DETAILED ACTION
Notice of Pre-AIA or AIA Status
The present application, filed on or after March 16, 2013, is being examined under the first inventor to file provisions of the AIA .
Information Disclosure Statement
The information disclosure statement (IDS) submitted on 11/15/24 and 7/16/26 is being considered by the examiner.
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) 1-20 is/are rejected under 35 U.S.C. 103 as being unpatentable over Karjeet e al. (US 2022/0311678) in view of Chong (US 2025/0300906).
Regarding claim 1, Karjee teaches a wireless transmit/receive unit (WTRU) (IoT device) comprising:
a processor configured to:
determine machine-task context information for execution of a machine-type task (deep neural network (DNN) execution) (see “At operation 305, the method 300 includes determining, by the IoT device 101, a split ratio the DNN based on at least one of an inference time of the DNN and a transmission time required for transmitting output of each layer of the DNN from the IoT device to the selected at least one edge device” [0058]), wherein the machine-task context information comprises at least one of application performance information, WTRU performance information (see “In an embodiment, the one or more inference parameter may comprise a predetermined computation time of each layer of the DNN on the IoT device obtained by benchmarking the DNN on the IoT device” [0058]), edge server (edge device) performance information (see “In an embodiment, the one or more inference parameter may comprise, a predetermined computation time of each layer of the DNN on the selected at least one edge device obtained by benchmarking the DNN on the selected at least one edge device” [0058]), or network (NW) performance information (throughput) (see “In an embodiment, the one or more inference parameter may comprise the throughput of the identified network determined based on a response time of a message sent from the IoT device to the selected at least one edge device” [0058]);
receive NW-related parameters, wherein the NW-related parameters comprise at least one of channel bandwidth (see “At operation 303, the method 300 includes identifying, by the IoT device 101, a network for connecting the IoT device with the selected at least one edge device based on available bandwidth and historical inference time records of a plurality of networks associated with the IoT device” [0057]), WTRU transmission power limits, or end-to-end latency requirements (latency constraints) for executing the machine-type task (see “In an embodiment, the method of disclosure helps in overcoming the incompatibility associated with remotely deployed cloud due to latency constraints and unreliable connectivity during poor network conditions by deploying the DNN to edge devices in the close proximity of IoT devices and splitting the DNN among the IoT and edge devices using dynamic split computing (DSC)” [0145]);
determine an inference method (remote or partial split) for the machine-type task based on the machine-task context information and the NW-related parameters (see “The computation may happen in two modes, namely, a remote mode and a partial split mode. The remote mode involves offloading the entire computation to the edge devices. Whereas in the partial split mode, the task is partitioned between the IoT device, and the edge device based on the computation power of the edge and the available network throughput” [0108] and “Referring to FIG. 4A, a dynamic split computation (DSC) mechanism is used to find an optimal splitting point of DNN layers 400. The splitting point divides the DNN layers into two parts. One inference part is computed by the IoT device 101, and another inference part is computed by the edge device 103 which is partitioned based on available bandwidth” [0061]); and
transmit an indication (second part of DNN) of the inference method to at least one of the NW or a remote server (see “In an embodiment, the splitting module 229 may be configured for splitting the DNN according to the optimal split ratio and transmitting the second part of the DNN to the selected edge device, wherein the first part of the DNN is executed on the IoT device and the second part of the DNN is executed on the selected edge device” [0052] wherein the second part of the split DNN is an “indication of the inference method” since it indicates a partial split mode to the selected edge device).
Karjee does not explicitly teach wherein the indication comprises at least one of a validity period or predicted network resource requirements for the inferencing method. In an analogous prior art reference, Chong teaches an indication (ML model) comprises at least one of a validity period (valid time) or predicted network resource requirements for the inferencing method (see “In addition, the foregoing ML model may be provided by a model providing network element (also referred to as a model provider)” [0075] which includes “The model usage scope requirement information may indicate the scope information of a model, such as the effective region, applicable data network name (Data Network Name, DNN), applicable slice, and valid time” [0097] which suggests a valid time may be indicated along with a ML model which is used for performing model inference). Therefore it would have been obvious to one of ordinary skill in the art before the effective filing date of the clamed invention to modify the invention of Karjee to allow the indication to comprise at least one of a validity period or predicted network resource requirements for the inferencing method, as taught by Chong, in order to specify a valued time period to perform the remote or partial split mode to meet latency requirements.
Regarding claim 2, Karjee teaches the WTRU of claim 1, wherein the application performance information comprises at least one of observed application round trip time (total inference time) (see total inference time is the inference time on edge to compute DNN layers [0061] and transport time from IoT device to edge device [0061] and transport time from edge to IoT device [0062] which reads on the claimed “observed application round trip time“), application round trip time thresholds, or data size (output size) related to the machine-type tasks (see “Since transport time will depend on the output size, given that network is consistent, the total inference time for each layer is identified just from output size of that layer” [0086]).
Regarding claim 3, Karjee teaches the WTRU of claim 1, wherein the WTRU performance information comprises at least one of compute delay (computation time) or computation load related to the execution of the machine-type tasks (see “In an embodiment, the one or more inference parameter may comprise a predetermined computation time of each layer of the DNN on the IoT device obtained by benchmarking the DNN on the IoT device” [0058]).
Regarding claim 4, Karjee teaches the WTRU of claim 1, wherein the edge server performance information comprises at least one of compute delay (computational latency) or computation load related to the execution of the machine-type tasks (see “In an embodiment, the computational latency 213 may indicate the amount of time the plurality of edge devices takes for a packet of data to be captured, transmitted, processed through the plurality of edge devices” [0045]).
Regarding claim 5, Karjee teaches the WTRU of claim 1, wherein the NW performance information comprises at least one of transport layer congestion (see “In an embodiment, the one or more inference parameter may comprise the throughput of the identified network determined based on a response time of a message sent from the IoT device to the selected at least one edge device” [0058] wherein throughput reads on the claimed “transport layer congestion” since the parameters are related (i.e. a higher throughput may indicate low or no congestion), packet drops, or buffer status related to the execution of the machine-type tasks.
Regarding claim 6, Karjee teaches the WTRU of claim 1, wherein the NW-related parameters comprise at least one of allocated bandwidth (available bandwidth) (see “At operation 303, the method 300 includes identifying, by the IoT device 101, a network for connecting the IoT device with the selected at least one edge device based on available bandwidth and historical inference time records of a plurality of networks associated with the IoT device” [0057]), NW backhaul latency, or packet drops.
Regarding claim 7, Karjee and Chong in combination teaches the WTRU of claim 1, wherein the processor is configured to:
determine the validity period of the inference method based on at least one of a number of slots, frames, or milliseconds for which the local, remote, or split inferencing method is determined to be valid (in the relied upon combination of Karjee and Chong, the remote or partial split mode of Karjee is modified to include a valid time as taught by Chong and therefore reads on a time for which a “remote, or split inferencing method is determined to be valid”).
Regarding claim 8, Karjee teaches the WTRU of claim 1, wherein the processor is configured to:
determine the inference method based on at least one of machine-task quality of service (QoS) requirements, a WTRU environment, or a wireless channel condition (network condition information) (see “As an example, the one or more network condition information 211 may include, without limitation, available bandwidth and historical inference time records” [0044]), wherein the inference method comprises local inferencing (on-device inference), remote inferencing (on-edge inference), or split inferencing (see “The FIG. 8, compares the inference time (in ms) of PoseNet model computed in three ways: on-device inference (i.e., completely on the RPi), on-edge inference (i.e., complete offload to the Galaxy S20 device) and split inference using the E-DSC algorithm over 500 iterations” [0129]).
Regarding claim 9, Karjee teaches the WTRU of claim 8, wherein:
the WTRU environment comprises at least one of a WTRU location, a number of objects near the WTRU, characteristics of the objects near the WTRU, or atmospheric conditions that affect machine-task application performance (this limitation is not given patentable weight since claim 8 from which this claim depends is written in an alternative "or" format which requires the selection of either A or B in order to meet the claimed limitations as is indicated by MPEP 2143.03 & MPEP 2111.04 Section ll); and
the wireless channel condition comprises at least one of a channel quality indicator (CQI), a reference signal received power (RSRP), or a path loss (see “In an embodiment, another approach to find anoptimal splitting point, a deep learning model with N layers is considered. Initially IoT device chooses the nearest neighbor edge based on path-loss model to execute K optimal split computing points among IoT and edge to compute the total inference time (ITotal)…” [0064]).
Regarding claim 10, Karjee teaches the WTRU of claim 1, wherein the indication of the inference method comprises at least one of predicted bandwidth requirements, throughput, or expected round-trip time for executing the inference method (these limitations are not given patentable weight since when viewed with claim 1 as a whole, they are recited in an alternative or “or” format (i.e. the indication of the inference method comprises at least one of a validity period, predicted network resource requirements, predicted bandwidth requirements, throughput, or expected round-trip time) and the combination of Karjee and Chong teaches the validity period as indicated above).
Claim 11 recites subject matter similar to claim 1 and is therefore rejected on the same basis.
Claim 12 recites subject matter similar to claim 2 and is therefore rejected on the same basis.
Claim 13 recites subject matter similar to claim 3 and is therefore rejected on the same basis.
Claim 14 recites subject matter similar to claim 4 and is therefore rejected on the same basis.
Claim 15 recites subject matter similar to claim 5 and is therefore rejected on the same basis.
Claim 16 recites subject matter similar to claim 6 and is therefore rejected on the same basis.
Claim 17 recites subject matter similar to claim 7 and is therefore rejected on the same basis.
Claim 18 recites subject matter similar to claim 8 and is therefore rejected on the same basis.
Claim 19 recites subject matter similar to claim 9 and is therefore rejected on the same basis.
Claim 20 recites subject matter similar to claim 10 and is therefore rejected on the same basis.
Conclusion
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Nam T Huynh whose telephone number is (571)272-5970. The examiner can normally be reached 9am-5pm.
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, Alison Slater can be reached at 571-270-0375. 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.
/NAM T HUYNH/Primary Examiner, Art Unit 2647
Prosecution Insights