DETAILED ACTION
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
The amendment submitted on 01/26/2026 has been received and considered by the Examiner. Claims 1, 12, and 20 were amended, and claims 1-20 remain pending.
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.
Allowable Subject Matter
Claims 5, 11, and 15-16 are objected to as being dependent upon a rejected base claim, but would be allowable if rewritten in independent form including all of the limitations of the base claim and any intervening claims.
Claims 5 and 15 are allowable because the prior art does not teach “a second request message” that is “used to indicate the density of the first reference signal” in combination with the other limitations of dependent claims 5 and 15.
Claim 11 is allowable because the prior art does not teach that “the density of the first reference signal is ½ or ¼ the density of the second reference signal” in combination with the other limitations of dependent claim 11.
Claim 16 is allowable because the prior art does not teach “periodically sending, by the second device, the second request message to the first device; or sending, by the second device, the second request message to the first device based on receiving second indication information from the first device, wherein the second indication information is used to indicate that the first neural network model is determined” in combination with the other limitations of dependent claim 16.
Response to Arguments
The Applicant argues on page 9 of their remarks that “Yoon fails to teach or suggest that the first CSI is obtained by the second device based on a part of the first reference signal received on a resource used to receive a low-density reference signal” (Applicant Remarks, p. 9). Regarding the mapping presented in the office action dated 12/29/2025, the Applicant adds that “[i]t is improper to interpret that the low-density CSI-RS of Yoon reads on two distinct features of ‘the first reference signal’ and ‘the resource used to receive a low-density reference signal’, as recited in claim 1” (Remarks, p. 9).
However, the Examiner respectfully submits that the previous office action did not directly map the “low density CSI-RS” of Yoon to both the claimed “first reference signal” and “the resource used to receive a low-density reference signal”. Rather, the fact that the low-density reference signal is transmitted shows that a resource exists to transmit it. A low-density reference signal cannot be transmitted without a corresponding resource, so its existence is evidence that the resource also exists.
The Applicant also writes on pages 9-10 of their remarks that “Yoon specifically requires performing both the beam measurement using a CSI-RS received based on the low-density configuration information and the time tracking using a CSI-RS received based on the high-density configuration information”, and the Applicant concludes from this that “[s]uch distinction further emphasizes that it is improper to interpret the low-density CSI-RS of Yoon to read on two distinct features of ‘the first reference signal’ and ‘the resource used to receive a low-density reference signal,’ as recited in claim 1 [emphasis in original]” (Remarks, p. 9-10).
Again, the Examiner respectfully reiterates that the low-density reference signal in Yoon does not directly equate to “the resources used to receive a low-density reference signal”. Rather, its existence is evidence that those resources exist. Thus, the rejection based on Chavva in view of Yoon and Zhou is properly maintained.
Claim Rejections - 35 USC § 103
The text of those sections of Title 35, U.S. Code not included in this action can be found in a prior Office action.
Claim(s) 1-2, 6, 9-10, 12-13, 17, and 20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Chavva et al. (US 2021/0351885 A1, hereinafter “Chavva”) in view of Yoon et al. (US 2021/0105111 A1, hereinafter “Yoon”) and further in view of Zhou et al. (US 2020/0106498 A1, hereinafter “Zhou”).
As to Claim 1, 12, and 20:
Chavva describes a method to optimize beam selection by processing a CSI report using machine learning.
Specifically, Chavva teaches:
Sending, by a first device, a first reference signal to a second device
Paragraph 0107 of Chavva discusses sending a “CSI-RS” based on “the CSI enable trigger” (Chavva, 0107).
Receiving, by the first device, first channel state information (CSI) from the second device
Paragraph 0179 of Chavva describes “send[ing] the first CSI report at time instance ‘t’” (Chavva, 0179).
The first CSI is obtained by the second device based on the first reference signal
Chavva adds in paragraph 0179 that “[t]he feedback parameters are computed using measurements performed using CSI-RS” (Chavva, 0179).
Obtaining, by the first device, second CSI based on the first CSI and a first neural network model
(“The CSI report is sent to a gNB, which includes feedback parameters, computed and predicted using ML. The feedback parameters are computed using measurements performed using CSI-RS.... the UE 601 sends the first CSI report at the time instance ‘t’” (Chavva, 0179).
Here, “sent” maps to “obtaining”,
“a gNB” maps to “the first device”,
“feedback parameters” maps to “second CSI”,
“using measurements performed using CSI-RS” maps to “based on the first CSI”, and
“computed and predicted using ML” maps to “based on ... a first neural network model”).
The second CSI is used to indicate channel information between the first device and the second device
(“The output of the neural network ... can be utilized for predicting the CQI and PMI” (Chavva, 0167).
Here, “the output of the neural network” maps to “the second CSI”,
“predicting the CQI” maps to “indicate channel information between the first device and the second device” since it is obvious from context that the CQI measurement is taken between the first device and the second device).
Updating ... the first neural network model
(Fig. 17 in Chavva shows the training process for a neural network.
Here, “update weights” in step 1703 maps to “updating ... the first neural network model”).
Chavva does not explicitly disclose:
A first neural network model deployed on the first device
However, Chavva does separately teach:
A first neural network model deployed on ... [a] device
Paragraph 0007 of Chavva describes a neural network that generates CSI reports deployed on a UE.
And:
The first device
Paragraph 0007 of Chavva describes a neural network that generates CSI reports deployed on a UE.
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to deploy the neural network described in Chavva on the base station as opposed to the UE. The computational results of the neural network are transferred to the network device anyway, so the choice of which device should host the model and perform computation is arbitrary.
Chavva also does not explicitly disclose:
Density of the first reference signal is less than or equal to a density of a second reference signal, and the second reference signal is a normal-density reference signal
When the density of the first reference signal is less than the density of the second reference signal, the first CSI is obtained by the second device based on the first reference signal
When the density of the first reference signal is equal to the density of the second reference signal, the first CSI is obtained by the second device based on a part of the first reference signal received on a resource used to receive a low-density reference signal
However, Yoon does describe a method to simultaneously configure high-density and low-density CSI reference signals.
Specifically, Yoon teaches:
Density of the first reference signal is less than or equal to a density of a second reference signal, and the second reference signal is a normal-density reference signal
Paragraph 0013 of Yoon describes configuring and sending sending low-density and high-density reference signals to measure CSI.
When the density of the first reference signal is less than the density of the second reference signal, the first CSI is obtained by the second device based on the first reference signal
Paragraph 0109 of Yoon describes “a CSI-RS used as the TFRS” configured with “higher density than the CSI-RS for BM”. Here, the “CSI-RS for BM” is the “first reference signal” with lower density.
When the density of the first reference signal is equal to the density of the second reference signal, the first CSI is obtained by the second device based on a part of the first reference signal received on a resource used to receive a low-density reference signal
Paragraph 0112 of Yoon describes sending a low-density reference signal together with a high-density reference signal: “the TFRS may be composed of a CSI-RS for BM [beam management] and a CSI-RS added for tracking” (Yoon, 0112).
Here, the combination of “the TFRS” and “a CSI-RS added” combined can be considered the “first reference signal”, another combined reference signal can be considered the “second reference signals”, meaning the first combined signals’ density will be “equal to the density of the second reference signal”.
The “CSI-RS for BM” is the “part of the first reference signal received on a resource used to receive a low-density reference signal” because the BM RS is low-density.
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Yoon’s practice of sending reference signals with different densities on the same resources into Chavva’s method for using machine learning to generate CSI feedback. The ability to use different reference signal densities allows for increased flexibility in Chavva’s AI-based CSI reporting.
The combination of Chavva and Yoon also does not explicitly disclose:
Updating, by the first device, the first neural network model based on a preset trigger condition being met
The preset trigger condition includes that a first timer expires, and the first timer is started when the first device sends the first reference signal to the second device
However, Zhou does describe a method for measuring and reporting results from an aperiodic reference signal.
Specifically, Zhou teaches:
Updating, by the first device ... based on a preset trigger condition being met
Paragraph 0006 of Zhou describes updating a beam configuration based on expiration of a timer (i.e. the “preset trigger condition”).
The preset trigger condition includes that a first timer expires
Paragraph 0006 of Zhou describes updating a beam configuration based on expiration of a timer (i.e. the “preset trigger condition”).
The first timer is started when the first device sends the first reference signal to the second device
Zhou states in paragraph 0101 that “base station 105-d may start the timer based on scheduling information for the aperiodic CSI-RS in the DCI” (Zhou, 0101).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to apply Zhou’s method for updating a configuration after receiving a reference signal for measuring CSI to neural network updates described in Chavva. If a device is currently monitoring for a CSI-RS, it doesn’t make sense to update either the beam configuration or a neural network used to measure CSI, so it would have been obvious to wait until after the monitoring period is over to retrain the AI model.
Claim 12 contains substantially the same subject matter as Claim 1 from the perspective of the UE.
Claim 20 contains substantially the same subject matter as Claim 1 in the form of an apparatus claim.
As to Claim 2:
Chavva teaches:
Determining, by the first device, the first neural network model
Paragraph 0131 of Chavva describes training a neural network.
Sending, by the first device, the second reference signal to the second device
Paragraph 0177 of Chavva describes a gNB sending a CSI-RS to a UE.
Receiving, by the first device, third CSI from the second device
Paragraph 0179 of Chavva describes a UE “periodically” receiving CSI and generating a CSI report which includes doing it for a “third” time.
The third CSI is obtained by the second device based on the second reference signal
Paragraph 0179 of Chavva describes a UE “periodically” receiving CSI and generating a CSI report which includes doing it for a “third” time.
Training, by the first device, a neural network based on the third CSI, to obtain the first neural network model
Paragraph 0131 of Chavva describes training a neural network based on generated CSI.
Chavva does not explicitly disclose:
Before the sending, by the first device, a first reference signal to a second device
However, Yoon does teach:
Before the sending, by the first device, a first reference signal to a second device
Paragraph 0012 of Yoon describes configuring a reference signal as high density or low density before sending a CSI-RS.
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Yoon’s practice of sending reference signals with different densities on the same resources into Chavva’s method for using machine learning to generate CSI feedback. The ability to use different reference signal densities allows for increased flexibility in Chavva’s AI-based CSI reporting.
As to Claim 6 and 17:
Chavva teaches:
Sending ... when determining the first neural network training model
Paragraph 0131 of Chavva describes training a neural network, which includes “send data such as the contents stored in the database 602b” (Chavva, 0131).
Chavva does not explicitly disclose:
Before sending, by a first device, a first reference signal to a second device
Sending, by the first device, first indication information to the second device
The first indication information is used to indicate the density of the first reference signal
However, Yoon does teach:
Before the sending, by the first device, a first reference signal to a second device
Paragraph 0012 of Yoon describes configuring a reference signal as high density or low density before sending a CSI-RS.
Sending, by the first device, first indication information to the second device
Paragraph 0012 of Yoon describes a base station sending “low-density configuration information” (i.e. “first indication information”) to a user equipment.
The first indication information is used to indicate the density of the first reference signal
Paragraph 0012 of Yoon describes a base station sending “low-density configuration information” (i.e. “first indication information”) to a user equipment.
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Yoon’s practice of sending reference signals with different densities on the same resources into Chavva’s method for using machine learning to generate CSI feedback. The ability to use different reference signal densities allows for increased flexibility in Chavva’s AI-based CSI reporting.
Claim 17 describes substantially the same subject matter as Claim 6 from the perspective of the UE
As to Claim 9:
Chavva teaches:
Sending, by the first device, a radio resource control (RRC) message to the second device
Paragraph 0091 of Chavva describes a UE “receiving a Radio Resource Configuration (RRC) message ... from the gNB” (Chavva, 0091).
The RRC message comprises ... information
Paragraph 0091 of Chavva describes a base station sending a RRC message to a UE containing “a feedback configuration for CSI-Reference Signals” (Chavva, 0091).
Chavva does not explicitly disclose:
Density configuration information of the first reference signal
However, Yoon does teach:
Density configuration information of the first reference signal
Paragraph 0012 of Yoon describes a base station configuring a UE with either a high-density or low-density reference signal configuration.
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Yoon’s practice of sending reference signals with different densities on the same resources into Chavva’s method for using machine learning to generate CSI feedback. The ability to use different reference signal densities allows for increased flexibility in Chavva’s AI-based CSI reporting.
As to Claim 10:
Chavva teaches:
Receiving, by the first device, a radio resource control RRC message from the second device
Paragraph 0091 of Chavva describes a base station sending a RRC message to a UE containing “a feedback configuration for CSI-Reference Signals” (Chavva, 0091).
The RRC message comprises ... information
Paragraph 0091 of Chavva describes a base station sending a RRC message to a UE containing “a feedback configuration for CSI-Reference Signals” (Chavva, 0091).
Chavva does not explicitly disclose:
Density configuration information of the first reference signal
However, Yoon does teach:
Density configuration information of the first reference signal
Paragraph 0012 of Yoon describes a base station configuring a UE with either a high-density or low-density reference signal configuration.
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Yoon’s practice of sending reference signals with different densities on the same resources into Chavva’s method for using machine learning to generate CSI feedback. The ability to use different reference signal densities allows for increased flexibility in Chavva’s AI-based CSI reporting.
As to Claim 13:
Chavva teaches:
Receiving, by the second device, the second reference signal from the first device
Paragraph 0177 of Chavva describes a base station sending a CSI-RS to a UE.
Sending, by the second device, third CSI to the first device
Paragraph 0179 of Chavva describes a UE periodically sending CSI reports which would include a “third CSI”).
The third CSI is obtained by the second device based on the second reference signal
Paragraph 0177 of Chavva clarifies that a UE generates CSI feedback using a “CSI-RS”.
The third CSI is used to train a neural network to obtain the first neural network model
Paragraph 0131 of Chavva describes training a neural network using a CSI report generated by the UE.
Chavva does not explicitly disclose:
Before receiving, by a second device, a first reference signal from a first device
However, Yoon does teach:
Before receiving, by a second device, a first reference signal from a first device
Paragraph 0012 of Yoon describes configuring a reference signal as high density or low density before sending a CSI-RS.
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Yoon’s practice of sending reference signals with different densities on the same resources into Chavva’s method for using machine learning to generate CSI feedback. The ability to use different reference signal densities allows for increased flexibility in Chavva’s AI-based CSI reporting.
Claim(s) 3-4 and 14 is/are rejected under 35 U.S.C. 103 as being unpatentable over Chavva (US 2021/0351885 A1) in view of Yoon (US 2021/0105111 A1) and Zhou (US 2020/0106498 A1) and further in view of Xue et al. (US 2021/0376895 A1, hereinafter “Xue”).
As to Claim 3 and 14:
Chavva does not explicitly disclose:
The third reference signal is a normal density reference signal
However, Yoon does teach:
The third reference signal is a normal density reference signal
Yoon describes a “CSI-RS” transmitted “according to the high-density configuration information” which is analogous to “a normal density reference signal” (Yoon, 0013).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate the standard density reference signal described in Yoon into Chavva’s method for using machine learning to estimate CSI. The standard density reference signal will remain readable in the presence of a reasonable amount of noise, ensuring that it can be demodulated and understood.
The combination of Chavva, Yoon, and Zhou also does not explicitly disclose:
Sending, by the first device, a third reference signal to the second device
Receiving, by the first device, fourth CSI from the second device
The fourth CSI is obtained by the second device based on the third reference signal
Training, by the first device, the neural network based on the fourth CSI, to obtain an updated first neural network model
However, Xue does describe a method for evaluating machine learning predictions of CSI.
Specifically, from this list, Xue at least teaches:
Sending, by the first device, a third reference signal to the second device
Paragraph 0006 of Xue describes a network device sending a CSI prediction model to a user equipment that it can use to “calculat[e] CSI based on downlink reference signal measurements” (Xue, 0006).
Receiving, by the first device, fourth CSI from the second device
Paragraph 0006 of Xue describes a UE reporting CSI measurements to a base station.
The fourth CSI is obtained by the second device based on the third reference signal
Paragraph 0006 of Xue further states that the UE obtains CSI based on a “downlink reference signal” (Xue, 0006).
Training, by the first device, the neural network based on the fourth CSI, to obtain an updated first neural network model
Paragraph 0053 describes the gNB sending a UE “an updated, retrained machine learning model” based on new CSI measurements.
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Xue’s practice of retraining a neural network based on updated CSI measurements into Chavva’s method for generating CSI feedback using machine learning. Continuing to update the model makes its predictions more refined and, presumably, more accurate.
Claim 14 describes substantially the same subject matter as claim 3 from the perspective of a user equipment.
As to Claim 4:
The combination of Chavva, Yoon, and Zhou does not explicitly disclose:
The preset trigger condition includes that the first device determines that demodulation performance of demodulating first data by the second device is less than a preset threshold, and the first data is sent by the first device based on the second CSI; or
The preset trigger condition includes that the first device receives a first request message from the second device, and the first request message is used to request to update the first neural network model
However, from this list, Xue at least teaches:
The preset trigger condition includes that the first device determines that demodulation performance of demodulating first data by the second device is less than a preset threshold, and the first data is sent by the first device based on the second CSI; or
Paragraph 0053 of Xue describes a base station reconfiguring a UE with an “updated, retrained machine learning model” if “the predictions are consistently inaccurate by more than a threshold amount” (Xue, 0053).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to incorporate Xue’s method for evaluating and updating a neural network into Chavva’s method for using a neural network to estimate CSI. Periodically retraining the neural network can make its prediction results more accurate as it continues to operate.
Claim(s) 7-8 and 18-19 is/are rejected under 35 U.S.C. 103 as being unpatentable over Chavva (US 2021/0351885 A1) in view of Yoon (US 2021/0105111 A1) and Zhou (US 2020/0106498 A1) and further in view of Ko et al. (US 2019/0140801 A1, hereinafter “Ko”).
As to Claim 7 and 18:
The combination of Chavva, Yoon, and Zhou does not explicitly disclose:
Based on the first device being a network device, the second request message is carried in uplink control information UCI; or
Based on the first device being a terminal device, the second request message is carried in downlink control information DCI
However, from this list, Ko does separately teach:
Based on the first device being a network device, the ... message is carried in uplink control information
Paragraphs 0071 and 0074 of Ko describe a UE sending uplink control information to the network.
And:
The second request message
Paragraph 0133 of Ko describes a UE “making a request for varying f RS density” (Ko, 0133).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to convey the request message for changing reference signal density through UCI, each of which Ko teaches separately, and incorporate them into Chavva’s method for measuring CSI using machine learning. A UCI is a common message format that can allow a terminal device to request an uplink device to change reference signal density, which is useful in Chavva’s method because it enables reference signals to use less power and resources when conditions allow.
Claim 18 describes substantially the same subject matter as Claim 7 from the perspective of the UE.
As to Claim 8 and 19:
Chavva does not explicitly disclose:
The first indication information
However, Yoon does teach:
The first indication information
Paragraph 0012 of Yoon describes configuring a UE with either a “low-density configuration” or a “high-density configuration” (Yoon, 0012).
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to include the control signaling disclosing reference signal density described in Yoon into Chavva’s method for determining an ML model to measure CSI. If reference signal density is variable, it makes sense to consider this when determining the model used to estimate CSI.
The combination of Chavva, Yoon, and Zhou also does not explicitly disclose:
Based on the first device being a network device, the first indication information is carried in downlink control information DCI; or
Based on the first device being a terminal device, the first indication information is carried in uplink control information
However, from this list, Ko does at least teach:
Based on the first device being a network device, the ... information is carried in downlink control information DCI
Paragraph 0063 of Ko describes a UE receiving DCI to allocate resources.
Thus, it would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to use the DCI disclosed in Ko to send downlink configuration information to the UE in Chavva’s method. DCI is a commonly available message format that a base station can use to configure a downlink device.
Claim 19 describes substantially the same subject matter as claim 8 from the perspective of the UE
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Benjamin Peter Welte whose telephone number is (703)756-5965. The examiner can normally be reached Monday - Friday, 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, Chirag Shah, can be reached at (571)272-3144. 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.
/B.P.W./Examiner, Art Unit 2477
/CHIRAG G SHAH/Supervisory Patent Examiner, Art Unit 2477