DETAILED ACTION
Notice of 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 .
Continued Examination Under 37 CFR 1.114
A request for continued examination under 37 CFR 1.114, including the fee set forth in 37 CFR 1.17(e), was filed in this application after final rejection. Since this application is eligible for continued examination under 37 CFR 1.114, and the fee set forth in 37 CFR 1.17(e) has been timely paid, the finality of the previous Office action has been withdrawn pursuant to 37 CFR 1.114. Applicant's submission filed on 06/04/2025 has been entered.
Response to Arguments
Applicant's arguments filed 06/04/2025 have been fully considered.
Applicant argues that Chen does not teach on amended claim 1 which requires a first network element with an NN encoder that generates a latent representation from its own operational state variables, and a second network element with an NN decoder that directly obtains and processes this latent representation from the first network element to produce a traffic engineering output. That it is the traffic prediction model of Chen that generates a traffic related output, not the decoder of Chen's autoencoder.
In response to the argument, Examiner respectfully disagrees.
First, the claim does not require “directly obtains and processes”. Further, “for the purpose of traffic engineering” as recited in the independent claims is intended use. As shown below, Chen teaches on the decoder receiving a latent representation directly from the encoder and processing it to generate an TE output.
Chen teaches the second network element having a NN decoder (Fig 8A & B, Decoder 8-5) configured to process the latent representation (Fig 8, vector of latent features z 8-3) in accordance with a traffic engineering (TE) function (ie. loss function) to obtain a TE output (ie. reproduction G 8-7); ([0091] Given this importance score, the loss function to transfer and/or fine tune the neural network (NN) at the target is now updated. [0105] This vector G, item 6-7 of FIG. 8B, is first mapped to a vector of latent features z by the encoder 8-1 of the similarity AE. z is then passed through the decoder 8-5 of the AE to generate a reproduction G 8-7, which aims to reproduce G 6-7 with high similarity.)
However, with the amendments to the independent claims, Chen no longer anticipates the independent claims.
Applicant argues that Chen does not teach or suggest an orchestrator that specifically deploys and places the distinct encoder and decoder functional entities as specified in amended claim 1, "identifies and selects an optimal set of network nodes for deployment, placement and distribution of encoders and decoders within these nodes in the network for the purpose of traffic engineering." .
In response to the argument, Examiner respectfully agrees.
Although the argument is persuasive, Chen still teaches on most of the limitations of the independent claims. An updated search was conducted and a prior art was discovered to read on the amended portion regarding the role of the orchestrator: US 2021/0357282 A1 (Verma)
Chen teaches on a management/orchestration server ([0010]). However, Chen is silent on the network orchestrator configured to identify and select an optimal set of network nodes for deployment, placement and distribution of encoders and decoders within the optimal set of network nodes in the network, the first network element and the second network element being part of the optimal set of network nodes.
Verma teaches the network orchestrator (Fig 2, anomaly prediction system 200) configured to identify and select an optimal set of network nodes for deployment, placement and distribution of encoders and decoders within the optimal set of network nodes in the network, the first network element and the second network element being part of the optimal set of network nodes. ([0045] The anomaly prediction system 200 includes a computer system 202 and a database 204. The computer system 202 includes at least one processor 206 for executing instructions. [0051] The processor 206 includes a data pre-processing engine 220, a log clustering engine 222, a first auto encoder 224 (including first density auto encoder 226 and first sequential auto encoder 228), a second auto encoder 230 (including second density auto encoder 232 and second sequential auto encoder 234), an ensemble manager 236, and a prediction engine 238.)
It would have been obvious to modify Chen per Verma as it would allow the modified system to provide desired neural network architecture from any placement of the orchestrator.
Please see rejection below:
Claim(s) 1-4, 6-11, 15-16, 19-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over US 2022/0053341 A1 (Chen) in view of US 2021/0357282 A1 (Verma).
Claim(s) 5, 12-14, 17-18 is/are rejected under 35 U.S.C. 103 as being unpatentable over US 2022/0053341 A1 (Chen) in view of US 2021/0357282 A1 (Verma) further in view of US 2023/0217308 A1 (Sandberg).
Claim Interpretation
This application includes one or more claim limitations that do not use the word “means,” but are nonetheless being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, because the claim limitation(s) uses a generic placeholder that is coupled with functional language without reciting sufficient structure to perform the recited function and the generic placeholder is not preceded by a structural modifier. Such claim limitation(s) is/are: “a first network element … configured to” & “a second network element configured to” & “a third network element configured to” in claim 1 and “a second network element … configured to” & “a third network element configured to” in claim 12.
Because this/these claim limitation(s) is/are being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, it/they is/are being interpreted to cover the corresponding structure described in the specification as performing the claimed function, and equivalents thereof. Support in the specification can be found: structure and algorithm: Fig 2A, 2B, 4-7, 10. [0063]-[0089].
If applicant does not intend to have this/these limitation(s) interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph, applicant may: (1) amend the claim limitation(s) to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph (e.g., by reciting sufficient structure to perform the claimed function); or (2) present a sufficient showing that the claim limitation(s) recite(s) sufficient structure to perform the claimed function so as to avoid it/them being interpreted under 35 U.S.C. 112(f) or pre-AIA 35 U.S.C. 112, sixth paragraph.
Claim Rejections - 35 USC § 112
112(b):
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.
Claims 1-20 are rejected under 35 U.S.C. 112(b) 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.
Claim 1 recites the limitation "for the purpose of traffic engineering" in line 15. This renders the claim unclear as there is insufficient antecedent basis for this limitation in the claim.
This same rejection applies to Claims 8, 15.
Claim 1 recites the limitation “the network” in line 15. This renders the claim unclear as there is insufficient antecedent basis for this limitation in the claim.
This same rejection applies to Claims 8, 15.
Claim 2 recites the limitation “the operational state variables” in line 1. This renders the claim unclear as there is insufficient antecedent basis for this limitation in the claim. Claim 2 depends on Claim 1 and Claim recites “operational state variables” twice and it is unclear as to which of these “the operational state variables” in Claim 2 is referring back to.
This same rejection applies to Claims 3 & 4.
Claim 8 recites the limitation "the second network element" in line 5. This renders the claim unclear as there is insufficient antecedent basis for this limitation in the claim.
All dependents are also rejected under the same basis/analysis as having the same deficiencies as the claims from which they depend.
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.
Claim(s) 1-4, 6-11, 15-16, 19-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over US 2022/0053341 A1 (Chen) in view of US 2021/0357282 A1 (Verma).
Regarding Claim 1:
Chen teaches A communication network ([0022] FIG. 4 illustrates a bounce diagram 4-9 showing communication and configuration among the entities of the system 2-50 (see Fig 2).) configured to employ machine learning to modify and adjust operational state variables related to traffic engineering functions TE functions ([0091] loss function) and network nodal parameters ([0075] allocation of resources), comprising:
a first network element having a neural network (NN) encoder (Fig 8A, Encoder 8-1), the NN encoder configured to obtain, from the first network element, input data (ie. first and second node vectors) and to process the input data to obtain a latent representation of the input data (ie. first and second latent vectors), ([0011] obtaining a second node vector from the target base station; obtaining a first latent vector as a first output of the autoencoder when the first node vector is input to the autoencoder; obtaining a second latent vector as a second output of the autoencoder when the second node vector is input to the autoencoder. [0072] FIG. 8A illustrates a system 8-9 including similarity network 4-40 and a similarity calculator 8-11. The similarity network 4-40 includes the encoder 8-1 and decoder 8-5. First base station statistics 3-5 (source) Gs are input to the encoder 8-1 and the latent vector Zs 5-5 is obtained. Second base station statistics 3-7 (target) Gr are input to the encoder 8-1 and the latent vector Zr 5-3 is obtained.)
the input data being values of operational state variables of the communication network obtained at the first network element; ([0031] Embodiments use available data from both data abundant reference nodes (base stations) and data limited regular nodes (base stations) without migration of traffic history logs. [0053] At operations 4-24 and 4-25 the target base station 3-9 and source base station 3-1 form respective predictions 3-16 and 4-34 of respective traffic 2-16 and 2-15. The target base station 3-9 then is configured based on prediction 3-16 at operation 4-16. The source base station 3-1 is configured based on prediction 4-34 at operation 4-27.)
and a second network element (Fig 8A & B, Decoder 8-5) configured to obtain the latent representation (Fig 8, vector of latent features z 8-3) from the first network element (Fig 8A, Encoder 8-1), ([0072] From the latent vector Zs 5-5, the decoder has been trained to reconstruct base station statistics.)
the second network element having a NN decoder (Fig 8A & B, Decoder 8-5) configured to process the latent representation (Fig 8, vector of latent features z 8-3) in accordance with a traffic engineering (TE) function (ie. loss function) to obtain a TE output (ie. reproduction G 8-7); ([0091] Given this importance score, the loss function to transfer and/or fine tune the neural network (NN) at the target is now updated. [0105] This vector G, item 6-7 of FIG. 8B, is first mapped to a vector of latent features z by the encoder 8-1 of the similarity AE. z is then passed through the decoder 8-5 of the AE to generate a reproduction G 8-7, which aims to reproduce G 6-7 with high similarity.)
and a third network element (Fig 2, Server 2-20) configured to act as a network orchestrator, ([0006] Provided herein is server configured to manage traffic prediction model transfer learning among cellular communications base stations (including as a non-limiting example, 5G base stations).)
Chen teaches on a management/orchestration server ([0010]). However, Chen is silent on the network orchestrator configured to identify and select an optimal set of network nodes for deployment, placement and distribution of encoders and decoders within the optimal set of network nodes in the network, the first network element and the second network element being part of the optimal set of network nodes.
Verma teaches, in the same field of endeavor, methods and systems of predicting server failures using machine learning, Abstract.
Verma also teaches the network orchestrator (Fig 2, anomaly prediction system 200) configured to identify and select an optimal set of network nodes for deployment, placement and distribution of encoders and decoders within the optimal set of network nodes in the network, the first network element (ie. a first auto encoder 224) and the second network element (ie. a second auto encoder 230) being part of the optimal set of network nodes. ([0045] The anomaly prediction system 200 includes a computer system 202 which includes at least one processor 206. [0051] The processor 206 includes a data pre-processing engine 220, a log clustering engine 222, a first auto encoder 224 (including first density auto encoder 226 and first sequential auto encoder 228), a second auto encoder 230 (including second density auto encoder 232 and second sequential auto encoder 234). [0086] Fig 5, The auto encoder is a neural network in which the encoder layer and the decoder layer have the same number of neurons (the same number of units).) The anomaly prediction system 200 is the orchestrator of the system.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention, to modify Chen per Verma to include the network orchestrator configured to identify and select an optimal set of network nodes for deployment, placement and distribution of encoders and decoders within the optimal set of network nodes in the network, the first network element and the second network element being part of the optimal set of network nodes. This would have been advantageous as discussed above, as it would allow the modified system to provide desired neural network architecture from any placement of the orchestrator.
Regarding Claim 2:
Chen (as modified by Verma) teaches the invention of claim 1 as described.
Chen teaches wherein the operational state variables of the communication network include at least one of an availability of computing resources in the communication network, ([0041] An output of operation 1-64 is predicted traffic at the future time of 12. Resources have been duly allocated and at 1-66 resources are sufficient to meet demand, and a number of UEs waiting for service is decreased.)
a packet size of packets transmitted in the communication network, ([0081] The prediction loss for a new data sample x[t] at the target base station 3-9 can be defined as in Eq. (4). [0082] where the summation is the size of the data set, y is the prediction of the ground truth y.)
and a metric obtained from a data flow, a traffic flow. ([0053] At operations 4-24 and 4-25 the target base station 3-9 and source base station 3-1 form respective predictions 3-16 and 4-34 of respective traffic 2-16 and 2-15. The target base station 3-9 then is configured based on prediction 3-16 at operation 4-16. The source base station 3-1 is configured based on prediction 4-34 at operation 4-27.)
Regarding Claim 3:
Chen (as modified by Verma) teaches the invention of claim 1 as described.
Chen teaches wherein the second network element is configured to modify (ie. configure) at least one of the operational state variables of the communication network in accordance with the TE output. ([0035] Embodiments are more fine-grained and node customized than previous approaches. Embodiments provide weight-level importance scores to balance between 1) keeping an NN weight learned for generality and 2) updating this weight to serve the specific target. Embodiments also customizes the model transfer for each individual edge node by adjusting the importance scores according to the similarity between the source and the target. [0053] At operations 4-24 and 4-5 the target base station 3-9 and source base station 3-1 form respective predictions 3-16 and 4-34 of respective traffic 2-16 and 2-15. The target base station 3-9 then is configured based on prediction 3-16 at operation 4-16. The source base station 3-1 is configured based on prediction 4-34 at operation 4-27.)
Regarding Claim 4:
Chen (as modified by Verma) teaches the invention of claim 1 as described.
Chen teaches wherein the TE output includes a prediction of traffic in the communication network, the prediction of the traffic including a prediction of one or more of the operational state variables; ([0053] At operations 4-24 and 4-5 the target base station 3-9 and source base station 3-1 form respective predictions 3-16 and 4-34 of respective traffic 2-16 and 2-15. The target base station 3-9 then is configured based on prediction 3-16 at operation 4-16. The source base station 3-1 is configured based on prediction 4-34 at operation 4-27.)
Regarding Claim 6:
Chen (as modified by Verma) teaches the invention of claim 1 as described.
Chen teaches that the network orchestrator is further configured: to configure the encoders and decoders via a training phase; ([0010] The similarity network includes an autoencoder, and execution of the program by the one or more processors is further configured to cause the server to train the similarity network. [0072] FIG. 8A illustrates a system 8-9 including similarity network 4-40 and a similarity calculator 8-11. The similarity network 4-40 includes the encoder 8-1 and decoder 8-5. )
Chen teaches on configuring encoders and decoders via a training phase ([0010][0072]). However, Chen is silent that the network orchestrator is further configured for maintenance of the encoders and decoders.
Verma teaches that the network orchestrator is further configured: to configure the encoders and decoders via a training phase; and for maintenance (ie. tuned/refined) of the encoders and decoders. ([0061] The first density auto encoder 226 learns a representation learning of the input vector and tries to reconstruct the same input vector as an output. To the extent the reconstructed output from the first density auto encoder 226 differs from the original input vector, various training techniques, (such as, back propagation, stochastic gradient descent, etc.,) may be employed to adjust various weights associated with the first density auto encoder 226 to reduce the reconstruction error and train the first density auto encoder 226. [0070] The machine learning models can initially be trained using one set of server logs and then later tuned/refined using an entirely different set of server logs or user feedback.)
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention, to modify Chen per Verma to include the network orchestrator is further configured for maintenance of the encoders and decoders. This would have been advantageous as discussed above, as it would allow the modified system to provide modifications as desired/required based on logs which allow for a more targeted maintenance.
Regarding Claim 7:
Chen (as modified by Verma) teaches the invention of claim 1 as described.
Chen teaches wherein: the communication network is an access network; ([0047] FIG. 4, system entities of the server 2-20, source base station 3-1, target base station 3-9, UE 2-11 and UE 2-13 are shown.)
the first network element is one of: an access network equipment; ([0047] FIG. 4, system entities of the server 2-20, source base station 3-1, target base station 3-9, UE 2-11 and UE 2-13 are shown.)
and the decoder network element is one of: an access network equipment. ([0047] FIG. 4, system entities of the server 2-20, source base station 3-1, target base station 3-9, UE 2-11 and UE 2-13 are shown.)
Regarding Claim 8:
Chen teaches A method for traffic engineering using a deployed neural network (NN) decoder in a communication network, ([0022] FIG. 4 illustrates a bounce diagram 4-9 showing communication and configuration among the entities of the system 2-50 (see Fig 2).) wherein the communication network is designed to employ machine learning to modify and adjust traffic engineering functions TE functions ([0091] loss function) and network nodal parameters ([0075] allocation of resources),
and wherein the communication network comprises a first network element having an NN encoder (Fig 8A, Encoder 8-1),
the second network element having the NN decoder, (Fig 8A, Decoder 8-5)
and a third network element (Fig 2, Server 2-20) configured to act as a network orchestrator ([0006] Provided herein is server configured to manage traffic prediction model transfer learning among cellular communications base stations (including as a non-limiting example, 5G base stations).)
the method comprising: obtaining, by the NN decoder (Fig 8A, Decoder 8-5), a latent representation (Fig 8, vector of latent features z 8-3) from the first network element (Fig 8A, Encoder 8-1) of the communication network, the latent representation having been generated by the NN encoder at the first network element from input data (ie. first and second node vectors), ([0011] obtaining a second node vector from the target base station; obtaining a first latent vector as a first output of the autoencoder when the first node vector is input to the autoencoder; obtaining a second latent vector as a second output of the autoencoder when the second node vector is input to the autoencoder. [0072] FIG. 8A illustrates a system 8-9 including similarity network 4-40 and a similarity calculator 8-11. The similarity network 4-40 includes the encoder 8-1 and decoder 8-5. First base station statistics 3-5 (source) Gs are input to the encoder 8-1 and the latent vector Zs 5-5 is obtained. Second base station statistics 3-7 (target) Gr are input to the encoder 8-1 and the latent vector Zr 5-3 is obtained.)
the input data being values of operational state variables (ie. traffic) of the communication network obtained at the first network element; ([0031] Embodiments use available data from both data abundant reference nodes (base stations) and data limited regular nodes (base stations) without migration of traffic history logs. [0053] At operations 4-24 and 4-25 the target base station 3-9 and source base station 3-1 form respective predictions 3-16 and 4-34 of respective traffic 2-16 and 2-15. The target base station 3-9 then is configured based on prediction 3-16 at operation 4-16. The source base station 3-1 is configured based on prediction 4-34 at operation 4-27.)
and processing, by the NN decoder (Fig 8A, Decoder 8-5), the latent representation (Fig 8, vector of latent features z 8-3) in accordance with a traffic engineering (TE) function (ie. loss function) to obtain a TE output (ie. reproduction G 8-7). ([0091] Given this importance score, the loss function to transfer and/or fine tune the neural network (NN) at the target is now updated. [0105] This vector G, item 6-7 of FIG. 8B, is first mapped to a vector of latent features z by the encoder 8-1 of the similarity AE. z is then passed through the decoder 8-5 of the AE to generate a reproduction G 8-7, which aims to reproduce G 6-7 with high similarity.)
Chen teaches on a management/orchestration server for the purpose of traffic engineering ([0010]). However, Chen is silent on a third network element configured to act as a network orchestrator that identifies and selects an optimal set of network nodes for deployment, placement and distribution of the NN encoder and the NN decoder within the network.
Verma teaches a third network element configured to act as a network orchestrator (Fig 2, anomaly prediction system 200) that identifies and selects an optimal set of network nodes for deployment, placement and distribution of the NN encoder and the NN decoder (Fig 5, The auto encoder is a neural network, both encoder/decoder pair) within the network. ([0045] The anomaly prediction system 200 includes a computer system 202 which includes at least one processor 206. [0051] The processor 206 includes a data pre-processing engine 220, a log clustering engine 222, a first auto encoder 224 (including first density auto encoder 226 and first sequential auto encoder 228), a second auto encoder 230 (including second density auto encoder 232 and second sequential auto encoder 234). [0086] Fig 5, The auto encoder is a neural network in which the encoder layer and the decoder layer have the same number of neurons (the same number of units).) The anomaly prediction system 200 is the orchestrator of the system.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention, to modify Chen per Verma to include a third network element configured to act as a network orchestrator that identifies and selects an optimal set of network nodes for deployment, placement and distribution of the NN encoder and the NN decoder within the network. This would have been advantageous as discussed above, as it would allow the modified system to provide desired neural network architecture from any placement of the orchestrator.
Regarding Claim 9:
Chen (as modified by Verma) teaches the invention of claim 8 as described.
Chen teaches wherein the operational state variables of the communication network include at least one of: an availability of computing resources in the communication network, ([0041] An output of operation 1-64 is predicted traffic at the future time of 12. Resources have been duly allocated and at 1-66 resources are sufficient to meet demand, and a number of UEs waiting for service is decreased.)
and a packet size of packets transmitted in the communication network. ([0081] The prediction loss for a new data sample x[t] at the target base station 3-9 can be defined as in Eq. (4). [0082] where the summation is the size of the data set, y is the prediction of the ground truth y.)
Regarding Claim 10:
Chen (as modified by Verma) teaches the invention of claim 8 as described.
Chen teaches wherein the first network element (Fig 8A, Decoder 8-5) is configured to modify (ie. configure) at least one of the operational state variables of the communication network in accordance with the TE output. ([0035] Embodiments are more fine-grained and node customized than previous approaches. Embodiments provide weight-level importance scores to balance between 1) keeping an NN weight learned for generality and 2) updating this weight to serve the specific target. Embodiments also customizes the model transfer for each individual edge node by adjusting the importance scores according to the similarity between the source and the target. [0053] At operations 4-24 and 4-5 the target base station 3-9 and source base station 3-1 form respective predictions 3-16 and 4-34 of respective traffic 2-16 and 2-15. The target base station 3-9 then is configured based on prediction 3-16 at operation 4-16. The source base station 3-1 is configured based on prediction 4-34 at operation 4-27.)
Regarding Claim 11:
Chen (as modified by Verma) teaches the invention of claim 8 as described.
Chen teaches wherein processing the latent representation in accordance with the TE function to obtain the TE output includes processing the latent representation in accordance with the TE function to obtain a prediction of one or more of the operational state variables, ([0053] At operations 4-24 and 4-5 the target base station 3-9 and source base station 3-1 form respective predictions 3-16 and 4-34 of respective traffic 2-16 and 2-15. The target base station 3-9 then is configured based on prediction 3-16 at operation 4-16. The source base station 3-1 is configured based on prediction 4-34 at operation 4-27.)
Regarding Claim 15:
Chen teaches A method for traffic engineering (TE) using a deployed neural network (NN) encoder in a communication network, ([0022] FIG. 4 illustrates a bounce diagram 4-9 showing communication and configuration among the entities of the system 2-50 (see Fig 2).) wherein the communication network is designed to employ machine learning to modify and adjust TE functions ([0091] loss function) and network nodal parameters ([0075] allocation of resources),
and wherein the communication network comprises a first network element having the NN encoder, (Fig 8A, Encoder 8-1)
a second network element having an NN decoder, (Fig 8A, Decoder 8-5)
and a third network element (Fig 2, Server 2-20) configured to act as a network orchestrator ([0006] Provided herein is server configured to manage traffic prediction model transfer learning among cellular communications base stations (including as a non-limiting example, 5G base stations).)
the method comprising, at the first network element (Fig 8A, Encoder 8-1) of a communication network: obtaining input data (ie. first and second node vectors) of the communication network, the input data being values of operational state variables (ie. respective traffic) of the communication network; ([0031] Embodiments use available data from both data abundant reference nodes (base stations) and data limited regular nodes (base stations) without migration of traffic history logs. [0053] At operations 4-24 and 4-25 the target base station 3-9 and source base station 3-1 form respective predictions 3-16 and 4-34 of respective traffic 2-16 and 2-15. The target base station 3-9 then is configured based on prediction 3-16 at operation 4-16. The source base station 3-1 is configured based on prediction 4-34 at operation 4-27.)
encoding, using the NN encoder (Fig 8A, Encoder 8-1), the input data to obtain a latent representation; ([0011] obtaining a second node vector from the target base station; obtaining a first latent vector as a first output of the autoencoder when the first node vector is input to the autoencoder; obtaining a second latent vector as a second output of the autoencoder when the second node vector is input to the autoencoder. [0072] FIG. 8A illustrates a system 8-9 including similarity network 4-40 and a similarity calculator 8-11. The similarity network 4-40 includes the encoder 8-1 and decoder 8-5. First base station statistics 3-5 (source) Gs are input to the encoder 8-1 and the latent vector Zs 5-5 is obtained. Second base station statistics 3-7 (target) Gr are input to the encoder 8-1 and the latent vector Zr 5-3 is obtained.)
providing the latent representation (Fig 8, vector of latent features z 8-3) to the second network element, the second network element having the NN decoder (Fig 8, vector of latent features z 8-3), with the NN decoder (Fig 8A, Decoder 8-5), in accordance with a TE function (ie. loss function) to obtain a TE output (ie. reproduction G 8-7). ([0091] Given this importance score, the loss function to transfer and/or fine tune the neural network (NN) at the target is now updated. [0105] This vector G, item 6-7 of FIG. 8B, is first mapped to a vector of latent features z by the encoder 8-1 of the similarity AE. z is then passed through the decoder 8-5 of the AE to generate a reproduction G 8-7, which aims to reproduce G 6-7 with high similarity.)
Chen teaches on a management/orchestration server ([0010]). However, Chen is silent on a third network element configured to act as a network orchestrator that identifies and selects an optimal set of network nodes for deployment, placement and distribution of the NN encoder and the NN decoder within the network and the NN decoder deployed thereat by the network orchestrator.
Verma teaches a third network element configured to act as a network orchestrator that identifies and selects an optimal set of network nodes for deployment, placement and distribution of the NN encoder and the NN decoder (Fig 5, The auto encoder is a neural network, both encoder/decoder pair) within the network and the NN decoder deployed thereat by the network orchestrator. ([0045] The anomaly prediction system 200 includes a computer system 202 which includes at least one processor 206. [0051] The processor 206 includes a data pre-processing engine 220, a log clustering engine 222, a first auto encoder 224 (including first density auto encoder 226 and first sequential auto encoder 228), a second auto encoder 230 (including second density auto encoder 232 and second sequential auto encoder 234). [0086] Fig 5, The auto encoder is a neural network in which the encoder layer and the decoder layer have the same number of neurons (the same number of units).) The anomaly prediction system 200 is the orchestrator of the system.
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention, to modify Chen per Verma to include a third network element configured to act as a network orchestrator that identifies and selects an optimal set of network nodes for deployment, placement and distribution of the NN encoder and the NN decoder within the network and the NN decoder deployed thereat by the network orchestrator. This would have been advantageous as discussed above, as it would allow the modified system to provide desired neural network architecture from any placement of the orchestrator.
Regarding Claim 16:
Chen (as modified by Verma) teaches the invention of claim 15 as described.
Chen teaches wherein the operational state variables of the communication network include at least one of an availability of computing resources in the communication network, ([0041] An output of operation 1-64 is predicted traffic at the future time of 12. Resources have been duly allocated and at 1-66 resources are sufficient to meet demand, and a number of UEs waiting for service is decreased.)
and a packet size of packets transmitted in the communication network. ([0081] The prediction loss for a new data sample x[t] at the target base station 3-9 can be defined as in Eq. (4). [0082] where the summation is the size of the data set, y is the prediction of the ground truth y.)
Regarding Claim 19:
Chen (as modified by Verma) teaches the invention of claim 8 as described.
Chen teaches A tangible, non-transitory computer-readable medium having stored thereon instructions to be performed by a processor ([0131] The one or more non-volatile memories 11-3 may include a non-transitory computer readable medium storing instructions for execution by the one or more hardware processors 11-9 to cause apparatus 11-1 to perform any of the methods of embodiments disclosed herein.) to perform the method of claim 8.
Regarding Claim 20:
Chen (as modified by Verma) teaches the invention of claim 15 as described.
Chen teaches A tangible, non-transitory computer-readable medium having stored thereon instructions to be performed by a processor ([0131] The one or more non-volatile memories 11-3 may include a non-transitory computer readable medium storing instructions for execution by the one or more hardware processors 11-9 to cause apparatus 11-1 to perform any of the methods of embodiments disclosed herein.) to perform the method of claim 15.
Claim(s) 5, 12-14, 17-18 is/are rejected under 35 U.S.C. 103 as being unpatentable over US 2022/0053341 A1 (Chen) in view of US 2021/0357282 A1 (Verma) further in view of US 2023/0217308 A1 (Sandberg).
Regarding Claim 5:
Chen (as modified by Verma) teaches the invention of claim 1 as described.
Chen teaches wherein the latent representation is an initial latent representation, wherein: the communication network comprises additional network elements (ie. base stations) each having a respective additional NN encoder (Fig 8A, encoder 8-1) ([0031] Embodiments use available data from both data abundant reference nodes (base stations) and data limited regular nodes (base stations) without migration of traffic history logs.) configured to obtain, from a respective additional network element, an additional latent representation (Fig 8, vector of latent features z 8-3) of respective additional input data (ie. first and second node vectors), ([0011] obtaining a second node vector from the target base station; obtaining a first latent vector as a first output of the autoencoder when the first node vector is input to the autoencoder; obtaining a second latent vector as a second output of the autoencoder when the second node vector is input to the autoencoder. [0072] FIG. 8A illustrates a system 8-9 including similarity network 4-40 and a similarity calculator 8-11. The similarity network 4-40 includes the encoder 8-1 and decoder 8-5. First base station statistics 3-5 (source) Gs are input to the encoder 8-1 and the latent vector Zs 5-5 is obtained. Second base station statistics 3-7 (target) Gr are input to the encoder 8-1 and the latent vector Zr 5-3 is obtained.)
the respective additional input data being related to additional values of operational state variables (ie. respective traffic) of the communication network obtained at the respective additional network element. ([0053] At operations 4-24 and 4-25 the target base station 3-9 and source base station 3-1 form respective predictions 3-16 and 4-34 of respective traffic 2-16 and 2-15. The target base station 3-9 then is configured based on prediction 3-16 at operation 4-16. The source base station 3-1 is configured based on prediction 4-34 at operation 4-27.)
Chen teaches on multiple network elements located on multiple base stations (each including two network elements, one encoder and one decoder) ([0031][0053]). However, Chen (as modified by Verma) is silent that the decoder (second network element) receives multiple latent representations from multiple encoders. Chen is silent that the second network element is configured to obtain the additional latent representations from the additional network elements, the second network element configured to process the additional latent representation and the initial representation in accordance with the TE function to obtain the TE output, wherein: the second network element is configured to obtain a concatenation of the initial latent representation with the additional latent representations, and the second network element is configured to process the concatenation in accordance with the TE function to obtain the TE output.
Sandberg teaches, in the same field of endeavor, systems and methods related to traffic flow prediction in a wireless network, Abstract.
Sandberg also teaches that the second network element (ie. decoder) is configured to obtain the additional latent representations from the additional network elements (ie. encoders chained together), the second network element configured to process the additional latent representation and the initial representation in accordance with the TE function to obtain the TE output, ([0125] The transformer consists of two main components: (1) a set of encoders chained together and (2) a set of decoders chained together. The function of each encoder is to process its input vectors to generate what are known as encodings, which contain information about the parts of the inputs which are relevant to each other. It passes its set of generated encodings to the next encoder as inputs. Each decoder does the opposite, taking all the encodings and processing them, using their incorporated contextual information to generate an output sequence.)
wherein: the second network element (ie. decoder) is configured to obtain a concatenation of the initial latent representation with the additional latent representations (ie. generated encodings from multiple encoders), and the second network element is configured to process the concatenation in accordance with the TE function to obtain the TE output. ([0125] The transformer consists of two main components: (1) a set of encoders chained together and (2) a set of decoders chained together. The function of each encoder is to process its input vectors to generate what are known as encodings, which contain information about the parts of the inputs which are relevant to each other. It passes its set of generated encodings to the next encoder as inputs. Each decoder does the opposite, taking all the encodings and processing them, using their incorporated contextual information to generate an output sequence.)
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention, to modify Chen (as modified by Verma) by modifying Chen per Sandberg to include that the second network element is configured to obtain the additional latent representations from the additional network elements, the second network element configured to process the additional latent representation and the initial representation in accordance with the TE function to obtain the TE output, wherein: the second network element is configured to obtain a concatenation of the initial latent representation with the additional latent representations, and the second network element is configured to process the concatenation in accordance with the TE function to obtain the TE output. This would have been advantageous as discussed above, as it would allow the combined system to provide a more relevant output by allowing the decoder to process multiple related encodings together.
Regarding Claim 12:
Chen (as modified by Verma) teaches the invention of claim 8 as described.
Chen teaches on multiple network elements located on multiple base stations (each including two network elements, one encoder and one decoder) ([0031][0053]). However, Chen (as modified by Verma) is silent that the decoder (second network element) receives multiple latent representations from multiple encoders. Chen (as modified by Verma) is silent wherein the latent representation is an initial latent representation, the method further comprising, at the first network element of the communication network: obtaining additional latent representations from respective additional network elements of the communication network, the additional latent representations representing respective additional input data obtained at the respective additional network elements, the additional input data being additional values of operational state variables of the communication network wherein processing the initial latent representation in accordance with the TE function to obtain the TE output includes processing the additional latent representations and the initial latent representation in accordance with the TE function to obtain the TE output.
Sandberg teaches wherein the latent representation is an initial latent representation, the method further comprising, at the first network element (ie. decoders chained together) of the communication network: obtaining additional latent representations (ie. multiple encodings) from respective additional network elements (ie. encoders chained together) of the communication network, the additional latent representations representing respective additional input data (ie. input vectors) obtained at the respective additional network elements, the additional input data being additional values of operational state variables (ie. dataset from traffic flows) of the communication network (ie. RAN) wherein processing the initial latent representation in accordance with the TE function to obtain the TE output includes processing the additional latent representations and the initial latent representation in accordance with the TE function to obtain the TE output. ([0120] As discussed below, during a training phase, the traffic type predictor 202 trains a model (e.g., a neural network) based on a dataset. The dataset is generated from actual traffic flows in the RAN. [0125] The transformer consists of two main components: (1) a set of encoders chained together and (2) a set of decoders chained together. The function of each encoder is to process its input vectors to generate what are known as encodings, which contain information about the parts of the inputs which are relevant to each other. It passes its set of generated encodings to the next encoder as inputs. Each decoder does the opposite, taking all the encodings and processing them, using their incorporated contextual information to generate an output sequence.)
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention, to modify Chen (as modified by Verma) by modifying Chen per Sandberg to include at the first network element of the communication network: obtaining additional latent representations from respective additional network elements of the communication network, the additional latent representations representing respective additional input data obtained at the respective additional network elements, the additional input data being additional values of operational state variables of the communication network wherein processing the initial latent representation in accordance with the TE function to obtain the TE output includes processing the additional latent representations and the initial latent representation in accordance with the TE function to obtain the TE output. This would have been advantageous as discussed above, as it would allow the combined system to provide a more relevant output by allowing the decoder to process multiple related encodings together.
Regarding Claim 13:
Chen (as modified by Verma & Sandberg) teaches the invention of claim 12 as described.
Chen teaches on multiple network elements located on multiple base stations (each including two network elements, one encoder and one decoder) ([0031][0053]). However, Chen (as modified by Verma) is silent that the decoder (second network element) receives multiple latent representations from multiple encoders. Chen (as modified by Verma) is silent on further comprising obtaining a concatenation of the initial latent representation with the additional latent representations, wherein processing the additional latent representations and the initial latent representation in accordance with the TE function to obtain the TE output includes processing the concatenation in accordance with the TE function to obtain the TE output.
Sandberg teaches obtaining a concatenation of the initial latent representation with the additional latent representations (ie. generated encodings from multiple encoders), wherein processing the additional latent representations and the initial latent representation in accordance with the TE function to obtain the TE output includes processing the concatenation in accordance with the TE function to obtain the TE output. ([0120] As discussed below, during a training phase, the traffic type predictor 202 trains a model (e.g., a neural network) based on a dataset. The dataset is generated from actual traffic flows in the RAN. [0125] The transformer consists of two main components: (1) a set of encoders chained together and (2) a set of decoders chained together. The function of each encoder is to process its input vectors to generate what are known as encodings, which contain information about the parts of the inputs which are relevant to each other. It passes its set of generated encodings to the next encoder as inputs. Each decoder does the opposite, taking all the encodings and processing them, using their incorporated contextual information to generate an output sequence.)
It would have been obvious to a person having ordinary skill in the art before the effective filing date of the claimed invention, to modify Chen (as modified by Verma) by modifying Chen per Sandberg to include obtaining a concatenation of the initial latent representation with the additional latent representations, wherein processing the additional latent representations and the initial latent representation in accordance with the TE function to obtain the TE output includes processing the concatenation in accordance with the TE function to obtain the TE output. This would have been advantageous as discussed above, as it would allow the combined system to provide a more relevant output by allowing the decoder to process multiple related encodings together.
Regarding Claim 14:
Chen (as modified by Verma) teaches the invention of claim 8 as described.
Chen teaches on multiple types of dev