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 .
DETAILED ACTION
Claims 1-20 are presented for examination.
Claim Rejections - 35 USC § 102
The following is a quotation of the appropriate paragraphs of 35 U.S.C. 102 that form the basis for the rejections under this section made in this Office action:
A person shall be entitled to a patent unless –
(a)(1) the claimed invention was patented, described in a printed publication, or in public use, on sale or otherwise available to the public before the effective filing date of the claimed invention.
(a)(2) the claimed invention was described in a patent issued under section 151, or in an application for patent published or deemed published under section 122(b), in which the patent or application, as the case may be, names another inventor and was effectively filed before the effective filing date of the claimed invention.
Claims 1 and 11 are rejected under 35 U.S.C. 102(a)(1) as anticipated by Higuchi (US 2022/0116456 A1)
As per claim 1, Higuchi teaches A method for allocating computing resources for a vehicle, the method comprising:
determining an optimal task configuration for a computing task based at least in part on a task constraint of the computing task; (Higuchi [0021] Using the state vector and the computational task identifier, the one or more processors, through the use of a utility function, determine a utility score of the computational task. The utility function may be a machine learning algorithm that was trained using reinforcement learning to anticipate the expected value of offloading each computational task to an external system [0043] In one example, the utility function(s) 250 employees a reinforcement learning algorithm to learn the optimal decision policy. The utility function(s) 250 may be trained from the past history of <state, action, reward> tuples, where the state is the state vector for the time the offloading decision was made, the action signifies if the task was (a) offloaded (b) locally processed or (c) discarded and the reward indicates the gain in terms of (i) improvement of application performance and (ii) savings of network/compute resources. In some embodiments, one may train the utility function by reinforcement learning (e.g., Q learning). Different applications and/or computational tasks may have different criteria on the task utility. Therefore, the utility function(s) 250 are preferably trained on a per-task-type basis. See paragraph 22 below for constraints of the tasks. See also paragraph 50 for further info about constraints and their role in state vector used for optimization).
The examiner believes this interpretation of optimality is consistent with what is disclosed in the specification ([0015] to determine the optimal task configuration, the vehicle controller is further programmed to determine a predicted performance of the computing task using a vehicle specific offloading machine learning model.)
determining a criticality level of the computing task based at least in part on the optimal task configuration and the task constraint of the computing task;
(Higuchi [0022] The utility score generally represents an improvement in the functioning of the application if the computational task is offloaded to an external system for processing. For example, if the application is an object detection application, the utility score could indicate a general difference between if the computational task associated with the application is performed by the vehicle processor or is offloaded to be processed by an external system. For example, the vehicle processor may only be able to execute a lightweight version of an object detection algorithm, while the external system could execute a much more computationally intensive [task constraint] but much more accurate object detection algorithm [0024] As such, the system and method for value-anticipating task offloading provides a solution for task offloading such that each computational task is evaluated to determine which computational tasks are the most important [criticality of the task] and would add the most value if they are offloaded. The computational tasks that are identified as having the most value [critical tasks] if they are offloaded can then be prioritized such that they are offloaded.)
routing the computing task to one of a plurality of remote server systems based at least in part on the criticality level of the computing task. (Higuchi Fig 6 Block 408 (offload the computational task to the external system for processing) and [0063] In step 406, the task manager module 220 may then cause the processor(s) 110 to evaluate the utility score and determine if the computational task should be offloaded to an external system for processing (step 408),)
Claim 11 is the same Fig 2B shows a plurality of remote server systems and communication between vehicles and remote servers. Fig 3 and 4 shows details of the vehicle controller (Fig 4 Block 230 shows the task scheduler)
As to claim 11, it is rejected based on the same reason as claim 1.
Claim Rejections - 35 USC § 103
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 of this title, 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 2 and 3 are rejected under 35 U.S.C. 103 as being unpatentable over Higuchi (US 2022/0116456 A1) in view of Sorrentino (US 2024/0414050 A1).
As per claim 2, Higuchi does not teach determining a predicted performance of the computing task based at least in part on a task configuration of the computing task; comparing the predicted performance of the computing task to the task constraint of the computing task; modifying the task configuration of the computing task in response to determining that the predicted performance of the computing task does not satisfy the task constraint of the computing task; and repeating the determining the predicted performance step, the comparing the predicted performance step, and the modifying the task configuration step until the optimal task configuration is identified, wherein the predicted performance of the computing task with the optimal task configuration satisfies the task constraint of the computing task.
However, Sorrentino teaches determining a predicted performance of the computing task based at least in part on a task configuration of the computing task; comparing the predicted performance of the computing task to the task constraint of the computing task; modifying the task configuration of the computing task in response to determining that the predicted performance of the computing task does not satisfy the task constraint of the computing task; and repeating the determining the predicted performance step, the comparing the predicted performance step, and the modifying the task configuration step until the optimal task configuration is identified, wherein the predicted performance of the computing task with the optimal task configuration satisfies the task constraint of the computing task. (Sorrentino Fig 5 and Fig 6 and [0043] In step S603, the UE adapts its behaviour based on the QoS prediction. As discussed above, where the device is a remote driving vehicle this adaptation could take the form of, for example, rerouting the vehicle to avoid an area of poor network coverage. Continuing with the example discussed above with reference to step S600, a UE interested in operating a video streaming application may reduce the resolution of the streamed video if the QoS prediction indicates that the available bandwidth would not support a high resolution video stream. The QoS prediction is analysed and a determination of the feasibility of the prediction (that is, the acceptability of the success metric) is generated and sent via the predictor to the UE in step S604. The sending of the feasibility information to the UE is optional, but may be of use if the UE is to adapt its behaviour on a short time scale).
It would have been obvious to a person in the ordinary skill in the art before the filing date of the claimed invention to combine Sorrentino with the system of Higuchi to modify the task configuration step. One having ordinary skill in the art would have been motivated to use Sorrentino into the system of Higuchi for the purpose of using QoS analysis in edge networks. (Sorrentino paragraph 02)
As per claim 3, Higuchi teaches determining the predicted performance of the computing task using a vehicle specific offloading machine learning model, wherein the vehicle specific offloading machine learning model is configured to receive the task configuration of the computing task and historical vehicle specific task performance data as an input, and wherein the vehicle specific offloading machine learning model is configured to provide the predicted performance of the computing task as an output. (Higuchi Fig 6 shows the one vehicle model, whereas Fig 7 shows a multi-vehicle model.. Here use Liu [0003] Computing tasks associated with the generated data may be offloaded by a vehicle for scheduling and processing at cloud and edge servers to reduce system delays. Dedicated short-range communication (DSRC) and cellular networks, for example, may be used by the vehicle to accelerate task processing by wirelessly sending the computing task to a nearby edge server. In this configuration, powerful computing servers in centralized data centers or distributed network edge servers may schedule and process complex computing tasks offloaded by the vehicle. Fully offloading vehicle computing tasks to a server for execution may excessively use wireless transmission bandwidth and system resources. Also, local processing of most tasks on the vehicle could lead to system delays due to limited computing power, memory resources, storage size, bandwidth, or the like.[0005] In one embodiment, example systems and methods relate to a manner of improving the scheduling of computing tasks of a vehicle to execute on a server. A scheduling system efficiently scheduling an offloaded computing task may be suboptimal at fairly reducing delays. In one approach, a scheduling system may allocate the server resources for offloaded computing tasks more efficiently by analyzing context information of a vehicle and machine learning. Therefore, an improved scheduling system is disclosed that schedules computing tasks in a manner that reduces end-to-end delay and improves fairness by analyzing the context information associated with the vehicle. In one approach, a scheduling system may fairly allocate resources of a server using machine learning. Moreover, the scheduling system may schedule a computing task such that context information and a task descriptor in an offloading request satisfy fair resource loading between the vehicle and the server. In one approach, the computing task may be partitioned into subtasks by a machine learning module according to the context information. A scheduling message is sent to the vehicle with scheduling and task partition information to offload subtasks to the server for execution. Thus, the scheduling system uses context information and machine learning to improve resource usage for the server for offloaded computing tasks to reduce delays and fairly allocate resources. [0048] For making an offloading scheduling decision, the vehicle scheduling module or the task partitioning module may utilize historical information, system status information, vehicle(s) status information, or the like stored in database 530. For example, system status information may be received from infrastructure component 540 to determine available server computation resources to execute offloaded tasks. In one approach, the infrastructure component 540 may be an edge server, a part of distributed network edge servers, or the like that is co-located or part of controller 520 (not shown). In other approaches, the infrastructure component 540 may be an edge server, a part of distributed network edge servers, or the like remote or independent of controller 520. In another approach, both controller 520 and infrastructure component 540 may be part of or integrated in a cellular base station, Node-b, eNodeB, or the like.)
Claims 4 and 5 are rejected under 35 U.S.C. 103 as being unpatentable over Higuchi (US 2022/0116456 A1) in view of Sorrentino (US 2024/0414050 A1) in further view of Chen (US 2024/0127105 A1).
As per claim 4, Higuchi and Sorrentino do not teach wherein the method further comprises training the vehicle specific offloading machine learning model, wherein training the vehicle specific offloading machine learning model further comprises: training a global offloading machine learning model based at least in part on historical global task performance data, wherein the global offloading machine learning model is trained using one of the plurality of remote server systems; deploying the global offloading machine learning model from the one of the plurality of remote server systems to a vehicle controller of the vehicle; and training the vehicle specific offloading machine learning model based at least in part on the global offloading machine learning model using the vehicle controller, wherein the vehicle specific offloading machine learning model is trained using remote learning.
However, Chen teaches training a global offloading machine learning model based at least in part on historical global task performance data, wherein the global offloading machine learning model is trained using one of the plurality of remote server systems; deploying the global offloading machine learning model from the one of the plurality of remote server systems to a vehicle controller of the vehicle; and training the vehicle specific offloading machine learning model based at least in part on the global offloading machine learning model using the vehicle controller, wherein the vehicle specific offloading machine learning model is trained using remote learning. (Chen [0019] In embodiments, the server 106 considers heterogeneity of the edge nodes, i.e., different sensors and different computing resources of the edge nodes when computing a global model based on the updated local models. For example, the server 106 considers metadata about vehicles received from the vehicles when determining weights for local models. The metadata includes, but not limited to, quality of data that corresponding vehicle uses for training, the number of sensors that corresponding vehicle has, a computing power of a processor of the corresponding vehicle, and the like. Details about computing a global model based on the updated local models will be described with reference to FIGS. 2-4 below. [0047] In FIG. 4, the system includes three vehicles 410, 420, 430 and an edge server 440. The system may include more than or less than three vehicles. In step S402, the edge server 440 initializes a global model. In step S404, the edge server 440 transmits the initialized global model to each of the vehicles 410, 420, 430. In step S406, each of the vehicles 410, 420, 430 trains the initialized global model using its local data such as images captured by the vehicles 410, 420, 430, respectively. In step S408, each of the vehicles 410, 420, 430 transmits the trained model and metadata for the vehicles 410, 420, 430 to the edge server 440. The metadata includes, but not limited to, quality of data that corresponding vehicle uses for training, the number of sensors that corresponding vehicle has, a computing power of a processor of the corresponding vehicle, and the like. The metadata may include the type of a vehicle (e.g., sedan, SUV, truck, etc.), the model of the vehicle, the year of the vehicle, and the like. In some embodiments, the metadata may include the location of the vehicle when images as training data were captured, the time when the images were captured, information on whether when the images were captured, and the like [0048] In step S410, the edge server 440 analyzes the contributions of the trained local models of the vehicles 410, 420, 430 based on the metadata and fuses the trained local models based the contributions to obtain an aggregated global model. In step S412, the edge server 440 transmits the aggregated global model to the vehicles 410, 420, 430. Then, each of the vehicles 410, 420, 430 again trains the aggregated global model using its local data. In addition, each of the vehicles 410, 420, 430 may perform vision-based lane centering using the aggregated global model or autonomous driving).
It would have been obvious to a person in the ordinary skill in the art before the filing date of the claimed invention to combine Chen with the system of Higuchi and Sorrentino to use a global and vehicle specific machine learning model. One having ordinary skill in the art would have been motivated to use Chen into the system of Higuchi and Sorrentino for the purpose of using a vehicular network that takes into account heterogeneous edge nodes that differ in computation resource and hardware elements of the edge nodes (Chen paragraph 03)
As per claim 5, Chen teaches training the global offloading machine learning model with historical vehicle specific task performance data to produce the vehicle specific offloading machine learning model. (Chen [0047] In FIG. 4, the system includes three vehicles 410, 420, 430 and an edge server 440. The system may include more than or less than three vehicles. In step S402, the edge server 440 initializes a global model. In step S404, the edge server 440 transmits the initialized global model to each of the vehicles 410, 420, 430. In step S406, each of the vehicles 410, 420, 430 trains the initialized global model using its local data such as images captured by the vehicles 410, 420, 430, respectively. In step S408, each of the vehicles 410, 420, 430 transmits the trained model and metadata for the vehicles 410, 420, 430 to the edge server 440. The metadata includes, but not limited to, quality of data that corresponding vehicle uses for training, the number of sensors that corresponding vehicle has, a computing power of a processor of the corresponding vehicle, and the like. The metadata may include the type of a vehicle (e.g., sedan, SUV, truck, etc.), the model of the vehicle, the year of the vehicle, and the like. In some embodiments, the metadata may include the location of the vehicle when images as training data were captured, the time when the images were captured, information on whether when the images were captured, and the like [0048] In step S410, the edge server 440 analyzes the contributions of the trained local models of the vehicles 410, 420, 430 based on the metadata and fuses the trained local models based the contributions to obtain an aggregated global model. In step S412, the edge server 440 transmits the aggregated global model to the vehicles 410, 420, 430. Then, each of the vehicles 410, 420, 430 again trains the aggregated global model using its local data. In addition, each of the vehicles 410, 420, 430 may perform vision-based lane centering using the aggregated global model or autonomous driving)
Claim 7 is rejected under 35 U.S.C. 103 as being unpatentable over Higuchi (US 2022/0116456 A1) in view of Sorrentino (US 2024/0414050 A1) in further view of Liu (US 2021/0406065 A1).
As per claim 7, Higuchi and Sorrentino do not teach modifying the task configuration of the computing task using the vehicle specific offloading machine learning model, wherein the vehicle specific offloading machine learning model is further configured to receive the task constraint of the computing task as the input and provide a modified task configuration for the computing task as the output.
However, Liu teaches modifying the task configuration of the computing task using the vehicle specific offloading machine learning model, wherein the vehicle specific offloading machine learning model is further configured to receive the task constraint of the computing task as the input and provide a modified task configuration for the computing task as the output. (Liu [0003] Computing tasks associated with the generated data may be offloaded by a vehicle for scheduling and processing at cloud and edge servers to reduce system delays. Dedicated short-range communication (DSRC) and cellular networks, for example, may be used by the vehicle to accelerate task processing by wirelessly sending the computing task to a nearby edge server. In this configuration, powerful computing servers in centralized data centers or distributed network edge servers may schedule and process complex computing tasks offloaded by the vehicle. Fully offloading vehicle computing tasks to a server for execution may excessively use wireless transmission bandwidth and system resources. Also, local processing of most tasks on the vehicle could lead to system delays due to limited computing power, memory resources, storage size, bandwidth, or the like.[0005] In one embodiment, example systems and methods relate to a manner of improving the scheduling of computing tasks of a vehicle to execute on a server. A scheduling system efficiently scheduling an offloaded computing task may be suboptimal at fairly reducing delays. In one approach, a scheduling system may allocate the server resources for offloaded computing tasks more efficiently by analyzing context information of a vehicle and machine learning. Therefore, an improved scheduling system is disclosed that schedules computing tasks in a manner that reduces end-to-end delay and improves fairness by analyzing the context information associated with the vehicle. In one approach, a scheduling system may fairly allocate resources of a server using machine learning. Moreover, the scheduling system may schedule a computing task such that context information and a task descriptor in an offloading request satisfy fair resource loading between the vehicle and the server. In one approach, the computing task may be partitioned into subtasks by a machine learning module according to the context information. A scheduling message is sent to the vehicle with scheduling and task partition [example of task configuration modification] information to offload subtasks to the server for execution. Thus, the scheduling system uses context information and machine learning to improve resource usage for the server for offloaded computing tasks to reduce delays and fairly allocate resources. [0048] For making an offloading scheduling decision, the vehicle scheduling module or the task partitioning module may utilize historical information, system status information, vehicle(s) status information, or the like stored in database 530. For example, system status information may be received from infrastructure component 540 to determine available server computation resources to execute offloaded tasks. In one approach, the infrastructure component 540 may be an edge server, a part of distributed network edge servers, or the like that is co-located or part of controller 520 (not shown). In other approaches, the infrastructure component 540 may be an edge server, a part of distributed network edge servers, or the like remote or independent of controller 520. In another approach, both controller 520 and infrastructure component 540 may be part of or integrated in a cellular base station, Node-b, eNodeB, or the like.
It would have been obvious to a person in the ordinary skill in the art before the filing date of the claimed invention to combine Liu with the system of Higuchi and Sorrentino to use vehicle specific offloading machine learning model. One having ordinary skill in the art would have been motivated to use Liu into the system of Higuchi and Sorrentino for the purpose of improving the scheduling of computing tasks within a vehicle to execute on a server. (Liu paragraph 01)
Claim 8 is rejected under 35 U.S.C. 103 as being unpatentable over Higuchi (US 2022/0116456 A1) in view of Sorrentino (US 2024/0414050 A1) in further view of Bernat (US 2020/0389410 A1) and De Angelo (US 12,340,395 B2)
As per claim 8, Higuchi and Sorrentino do not teach determining the criticality level of the computing task based at least in part on the predicted performance of the computing task with the optimal task configuration, wherein the criticality level includes one of: a low criticality level, a normal criticality level, a high criticality level, and a very high criticality level.
However, Bernat teaches determining the criticality level of the computing task based at least in part on the predicted performance of the computing task with the optimal task configuration, wherein the criticality level includes one of: a low criticality level, a normal criticality level, a high criticality level, and a very high criticality level; (Bernat [0017] The example “Result Time” service parameter specifies a deadline or a round-trip latency defining a time-constraint by which a result of a function as a service is expected to be received at a requesting client device 105. As such, the “Result Time” service parameter corresponds to a deadline service parameter or a round-trip latency service parameter of the service request 104. In other examples, the “Result Time” service parameter may be omitted from the service request 104, and a deadline service parameter or a round-trip latency service parameter may instead be provided via the “SLA ID” service parameter as described below. The example “Priority ID” service parameter is to identify a priority level or priority policy that specifies a priority that should be assigned to the corresponding service request 104. In some examples, the “priority ID” service parameter is expressed as a “high priority,” a “normal priority,” a “low priority,” and/or a numeric-based priority (e.g., “1” being the highest priority and “10” being the lowest priority))
The specification does not establish standards that distinguish among these criticality levels. For purposes of examination, they will be treated as different levels of criticality.
It would have been obvious to a person in the ordinary skill in the art before the filing date of the claimed invention to combine Bernat with the system of Higuchi and Sorrentino to use tasks of varying criticality. One having ordinary skill in the art would have been motivated to use Bernat into the system of Higuchi and Sorrentino for the purpose of scheduling service requests in a network computing system using hardware queue managers. (Bernat paragraph 01)
Bernat does not teach comparing the criticality level to an allowed criticality level of the task constraint; and modifying the criticality level in response to determining that the criticality level does not satisfy the allowed criticality level.
However, DeAngelo teaches comparing the criticality level to an allowed criticality level of the task constraint; and modifying the criticality level in response to determining that the criticality level does not satisfy the allowed criticality level. (De Angelo [claim 25]. The system of claim 18, wherein the robotic devices or vehicles are configured to adjust their operations autonomously in response to detected events, including modifying their task prioritization, movement trajectory, or synchronization of tasks with the other robotic devices or vehicles).
It would have been obvious to a person in the ordinary skill in the art before the filing date of the claimed invention to combine De Angelo with the system of Higuchi and Sorrentino and Bernat to compare criticality level. One having ordinary skill in the art would have been motivated to use De Angelo into the system of Higuchi and Sorrentino and Bernat for the purpose of improving user engagement through responsive iterative and cyclic interaction using mobile devices. (DeAngelo col 1, lines 39-41)
Claim 10 is rejected under 35 U.S.C. 103 as being unpatentable over Higuchi (US 2022/0116456 A1) in view of Sorrentino (US 2024/0414050 A1) in further view of Ding (US 2020/0178198 A1)
As per claim 10, Higuchi and Sorrentino do not teach wherein the task constraint of the computing task includes at least a maximum end-to-end roundtrip latency for the computing task.
However, Ding teaches wherein the task constraint of the computing task includes at least a maximum end-to-end roundtrip latency for the computing task. (Ding [0150] In various embodiments, the MEC-O 1321 selects one or more MEC servers 1236 for computational intensive tasks. The selected one or more MEC servers 1236 may offload computational tasks of a UE application 1305 based on various operational parameters, such as network capabilities and conditions, computational capabilities and conditions, application requirements, and/or other like operational parameters. The application requirements may be rules and requirements associated to/with one or more MEC Apps 1336, such as deployment model of the application (e.g., whether it is one instance per user, one instance per host, one instance on each host, etc.); required virtualized resources (e.g., compute, storage, network resources, including specific hardware support); latency requirements (e.g., maximum latency, how strict the latency constraints are, latency fairness between users);
It would have been obvious to a person in the ordinary skill in the art before the filing date of the claimed invention to combine Ding with the system of Higuchi to use a maximum end-to-end roundtrip latency for the computing task. One having ordinary skill in the art would have been motivated to use Ding into the system of Higuchi for the purpose of using edge Computing technologies for supporting vehicle-to-everything (V2X) communications. (Ding paragraph 02)
Claims 12 and 13 are rejected under 35 U.S.C. 103 as being unpatentable over Higuchi (US 2022/0116456 A1) in view of Liu (US 2021/0406065 A1) in further view of Sorrentino (US 2024/0414050 A1).
As per claim 12, Higuchi does not teach determine a predicted performance of the computing task using a vehicle specific offloading machine learning model, wherein the vehicle specific offloading machine learning model is configured to receive a task configuration of the computing task and historical vehicle specific task performance data as an input, and wherein the vehicle specific offloading machine learning model is configured to provide the predicted performance of the computing task as an output;
However, Liu teaches determine a predicted performance of the computing task using a vehicle specific offloading machine learning model, wherein the vehicle specific offloading machine learning model is configured to receive a task configuration of the computing task and historical vehicle specific task performance data as an input, and wherein the vehicle specific offloading machine learning model is configured to provide the predicted performance of the computing task as an output; (Liu [0003] Computing tasks associated with the generated data may be offloaded by a vehicle for scheduling and processing at cloud and edge servers to reduce system delays. Dedicated short-range communication (DSRC) and cellular networks, for example, may be used by the vehicle to accelerate task processing by wirelessly sending the computing task to a nearby edge server. In this configuration, powerful computing servers in centralized data centers or distributed network edge servers may schedule and process complex computing tasks offloaded by the vehicle. Fully offloading vehicle computing tasks to a server for execution may excessively use wireless transmission bandwidth and system resources. Also, local processing of most tasks on the vehicle could lead to system delays due to limited computing power, memory resources, storage size, bandwidth, or the like.[0005] In one embodiment, example systems and methods relate to a manner of improving the scheduling of computing tasks of a vehicle to execute on a server. A scheduling system efficiently scheduling an offloaded computing task may be suboptimal at fairly reducing delays. In one approach, a scheduling system may allocate the server resources for offloaded computing tasks more efficiently by analyzing context information of a vehicle and machine learning. Therefore, an improved scheduling system is disclosed that schedules computing tasks in a manner that reduces end-to-end delay and improves fairness by analyzing the context information associated with the vehicle. In one approach, a scheduling system may fairly allocate resources of a server using machine learning. Moreover, the scheduling system may schedule a computing task such that context information and a task descriptor in an offloading request satisfy fair resource loading between the vehicle and the server. In one approach, the computing task may be partitioned into subtasks by a machine learning module according to the context information. A scheduling message is sent to the vehicle with scheduling and task partition information to offload subtasks to the server for execution. Thus, the scheduling system uses context information and machine learning to improve resource usage for the server for offloaded computing tasks to reduce delays and fairly allocate resources. [0048] For making an offloading scheduling decision, the vehicle scheduling module or the task partitioning module may utilize historical information, system status information, vehicle(s) status information, or the like stored in database 530. For example, system status information may be received from infrastructure component 540 to determine available server computation resources to execute offloaded tasks. In one approach, the infrastructure component 540 may be an edge server, a part of distributed network edge servers, or the like that is co-located or part of controller 520 (not shown). In other approaches, the infrastructure component 540 may be an edge server, a part of distributed network edge servers, or the like remote or independent of controller 520. In another approach, both controller 520 and infrastructure component 540 may be part of or integrated in a cellular base station, Node-b, eNodeB, or the like.)
It would have been obvious to a person in the ordinary skill in the art before the filing date of the claimed invention to combine Liu with the system of Higuchi to use a vehicle specific offloading machine learning model. One having ordinary skill in the art would have been motivated to use Liu into the system of Higuchi for the purpose of improving the scheduling of computing tasks within a vehicle to execute on a server. (Liu paragraph 01)
Liu does not teach compare the predicted performance of the computing task to the task constraint of the computing task; and modify the task configuration of the computing task in response to determining that the predicted performance of the computing task does not satisfy the task constraint of the computing task.
However, Sorrentino teaches compare the predicted performance of the computing task to the task constraint of the computing task; and modify the task configuration of the computing task in response to determining that the predicted performance of the computing task does not satisfy the task constraint of the computing task. (Sorrentino Fig 5 and Fig 6 and [0043] In step S603, the UE adapts its behaviour based on the QoS prediction. As discussed above, where the device is a remote driving vehicle this adaptation could take the form of, for example, rerouting the vehicle to avoid an area of poor network coverage. Continuing with the example discussed above with reference to step S600, a UE interested in operating a video streaming application may reduce the resolution of the streamed video if the QoS prediction indicates that the available bandwidth would not support a high resolution video stream. The QoS prediction is analysed and a determination of the feasibility of the prediction (that is, the acceptability of the success metric) is generated and sent via the predictor to the UE in step S604. The sending of the feasibility information to the UE is optional, but may be of use if the UE is to adapt its behaviour on a short time scale).
It would have been obvious to a person in the ordinary skill in the art before the filing date of the claimed invention to combine Sorrentino with the system of Higuchi and Liu to compare the predicted performance of the computing task to the task constraint of the computing task. One having ordinary skill in the art would have been motivated to use Sorrentino into the system of Higuchi and Liu for the purpose of using QoS analysis in edge networks. (Sorrentino paragraph 02)
As per claim 13, Liu teaches modify the task configuration of the computing task using the vehicle specific offloading machine learning model, wherein the vehicle specific offloading machine learning model is further configured to receive the task constraint of the computing task as the input and provide a modified task configuration for the computing task as the output. (Liu [0003] Computing tasks associated with the generated data may be offloaded by a vehicle for scheduling and processing at cloud and edge servers to reduce system delays. Dedicated short-range communication (DSRC) and cellular networks, for example, may be used by the vehicle to accelerate task processing by wirelessly sending the computing task to a nearby edge server. In this configuration, powerful computing servers in centralized data centers or distributed network edge servers may schedule and process complex computing tasks offloaded by the vehicle. Fully offloading vehicle computing tasks to a server for execution may excessively use wireless transmission bandwidth and system resources. Also, local processing of most tasks on the vehicle could lead to system delays due to limited computing power, memory resources, storage size, bandwidth, or the like.[0005] In one embodiment, example systems and methods relate to a manner of improving the scheduling of computing tasks of a vehicle to execute on a server. A scheduling system efficiently scheduling an offloaded computing task may be suboptimal at fairly reducing delays. In one approach, a scheduling system may allocate the server resources for offloaded computing tasks more efficiently by analyzing context information of a vehicle and machine learning. Therefore, an improved scheduling system is disclosed that schedules computing tasks in a manner that reduces end-to-end delay and improves fairness by analyzing the context information associated with the vehicle. In one approach, a scheduling system may fairly allocate resources of a server using machine learning. Moreover, the scheduling system may schedule a computing task such that context information and a task descriptor in an offloading request satisfy fair resource loading between the vehicle and the server. In one approach, the computing task may be partitioned into subtasks by a machine learning module according to the context information. A scheduling message is sent to the vehicle with scheduling and task partition [example of task configuration modification] information to offload subtasks to the server for execution. Thus, the scheduling system uses context information and machine learning to improve resource usage for the server for offloaded computing tasks to reduce delays and fairly allocate resources. [0048] For making an offloading scheduling decision, the vehicle scheduling module or the task partitioning module may utilize historical information, system status information, vehicle(s) status information, or the like stored in database 530. For example, system status information may be received from infrastructure component 540 to determine available server computation resources to execute offloaded tasks. In one approach, the infrastructure component 540 may be an edge server, a part of distributed network edge servers, or the like that is co-located or part of controller 520 (not shown). In other approaches, the infrastructure component 540 may be an edge server, a part of distributed network edge servers, or the like remote or independent of controller 520. In another approach, both controller 520 and infrastructure component 540 may be part of or integrated in a cellular base station, Node-b, eNodeB, or the like)
Claims 14 and 15 are rejected under 35 U.S.C. 103 as being unpatentable over Higuchi (US 2022/0116456 A1) in view of Liu (US 2021/0406065 A1) in further view of Sorrentino (US 2024/0414050 A1) and Chen (US 2024/0127105 A1).
As per claim 14, Higuchi and Liu an Sorrentino do not teach wherein the plurality of remote server systems includes at least one server controller and at least one server communication system in electrical communication with the at least one server controller, and wherein the at least one server controller is programmed to: train a global offloading machine learning model based at least in part on historical global task performance data; and transmit the global offloading machine learning model from the at least one server controller to the vehicle communication system using the at least one server communication system..
However, Chen teaches wherein the plurality of remote server systems includes at least one server controller and at least one server communication system in electrical communication with the at least one server controller, and wherein the at least one server controller is programmed to: train a global offloading machine learning model based at least in part on historical global task performance data; and transmit the global offloading machine learning model from the at least one server controller to the vehicle communication system using the at least one server communication system. (Chen [0019] In embodiments, the server 106 considers heterogeneity of the edge nodes, i.e., different sensors and different computing resources of the edge nodes when computing a global model based on the updated local models. For example, the server 106 considers metadata about vehicles received from the vehicles when determining weights for local models. The metadata includes, but not limited to, quality of data that corresponding vehicle uses for training, the number of sensors that corresponding vehicle has, a computing power of a processor of the corresponding vehicle, and the like. Details about computing a global model based on the updated local models will be described with reference to FIGS. 2-4 below. [0047] In FIG. 4, the system includes three vehicles 410, 420, 430 and an edge server 440. The system may include more than or less than three vehicles. In step S402, the edge server 440 initializes a global model. In step S404, the edge server 440 transmits the initialized global model to each of the vehicles 410, 420, 430. In step S406, each of the vehicles 410, 420, 430 trains the initialized global model using its local data such as images captured by the vehicles 410, 420, 430, respectively. In step S408, each of the vehicles 410, 420, 430 transmits the trained model and metadata for the vehicles 410, 420, 430 to the edge server 440. The metadata includes, but not limited to, quality of data that corresponding vehicle uses for training, the number of sensors that corresponding vehicle has, a computing power of a processor of the corresponding vehicle, and the like. The metadata may include the type of a vehicle (e.g., sedan, SUV, truck, etc.), the model of the vehicle, the year of the vehicle, and the like. In some embodiments, the metadata may include the location of the vehicle when images as training data were captured, the time when the images were captured, information on whether when the images were captured, and the like [0048] In step S410, the edge server 440 analyzes the contributions of the trained local models of the vehicles 410, 420, 430 based on the metadata and fuses the trained local models based the contributions to obtain an aggregated global model. In step S412, the edge server 440 transmits the aggregated global model to the vehicles 410, 420, 430. Then, each of the vehicles 410, 420, 430 again trains the aggregated global model using its local data. In addition, each of the vehicles 410, 420, 430 may perform vision-based lane centering using the aggregated global model or autonomous driving).
It would have been obvious to a person in the ordinary skill in the art before the filing date of the claimed invention to combine Chen with the system of Higuchi and Liu and Sorrentino to train a global offloading machine learning model. One having ordinary skill in the art would have been motivated to use Chen into the system of Higuchi and Liu and Sorrentino for the purpose of using a vehicular network that takes into account heterogeneous edge nodes that differ in computation resource and hardware elements of the edge nodes (Chen paragraph 03)
As per claim 15, Chen teaches The system of claim 14, wherein the vehicle controller is further programmed to: receive the global offloading machine learning model using the vehicle communication system; and train the vehicle specific offloading machine learning model based at least in part on the global offloading machine learning model using the vehicle controller, wherein the vehicle specific offloading machine learning model is trained using remote learning. (Chen [0047] In FIG. 4, the system includes three vehicles 410, 420, 430 and an edge server 440. The system may include more than or less than three vehicles. In step S402, the edge server 440 initializes a global model. In step S404, the edge server 440 transmits the initialized global model to each of the vehicles 410, 420, 430. In step S406, each of the vehicles 410, 420, 430 trains the initialized global model using its local data such as images captured by the vehicles 410, 420, 430, respectively. In step S408, each of the vehicles 410, 420, 430 transmits the trained model and metadata for the vehicles 410, 420, 430 to the edge server 440. The metadata includes, but not limited to, quality of data that corresponding vehicle uses for training, the number of sensors that corresponding vehicle has, a computing power of a processor of the corresponding vehicle, and the like. The metadata may include the type of a vehicle (e.g., sedan, SUV, truck, etc.), the model of the vehicle, the year of the vehicle, and the like. In some embodiments, the metadata may include the location of the vehicle when images as training data were captured, the time when the images were captured, information on whether when the images were captured, and the like [0048] In step S410, the edge server 440 analyzes the contributions of the trained local models of the vehicles 410, 420, 430 based on the metadata and fuses the trained local models based the contributions to obtain an aggregated global model. In step S412, the edge server 440 transmits the aggregated global model to the vehicles 410, 420, 430. Then, each of the vehicles 410, 420, 430 again trains the aggregated global model using its local data. In addition, each of the vehicles 410, 420, 430 may perform vision-based lane centering using the aggregated global model or autonomous driving)
Claim 16 is rejected under 35 U.S.C. 103 as being unpatentable over Higuchi (US 2022/0116456 A1) in view of Bernat (US 2020/0389410 A1)
As per claim 16, Higuchi does not teach determine the criticality level of the computing task, the vehicle controller is further programmed to: determine the criticality level of the computing task based at least in part on the predicted performance of the computing task with the optimal task configuration, wherein the criticality level includes one of: a low criticality level, a normal criticality level, a high criticality level, and a very high criticality level.
However, Bernat teaches determine the criticality level of the computing task based at least in part on the predicted performance of the computing task with the optimal task configuration, wherein the criticality level includes one of: a low criticality level, a normal criticality level, a high criticality level, and a very high criticality level. (Bernat [0017] The example “Result Time” service parameter specifies a deadline or a round-trip latency defining a time-constraint by which a result of a function as a service is expected to be received at a requesting client device 105. As such, the “Result Time” service parameter corresponds to a deadline service parameter or a round-trip latency service parameter of the service request 104. In other examples, the “Result Time” service parameter may be omitted from the service request 104, and a deadline service parameter or a round-trip latency service parameter may instead be provided via the “SLA ID” service parameter as described below. The example “Priority ID” service parameter is to identify a priority level or priority policy that specifies a priority that should be assigned to the corresponding service request 104. In some examples, the “priority ID” service parameter is expressed as a “high priority,” a “normal priority,” a “low priority,” and/or a numeric-based priority (e.g., “1” being the highest priority and “10” being the lowest priority))
It would have been obvious to a person in the ordinary skill in the art before the filing date of the claimed invention to combine Bernat with the system of Higuchi to use tasks of varying criticality. One having ordinary skill in the art would have been motivated to use Bernat into the system of Higuchi for the purpose of scheduling service requests in a network computing system using hardware queue managers. (Bernat paragraph 01)
Claim 18 is rejected under 35 U.S.C. 103 as being unpatentable over Higuchi (US 2022/0116456 A1) in view of Chen (US 2024/0127105 A1)
As per claim 18, Higuchi teaches A system for allocating computing resources for a vehicle, the system comprising: a plurality of remote server systems including: (Higuchi [0030] …the computational task may be sent to an edge server, such as edge server 102. Further still, the computational task may be sent to a remote server, such as remote server 104, via network 105)
determine a criticality level of the computing task based at least in part on the optimal task configuration and the task constraint of the computing task, wherein the criticality level includes one of: a low criticality level, a normal criticality level, a high criticality level, and a very high criticality level; (Higuchi [0022] The utility score generally represents an improvement in the functioning of the application if the computational task is offloaded to an external system for processing. For example, if the application is an object detection application, the utility score could indicate a general difference between if the computational task associated with the application is performed by the vehicle processor or is offloaded to be processed by an external system. For example, the vehicle processor may only be able to execute a lightweight version of an object detection algorithm, while the external system could execute a much more computationally intensive [task constraint] but much more accurate object detection algorithm [0024] As such, the system and method for value-anticipating task offloading provides a solution for task offloading such that each computational task is evaluated to determine which computational tasks are the most important [very high criticality level] and would add the most value if they are offloaded. The computational tasks that are identified as having the most value [very critical tasks] if they are offloaded can then be prioritized such that they are offloaded.)
route the computing task to one of the plurality of remote server systems using the vehicle communication system based at least in part on the criticality level of the computing task. (Higuchi Fig 6 Block 408 (offload the computational task to the external system for processing) and [0063] In step 406, the task manager module 220 may then cause the processor(s) 110 to evaluate the utility score and determine if the computational task should be offloaded to an external system for processing (step 408),)
Higuchi does not teach a server communication system; and a server controller in electrical communication with the server communication system, wherein the server controller is programmed to: train a global offloading machine learning model based at least in part on historical global task performance data;
transmit the global offloading machine learning model using the server communication system; and a vehicle system including: a vehicle communication system in wireless communication with the server communication system; a vehicle controller in electrical communication with the vehicle communication system, wherein the vehicle controller is programmed to: receive the global offloading machine learning model using the vehicle communication system; train a vehicle specific offloading machine learning model based at least in part on the global offloading machine learning model using the vehicle controller; determine an optimal task configuration for a computing task using the vehicle specific offloading machine learning model based at least in part on a task constraint of the computing task;
However, Chen teaches a server communication system; and a server controller in electrical communication with the server communication system, wherein the server controller is programmed to: (Chen Fig 2 Bock 106 see also Block 247 (Global model update module))
train a global offloading machine learning model based at least in part on historical global task performance data; (Chen [0019] In embodiments, the server 106 considers heterogeneity of the edge nodes, i.e., different sensors and different computing resources of the edge nodes when computing a global model based on the updated local models. For example, the server 106 considers metadata about vehicles received from the vehicles when determining weights for local models. The metadata includes, but not limited to, quality of data that corresponding vehicle uses for training, the number of sensors that corresponding vehicle has, a computing power of a processor of the corresponding vehicle, and the like. Details about computing a global model based on the updated local models will be described with reference to FIGS. 2-4 below)
transmit the global offloading machine learning model using the server communication system; and a vehicle system including:
a vehicle communication system in wireless communication with the server communication system; (Chen [0017] In some embodiments, the edge nodes 101, 103, 105, 107, and 109 are vehicle nodes, and the vehicles may communicate with a centralized server such as the server 106 via an edge server).
a vehicle controller in electrical communication with the vehicle communication system, wherein the vehicle controller is programmed to: receive the global offloading machine learning model using the vehicle communication system; (Chen [0007] In another embodiment, a system for contribution-aware federated learning is provided. The system includes a server and a plurality of vehicles. Each of the plurality of vehicles includes a controller programmed to: train a local machine learning model using first local data; obtain metadata for hardware elements of corresponding vehicle; transmit the trained local machine learning model and the metadata to a server; receive an aggregated machine learning model from the server [global model]; and train the aggregated machine learning model using second local data. The server generates the aggregated machine learning model based on the trained local machine learning models and the metadata received from the plurality of vehicles [0047] In FIG. 4, the system includes three vehicles 410, 420, 430 and an edge server 440. The system may include more than or less than three vehicles. In step S402, the edge server 440 initializes a global model. In step S404, the edge server 440 transmits the initialized global model to each of the vehicles 410, 420, 430. In step S406, each of the vehicles 410, 420, 430 trains the initialized global model using its local data such as images captured by the vehicles 410, 420, 430, respectively. In step S408, each of the vehicles 410, 420, 430 transmits the trained model and metadata for the vehicles 410, 420, 430 to the edge server 440)
train a vehicle specific offloading machine learning model based at least in part on the global offloading machine learning model using the vehicle controller; determine an optimal task configuration for a computing task using the vehicle specific offloading machine learning model based at least in part on a task constraint of the computing task; (Chen [0048] In step S410, the edge server 440 analyzes the contributions of the trained local models of the vehicles 410, 420, 430 based on the metadata and fuses the trained local models based the contributions to obtain an aggregated global model. In step S412, the edge server 440 transmits the aggregated global model to the vehicles 410, 420, 430. Then, each of the vehicles 410, 420, 430 again trains the aggregated global model using its local data. In addition, each of the vehicles 410, 420, 430 may perform vision-based lane centering using the aggregated global model or autonomous driving)
It would have been obvious to a person in the ordinary skill in the art before the filing date of the claimed invention to combine Chen with the system of Higuchi to train a global and a local machine learning model. One having ordinary skill in the art would have been motivated to use Chen into the system of Higuchi for the purpose of using a vehicular network that takes into account heterogeneous edge nodes that differ in computation resource and hardware elements of the edge nodes (Chen paragraph 03)
Allowable Subject Matter
Claims 6,9,17,19 and 20 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.
Conclusion
The prior art made of record and not relied upon is considered pertinent to applicant's disclosure.
US 20230269766 A1 – discloses training a global machine learning model using a learning server and a set of vehicle agents connected to roadside units (RSUs). The method includes steps of selecting vehicle agents from a pool of the vehicle agents connected to the RSUs, associating the selected vehicle agents and the RSUs respectively based on distances from the selected vehicle agents to the RSUs configured to provide measurements of the distances to the learning server, and transmitting a global model, a selected agent set and deadline thresholds in each global training round to the RSUs configured to transmit the global model and training deadlines to the selected vehicle agents. The associated RSUs compute the training deadlines of the corresponding selected vehicle agents and the selected vehicle agents locally train the global model independently using the local datasets collected by the on-board sensors of the selected vehicle agents to generate locally trained models. The method further includes aggregating the locally trained models from the selected vehicle agents via the associated RSUs to update the global model until the global model reaches an expected level of precision.
US 20230040264 A1 – discloses selecting an optimal edge computing node in an Internet of vehicle (IoV) environment. The method includes: acquiring and analyzing properties of computing tasks of a vehicle in the IoV environment; acquiring and analyzing properties of different edge computing nodes; computing matching degrees between the properties of the computing tasks and the properties of the nodes; analyzing computing demands of different tasks, and assigning weights to different types of matching degrees; and selecting a node having an optimal sum for products of the matching degrees and the weights as an optimal edge computing node to compute each of the computing tasks of the vehicle.
US 20220122011 A1 – discloses receiving (101) a first set of vehicle data and a second set of vehicle data, the vehicle data comprising information about each vehicle of the plurality of vehicles, each vehicle operating along at least one fixed route, receiving (102) a first set of environmental data and a second set of environmental data, the environmental data comprising information about each fixed route, and estimating (103), by means of the global self-learning model and each local-self learning model, a schedule parameter for each vehicle of the plurality of vehicles based on the received first set of vehicle data, the received first set of environmental data, the received second set of vehicle data, the received second set of environmental data, and a predefined interaction model between the global self-learning model and each local-self learning model. The method further comprises receiving (104) a measured schedule parameter for each vehicle, comparing (105) the estimated schedule parameter with the received measured schedule parameter, and updating (106) the global self-learning model and each local self-learning model based on the comparison of the estimated schedule parameter with the received measured schedule parameter.
US 20220083391 A1 – discloses improving execution of processing requests by an edge server. In one embodiment, a method includes predicting a number of computing requests from vehicles for execution by the edge server using a prediction solver for a time period that is forthcoming. The prediction solver may predict the number of computing requests using a prediction model selected in association with service constraints of the edge server and information from an additional server. The method also includes determining a request handling scheme using an optimization solver according to the number of computing requests, the service constraints of the edge server, and a service area of the edge server. The method also includes communicating the request handling scheme and a resource schedule to the edge server on a condition that a resources criteria are satisfied for the time period.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to MEHRAN KAMRAN whose telephone number is (571)272-3401. The examiner can normally be reached on 9-5.
If attempts to reach the examiner by telephone are unsuccessful, the examiner’s supervisor, April Blair can be reached on (571)270-1014. 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.
/MEHRAN KAMRAN/ Primary Examiner, Art Unit 2196