIs from a cluster of cells based on standardized KPIs so that the machine learning system can be identically used in different networks using different vendor equipment without requiring modify the learning system to account for vendor specific defined KPIs.
Claim 14 is rejected under 35 U.S.C. 103 as being unpatentable over Yousefi’zadeh et al. (US 10841853)(herein after “Yousefi”) in view of Previti et al. (US 2023/0262521)(herein after “Previti”) and Figura et al. (US 2013/0182700)(herein after “Figura”), as applied to claim 13, and further in view of Maheshwari et al. (US 2016/0104076)(herein after “Maheshwari”).
Regarding claim 14, Yousefi in view of Previti and further in view of Figura disclose the method of claim 13.
Yousefi discloses wherein: the raw values are collected at collection times over the observation period (Fig 11 & col 3, lines 50-52 disclose identifying and importing collected periodical cell measurements over a refresh timer interval (i.e. observation period.).).
Yousefi fails to disclose wherein: the raw values are raw KPIs; the raw KPI values are arranged into bins for the KPI; and determining the representative KPI value comprises: determining a cumulative for each bin of raw KPI values at each of the collection times over the observation period; normalizing the cumulative for each of the bins at each of the collection times; and calculating a median for each of the bins over the observation period to determine the representative KPI value for the KPI.
However, Mehashwari further teaches wherein: the raw values are raw KPIs ([0240] discloses raw KPIs such as CPU usage, memory usage or request response time.);
the raw KPI values are arranged into bins for the KPI ([0627] discloses bucketing of calculated values (i.e. raw KPIs) grouped together for determining a KPI.); and
determining the representative KPI value comprises: determining a cumulative for each bin of raw KPI values at each of the collection times over the observation period ([0500] discloses an aggregate KPI (i.e. cumulative KPI) that is an aggregate of the individual KPIs (i.e. raw KPIs) for a period of time (e.g. every second).);
normalizing the cumulative for each of the bins at each of the collection times ([0628] discloses normalizing each KPI by multiplying by a weighting value.); and
calculating a median for each of the bins over the observation period to determine the representative KPI value for the KPI ([1201] discloses that the KPI values (i.e. representative KPI values) may be transformed into the statistical median value.).
Therefore, it would have been obvious to someone having ordinary skill in the art prior to the effective filing date of the claimed invention to have the method of claim 13 that collects raw values over a refresh time interval, as disclosed by Yousefi in view of Previti and further in view of Figura, wherein the raw values are individual KPIs; the individual KPI values are arranged into buckets (i.e. bins) for the KPI; and determining the representative KPI value comprises: determining an aggregate (i.e. cumulative) for each bucket of individual KPI values at each of the collection times over a period of time; normalizing the cumulative for each of the buckets at each of the collection times; and transforming (i.e. calculating) a median for each of the buckets over the period of time to represent the KPI, as taught by Mehashwari. The motivation to do so would be to have a method using machine learning that can be trained through being provided periodical representative KPIs based on median calculations of the individual KPIs determined through aggregate, normalized buckets of individual KPIs, from a cluster of cells to improve the training data by removing outlier individual KPIs from skewing the representative KPI calculations.
Claim 20 is rejected under 35 U.S.C. 103 as being unpatentable over Yousefi’zadeh et al. (US 10841853)(herein after “Yousefi”) in view of Previti et al. (US 2023/0262521)(herein after “Previti”), as applied to claim 16, and further in view of Figura et al. (US 2013/0182700)(herein after “Figura”).
Regarding claim 20, Yousefi in view of Previti disclose the computer readable medium of claim 16.
Yousefi discloses further comprising: identifying a plurality of raw values collected over an observation period (Fig 11 & col 6, lines 18-20 disclose identifying and importing collected periodical cell measurements over a refresh timer interval (i.e. observation period.);
determining a representative KPI value for the KPI over the observation period based on the raw KPI values; and
storing the raw values for the KPI as the training data (Fig 11 & col 6, lines 18-20 disclose importing (i.e. storing and then importing) the collected periodical cell measurements as training data for LLSR, ARIMA or MPPDL learning algorithms.).
Yousefi fails to disclose wherein the raw values represent raw Key Performance indicator (KPI) values; and determining a representative KPI value for the KPI over the observation period based on the raw KPI values; and storing the representative KPI value as the training data.
However, Figura teaches wherein the raw values represent raw Key Performance indicator (KPI) values ([0025] discloses a plurality of KPIs such as minimum voice quality score per user, average call duration per network node, number of calls per region, etc.); and
determining a representative KPI value for the KPI over the observation period based on the raw KPI values ([0092] discloses a normal KPI (i.e. representative KPI) based on discording outlier KPI values (i.e. raw outlier KPI values), such as the minimum or maximum KPI value, and averaging the remaining KPI values (i.e. remaining raw KPI values) within a learning period.); and
storing the representative KPI value as the training data ([0092] discloses a normal KPI (i.e. representative KPI) that could be stored as training data in place of raw KPI values.).
Therefore, it would have been obvious to someone having ordinary skill in the art prior to the effective filing date of the claimed invention to have the computer readable medium of claim 16 that identifies and stores raw values collected over an observation period as training data, as disclosed by Yousefi, wherein the raw values represent raw Key Performance indicator (KPI) values; and determining a representative KPI value for the KPI over the observation period based on the raw KPI values (i.e. that can be stored in place of storing the raw values), as taught by Figura. The motivation to do so would be to have a non-transitory computer readable medium comprising memory with instructions that cause a processor to use machine learning that can be trained through being provided periodical cell measured average KPIs from a cluster of cells based on standardized KPIs so that the machine learning system can be identically used in different networks using different vendor equipment without requiring modify the learning system to account for vendor specific defined KPIs.
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.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to JAMES P SEYMOUR whose telephone number is (571)272-7654. The examiner can normally be reached M-F 8-5 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, Nishant Divecha can be reached at 571-270-3125. 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.
/JAMES P SEYMOUR/Examiner, Art Unit 2419
/Nishant Divecha/Supervisory Patent Examiner, Art Unit 2419