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 . This action is made final.
Claims 1-20 filed on 03/03/2026 have been reviewed and considered by this office action.
Claims 1, 9, and 16 have been amended.
Claim 20 has been newly added.
Drawings
The drawings are objected to as failing to comply with 37 CFR 1.84(p)(5) because they include the following reference characters not mentioned in the description: 1034. Examiner believes “At block 1032” in [0104] was meant to be corrected to “At block 1034.”
Corrected drawing sheets in compliance with 37 CFR 1.121(d), or amendment to the specification to add the reference character(s) in the description in compliance with 37 CFR 1.121(b) are required in reply to the Office action to avoid abandonment of the application. Any amended replacement drawing sheet should include all of the figures appearing on the immediate prior version of the sheet, even if only one figure is being amended. Each drawing sheet submitted after the filing date of an application must be labeled in the top margin as either “Replacement Sheet” or “New Sheet” pursuant to 37 CFR 1.121(d). If the changes are not accepted by the examiner, the applicant will be notified and informed of any required corrective action in the next Office action. The objection to the drawings will not be held in abeyance.
Claim Objections
Claim 9 is objected to because of the following informalities:
In claims 3 and 10, “to the to the” should read “to the”
In claim 9, “performing a sensor task” should read “perform a sensor task” and “performing the ML task” should read “perform the ML task”
Claims 6, 13, and 14 recite the limitation “the industrial operation.” There is insufficient antecedent basis for this limitation in the claim. The limitation will be interpreted to recite “the industrial system.”
Claims 16 was amended to “communicating with an APL switch of the industrial system via standard Ethernet; communicating with at least one APL-based edge device of an industrial system using APL Ethernet via the APL switch.” It would be clearer to write “communicating with an APL switch of an industrial system via standard Ethernet; communicating with at least one APL-based edge device of the industrial system using APL Ethernet via the APL switch.”
In claim 16, “wherein at least one of the first inferences or the second inferences are based” should read “wherein at least one of the first inferences or the second inferences is based”
Claims 16 recites the limitation “for processing by the ML model.” There is insufficient antecedent basis for this limitation in the claim. The limitation will be interpreted to recite “for processing by the second ML model.”
In claim 20, “The method of claim 11” is wrong because claim 11 is an APL-based edge device. If intended to depend from method claim 1 (or method claim 16), it should read “The method of claim 1” (or 16). If intended to depend from device claim 9, it should read “The APL-based edge device of claim 11.”
Appropriate correction is required.
Response to Arguments
Applicant’s arguments, filed 03/03/2026, regarding the rejections under 35 U.S.C. § 103 have been fully considered and are persuasive. Therefore, the rejection has been withdrawn. However, upon further consideration, a new ground of rejection is made in view of STMicroelectronics.
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.
Claims 1-20 are rejected under 35 U.S.C. 103 as being unpatentable over Sharma et al. (US 2020/0327371 A1), in view of Garcia (US 2022/0263717 A1), and in view of STMicroelectronics (STMicroelectronics. “ASM330LHHX Datasheet.” Internet Archive, 01 Jul. 2022).
Regarding claim 1, Sharma teaches a method comprising:
receiving input data from a sensor ([0013]: “The edge computing platform receives a first sensor data stream from a first sensor of the plurality of sensors”) in an event-driven framework that is compliant with power constraints of an [0074]: “The applications can be triggered by or receive input, or both, from the complex event processing (CEP) engine 429, also referred to as the analytics engine, which is preferably adapted to run on low footprint machines”);
([0149]: “Machine learning has evolved as a key computation construct in automating discovery of patterns in data and using the models built to produce intelligent predictions or inferences in a variety of industrial verticals”; [0085]: “an application executing on an example intelligent edge platform according to the invention may monitor and analyze locally and in real-time sensor data from pumps in an industrial IIoT environment. In one example, based on the real-time analysis of the data, which may include the use of machine learning models, an application may output in real-time a predictive maintenance schedule for the pumps, or may automatically take action in the local network to redirect flow around a pump to prevent costly damage due to a cavitation or other event detected or predicted”); and
wherein the input data is received by the ([0085]: “an application executing on an example intelligent edge platform according to the invention may monitor and analyze locally and in real-time sensor data from pumps in an industrial IIoT environment. In one example, based on the real-time analysis of the data, which may include the use of machine learning models, an application may output in real-time a predictive maintenance schedule for the pumps, or may automatically take action in the local network to redirect flow around a pump to prevent costly damage due to a cavitation or other event detected or predicted”).
While Sharma teaches edge devices ([0012]: “Before being deployed, the model is edge-converted (‘edge-ified’) to run optimally with the constrained resources of the edge device and with the same or better level of accuracy”), Sharma does not explicitly teach APL-based edge devices.
Garcia teaches APL-based devices ([0019: “the display device is an APL display device and the specified physical layer is an APL transmission layer. Thus, an APL-based Ethernet system is formed”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to adapt the method of Sharma to incorporate the teachings of Garcia so as to include APL-based devices. Doing so would allow an APL-based edge device to communicate over large distances, with the aim of controlling industrial systems where devices are widely distributed (Garcia, [0019]: “APL simultaneously allows electrical power supply as well as communication via the two-wire of the two-wire Ethernet devices and is suitable for Ex applications (e.g. intrinsic safety according to IEC 60079-11). The possible cable lengths of an APL network are adapted to the requirements of process automation and are thus significantly longer than conventional Ethernet networks with a different physical layer”).
While Sharma teaches performing the ML task to analyze the input data and making an inference on the input data about the one or more aspects of the industrial system ([0013]: “the machine learning model operates on the first sensor data stream and produces a stream of first inferences about a first network device in real-time”), Sharma does not explicitly teach that the input data is buffered. Sharma and Garcia do not explicitly teach “performing a sensor task that buffers the input data according to timer, interrupt, and data availability events.”
STMicroelectronics teaches performing a sensor task that buffers the input data (Page 33, Section 6.5: “The ASM330LHHX embeds 3 KB of data in FIFO to store the following data: • Gyroscope • Accelerometer • External sensors (up to 4) • Timestamp • Temperature”) according to timer (Page 33, Section 6.5: “If a change in the ODR or BDR (batch data rate) configuration is performed, the application can correctly reconstruct the timestamp and know exactly when the change was applied without disabling FIFO batching”; Page 35, Section 6.5.7: “In addition, it is possible to configure a counter of the batch events of accelerometer or gyroscope sensors. The flag COUNTER_BDR_IA in FIFO_STATUS2 (3Bh) alerts that the counter reaches a selectable threshold (CNT_BDR_TH_[10:0] field in COUNTER_BDR_REG1 (0Bh) and COUNTER_BDR_REG2 (0Ch)). This allows triggering the reading of FIFO with the desired latency of one single sensor”), interrupt, and data availability events (Page 33, Section 6.5: “Writing external sensor data in FIFO can be triggered by the accelerometer data-ready signal or by an external sensor interrupt”) and
performing the ML task to analyze the input data and make an inference on the buffered input data about the one or more aspects of the industrial system (Page 6, Section 2.2: “The ASM330LHHX embeds a dedicated core for machine learning processing that provides system flexibility, allowing some algorithms run in the application processor to be moved to the MEMS sensor with the advantage of consistent reduction in power consumption. Machine learning core logic allows identifying if a data pattern (for example motion, pressure, temperature, magnetic data, and so forth) matches a user-defined set of classes. Typical examples of applications could be activity detection like running, walking, driving, and so forth. The ASM330LHHX machine learning core works on data patterns coming from the accelerometer and gyroscope sensors, but it is also possible to connect and process external sensor data (like magnetometer) by using the sensor hub feature (mode 2)”).
It would have been obvious to one of ordinary skill in the art before the effective filing date of the claimed invention to adapt the method of Sharma in view of Garcia to incorporate the teachings of STMicroelectronics so as to include performing a sensor task that buffers the input data according to timer, interrupt, and data availability events and performing the ML task to analyze the input data and make an inference on the buffered input data about the one or more aspects of the industrial system. Doing so would allow sensor data buffering and machine learning processes to be offloaded with the aim of reducing power consumption (Page 3, Section 1: “The device also includes digital features like a finite state machine and an ST proprietary machine learning core, allowing defined motion pattern detection or some complex algorithms run in the application processor to be moved to the MEMS sensor with the advantage of consistent reduction in power consumption”).
Regarding claim 2, Sharma in view of Garcia and STMicroelectronics teaches the method of claim 1.
Sharma further teaches wherein the method further comprises: training the ML model using an external computing device; and deploying the ML model on the APL-based edge device ([0159]: “Model creation and training may still be accomplished in the cloud, where significant compute and storage resources are available. Once a model is trained, it can then be 'edge-ified' as described herein and pushed to the edge for live execution”).
Regarding claim 3, Sharma in view of Garcia and STMicroelectronics teaches the method of claim 1.
While Sharma teaches triggering actions in an industrial system, ([0085]: “an application executing on an example intelligent edge platform according to the invention may monitor and analyze locally and in real-time sensor data from pumps in an industrial IIoT environment. In one example, based on the real-time analysis of the data, which may include the use of machine learning models, an application may output in real-time a predictive maintenance schedule for the pumps, or may automatically take action in the local network to redirect flow around a pump to prevent costly damage due to a cavitation or other event detected or predicted”) Sharma does not explicitly teach “wherein the action is caused by controlling one or more actuators communicatively coupled to the to the APL-based edge device.”
Garcia further teaches wherein the action is caused by controlling one or more actuators communicatively coupled to the to the APL-based edge device ([0009]: “The two-wire Ethernet network has a specified protocol and one or more different higher layer network protocols above the physical layer protocol specified for these network protocols, e.g. an APL (Advanced Physical Layer)”; [0017]: “Subscriber devices here are, for example, sensors, actuators such as valves, pumps, motors, control devices, etc., or other display devices. The devices may be two-wire Ethernet devices, which can also be specified according to Ex (explosive) requirements and used in the network”).
Regarding claim 4, Sharma in view of Garcia and STMicroelectronics teaches the method of claim 1.
Sharma further teaches further comprising outputting the inference by transmitting the inference to a second APL edge device that causes the action ([0012]: “Multiple edge-based models communicate and are chained using a topic-based publish-subscribe infrastructure”).
Regarding claim 5, Sharma in view of Garcia and STMicroelectronics teaches the method of claim 1.
Sharma further teaches further comprising outputting the inference by transmitting the inference to a cloud-based processor that causes a second APL edge device to cause the action ([0159]: “inferences produced by models at the edge can be frequently sent to the cloud to further tune the models, and the updated models pushed back to the edge in a highly iterative, closed-loop fashion”).
Regarding claim 6, Sharma in view of Garcia and STMicroelectronics teaches the method of claim 1.
Sharma further teaches wherein the input data includes sensor data output by a sensor ([0013]: “The edge computing platform receives a first sensor data stream from a first sensor of the plurality of sensors”),
wherein analyzing the input data and the making the inference about the one or more aspects of the industrial operation includes detecting and characterizing normal operating conditions associated with the one or more aspects of the industrial system based on the sensor input data ([0155]: “A concrete example of this can be illustrated in the context of running predictive maintenance analytics for elevators at the edge of an IoT network”; [0157]: “the prediction (ML inference) required at the edge is typically at a much more granular level. As per the above example, it is desired to predict and examine elevator states just as they are happening based on all of the data values generated by the sensors, instead of examining the past”), and
recognizing abnormal conditions associated with the one or more aspects of the industrial system based on the sensor input data ([0156]: “an 'edge-ified' ML model as described herein is able to operate on the edge with constrained resources and in real-time on streaming sensor data indicating present elevator conditions, is able to predict essentially in real-time an imminent failure, and thus is able to immediately trigger a warning or remedial action”).
Regarding claim 7, Sharma in view of Garcia and STMicroelectronics teaches the method of claim 1.
Sharma further teaches wherein the input data includes machine-learning result data provided by another APL device or a cloud device that analyzed sensor data using a trained ML model ([0159]: “inferences produced by models at the edge can be frequently sent to the cloud to further tune the models, and the updated models pushed back to the edge in a highly iterative, closed-loop fashion”),
wherein analyzing the input data and the making the inference about the one or more aspects of the industrial system includes detecting and characterizing normal operating conditions associated with the one or more aspects of the industrial system based on the sensor input data ([0155]: “A concrete example of this can be illustrated in the context of running predictive maintenance analytics for elevators at the edge of an IoT network”; [0157]: “the prediction (ML inference) required at the edge is typically at a much more granular level. As per the above example, it is desired to predict and examine elevator states just as they are happening based on all of the data values generated by the sensors, instead of examining the past”), and
recognizing abnormal conditions associated with the one or more aspects of the industrial system based on the machine-learning result data ([0156]: “an 'edge-ified' ML model as described herein is able to operate on the edge with constrained resources and in real-time on streaming sensor data indicating present elevator conditions, is able to predict essentially in real-time an imminent failure, and thus is able to immediately trigger a warning or remedial action”).
Regarding claim 8, Sharma in view of Garcia and STMicroelectronics teaches the method of claim 1.
Sharma further teaches further comprising: generating, using the inference, control data configured to control a device; and outputting the control data ([0230]: “The predictions, inferences, and other analytics results and intelligence can be transferred from the edge platform to the cloud 412 via the data publisher 570, as well as to other edge platform instances, to the local network, or to other networks. They can also be used to determine whether to take actions in the local network with respect to control systems, machines, sensors and devices 523, or the like, or to provide information, or a combination, alarms, warnings, and others, for example to a management system user interface 908 of the local network”).
Regarding claim 9, Sharma teaches an Advanced Physical Layer (APL)-based edge device comprising:
an embedded processing device and memory, the memory configured to store a plurality of programmable instructions, and the processing device in communication with the memory ([0055]: “A computer-implemented or computer-executable version or computer program product incorporating the invention or aspects thereof may be embodied using, stored on, or associated with computer-readable medium. A computer-readable medium may include any medium that participates in providing instructions to one or more processors for execution. Such a medium may take many forms including, but not limited to, nonvolatile, volatile, and transmission media. Non-volatile media may include, for example, flash memory, or optical or magnetic disks. Volatile media includes static or dynamic memory, such as cache memory or RAM”),
wherein the processing device, upon execution of the plurality of programmable instructions is configured to: receive input data from a sensor ([0013]: “The edge computing platform receives a first sensor data stream from a first sensor of the plurality of sensors”) in an event-driven framework that is compliant with power constraints of the ([0074]: “The applications can be triggered by or receive input, or both, from the complex event processing (CEP) engine 429, also referred to as the analytics engine, which is preferably adapted to run on low footprint machines”);
having an ML model that is trained to analyze the input data and make inferences about one or more aspects of an industrial system based on the input data ([0149]: “Machine learning has evolved as a key computation construct in automating discovery of patterns in data and using the models built to produce intelligent predictions or inferences in a variety of industrial verticals”; [0085]: “an application executing on an example intelligent edge platform according to the invention may monitor and analyze locally and in real-time sensor data from pumps in an industrial IIoT environment. In one example, based on the real-time analysis of the data, which may include the use of machine learning models, an application may output in real-time a predictive maintenance schedule for the pumps, or may automatically take action in the local network to redirect flow around a pump to prevent costly damage due to a cavitation or other event detected or predicted”); and
wherein the input data is received by the ([0085]: “an application executing on an example intelligent edge platform according to the invention may monitor and analyze locally and in real-time sensor data from pumps in an industrial IIoT environment. In one example, based on the real-time analysis of the data, which may include the use of machine learning models, an application may output in real-time a predictive maintenance schedule for the pumps, or may automatically take action in the local network to redirect flow around a pump to prevent costly damage due to a cavitation or other event detected or predicted”).
While Sharma teaches edge devices ([0012]: “Before being deployed, the model is edge-converted (‘edge-ified’) to run optimally with the constrained resources of the edge device and with the same or better level of accuracy”), Sharma does not explicitly teach APL-based edge devices.
Garcia teaches APL-based devices ([0019: “the display device is an APL display device and the specified physical layer is an APL transmission layer. Thus, an APL-based Ethernet system is formed”).
The reasons to combine Garcia into Sharma are the same as articulated in the rejection of claim 1 above.
While Sharma teaches performing the ML task to analyze the input data and making an inference on the input data about the one or more aspects of the industrial system ([0013]: “the machine learning model operates on the first sensor data stream and produces a stream of first inferences about a first network device in real-time”), Sharma does not explicitly teach that the input data is buffered. Sharma and Garcia do not explicitly teach “performing a sensor task that buffers the input data according to timer, interrupt, and data availability events.”
STMicroelectronics teaches performing a sensor task that buffers the input data (Page 33, Section 6.5: “The ASM330LHHX embeds 3 KB of data in FIFO to store the following data: • Gyroscope • Accelerometer • External sensors (up to 4) • Timestamp • Temperature”) according to timer (Page 33, Section 6.5: “If a change in the ODR or BDR (batch data rate) configuration is performed, the application can correctly reconstruct the timestamp and know exactly when the change was applied without disabling FIFO batching”; Page 35, Section 6.5.7: “In addition, it is possible to configure a counter of the batch events of accelerometer or gyroscope sensors. The flag COUNTER_BDR_IA in FIFO_STATUS2 (3Bh) alerts that the counter reaches a selectable threshold (CNT_BDR_TH_[10:0] field in COUNTER_BDR_REG1 (0Bh) and COUNTER_BDR_REG2 (0Ch)). This allows triggering the reading of FIFO with the desired latency of one single sensor”), interrupt, and data availability events (Page 33, Section 6.5: “Writing external sensor data in FIFO can be triggered by the accelerometer data-ready signal or by an external sensor interrupt”) and
performing the ML task to analyze the input data and make an inference on the buffered input data about the one or more aspects of the industrial system (Page 6, Section 2.2: “The ASM330LHHX embeds a dedicated core for machine learning processing that provides system flexibility, allowing some algorithms run in the application processor to be moved to the MEMS sensor with the advantage of consistent reduction in power consumption. Machine learning core logic allows identifying if a data pattern (for example motion, pressure, temperature, magnetic data, and so forth) matches a user-defined set of classes. Typical examples of applications could be activity detection like running, walking, driving, and so forth. The ASM330LHHX machine learning core works on data patterns coming from the accelerometer and gyroscope sensors, but it is also possible to connect and process external sensor data (like magnetometer) by using the sensor hub feature (mode 2)”).
The reasons to combine STMicroelectronics into Sharma in view of Garcia are the same as articulated in the rejection of claim 1 above.
Regarding claim 10, Sharma in view of Garcia and STMicroelectronics teaches the APL-based edge device of claim 9.
While Sharma teaches triggering actions in an industrial system, ([0085]: “an application executing on an example intelligent edge platform according to the invention may monitor and analyze locally and in real-time sensor data from pumps in an industrial IIoT environment. In one example, based on the real-time analysis of the data, which may include the use of machine learning models, an application may output in real-time a predictive maintenance schedule for the pumps, or may automatically take action in the local network to redirect flow around a pump to prevent costly damage due to a cavitation or other event detected or predicted”) Sharma does not explicitly teach “wherein the action is caused by controlling one or more actuators communicatively coupled to the to the APL-based edge device.”
Garcia further teaches wherein the action is caused by controlling one or more actuators communicatively coupled to the to the APL-based edge device ([0009]: “The two-wire Ethernet network has a specified protocol and one or more different higher layer network protocols above the physical layer protocol specified for these network protocols, e.g. an APL (Advanced Physical Layer)”; [0017]: “Subscriber devices here are, for example, sensors, actuators such as valves, pumps, motors, control devices, etc., or other display devices. The devices may be two-wire Ethernet devices, which can also be specified according to Ex (explosive) requirements and used in the network”).
Regarding claim 11, Sharma in view of Garcia and STMicroelectronics teaches the APL-based edge device of claim 9.
Sharma further teaches wherein the processing device, upon execution of the plurality of programmable instructions, is further configured to: output the inference by transmitting the inference to a second APL edge device that causes the action ([0012]: “Multiple edge-based models communicate and are chained using a topic-based publish-subscribe infrastructure”).
Regarding claim 12, Sharma in view of Garcia and STMicroelectronics teaches the APL-based edge device of claim 9.
Sharma further teaches wherein the processing device, upon execution of the plurality of programmable instructions, is further configured to output the inference by transmitting the inference to a cloud-based processor that causes a second APL edge device to cause the action ([0159]: “inferences produced by models at the edge can be frequently sent to the cloud to further tune the models, and the updated models pushed back to the edge in a highly iterative, closed-loop fashion”).
Regarding claim 13, Sharma in view of Garcia and STMicroelectronics teaches the APL-based edge device of claim 9.
Sharma further teaches wherein the input data includes sensor data output by a sensor ([0013]: “The edge computing platform receives a first sensor data stream from a first sensor of the plurality of sensors”),
wherein analyzing the input data and the making the inference about the one or more aspects of the industrial operation includes detecting and characterizing normal operating conditions associated with the one or more aspects of the industrial system based on the sensor input data ([0155]: “A concrete example of this can be illustrated in the context of running predictive maintenance analytics for elevators at the edge of an IoT network”; [0157]: “the prediction (ML inference) required at the edge is typically at a much more granular level. As per the above example, it is desired to predict and examine elevator states just as they are happening based on all of the data values generated by the sensors, instead of examining the past”), and
recognizing abnormal conditions associated with the one or more aspects of the industrial system based on the sensor input data ([0156]: “an 'edge-ified' ML model as described herein is able to operate on the edge with constrained resources and in real-time on streaming sensor data indicating present elevator conditions, is able to predict essentially in real-time an imminent failure, and thus is able to immediately trigger a warning or remedial action”).
Regarding claim 14, Sharma in view of Garcia and STMicroelectronics teaches the APL-based edge device of claim 9.
Sharma further teaches wherein the input data includes machine-learning result data provided by another APL device or a cloud device that analyzed sensor data using a trained ML model ([0159]: “inferences produced by models at the edge can be frequently sent to the cloud to further tune the models, and the updated models pushed back to the edge in a highly iterative, closed-loop fashion”),
wherein analyzing the input data and the making the inference about the one or more aspects of the industrial operation includes detecting and characterizing normal operating conditions associated with the one or more aspects of the industrial system based on the sensor input data ([0155]: “A concrete example of this can be illustrated in the context of running predictive maintenance analytics for elevators at the edge of an IoT network”; [0157]: “the prediction (ML inference) required at the edge is typically at a much more granular level. As per the above example, it is desired to predict and examine elevator states just as they are happening based on all of the data values generated by the sensors, instead of examining the past”), and
recognizing abnormal conditions associated with the one or more aspects of the industrial system based on the machine-learning result data ([0156]: “an 'edge-ified' ML model as described herein is able to operate on the edge with constrained resources and in real-time on streaming sensor data indicating present elevator conditions, is able to predict essentially in real-time an imminent failure, and thus is able to immediately trigger a warning or remedial action”).
Regarding claim 15, Sharma in view of Garcia and STMicroelectronics teaches the APL-based edge device of claim 9.
Sharma further teaches wherein the processing device, upon execution of the plurality of programmable instructions, is further configured to: generate, using the inference, control data configured to control a device; and output the control data ([0230]: “The predictions, inferences, and other analytics results and intelligence can be transferred from the edge platform to the cloud 412 via the data publisher 570, as well as to other edge platform instances, to the local network, or to other networks. They can also be used to determine whether to take actions in the local network with respect to control systems, machines, sensors and devices 523, or the like, or to provide information, or a combination, alarms, warnings, and others, for example to a management system user interface 908 of the local network”).
Regarding claim 16, Sharma teaches a method performed by a cloud-based computing device of an industrial system, the method comprising:
receiving first inferences about one or more aspects of the industrial system from a first ML model deployed on the at least one ([0020]: “one or more of the first inferences produced by the edge-converted model and one or more data points from the first sensor data stream may be transmitted to the remote cloud network for evaluation”) and/or transmitting data to a second ML model deployed on the at least one ([0028]: “A second edge-based machine learning model may access the stream of first inferences by subscribing to the second topic name, operate on the stream of first inferences, and produce a stream of second inferences about one or more of the plurality of network devices. A determination may be made from the stream of second inferences whether to take an action affecting one or more of the network devices”).
While Sharma teaches edge devices ([0012]: “Before being deployed, the model is edge-converted (‘edge-ified’) to run optimally with the constrained resources of the edge device and with the same or better level of accuracy”), Sharma does not explicitly teach APL-based edge devices.
Garcia teaches APL-based devices ([0019: “the display device is an APL display device and the specified physical layer is an APL transmission layer. Thus, an APL-based Ethernet system is formed”),
communicating with an information technology (IT)-based network via standard Ethernet ([0021]: “The Ethernet, which is used in the control room and forms the so-called backbone, is connected to the controller or PLC or control system”);
communicating with an APL switch of the industrial system via standard Ethernet ([0019]: “an APL-based Ethernet system is formed”); and
communicating with at least one APL-based edge device of an industrial system using APL Ethernet via the APL switch ([0021]: “different APL field switch modules arranged up to Ex zone 1, to which 1 to n APL field devices can each be connected in point-to-point wiring, depending on the number of channels. These supply themselves from the APL system in terms of energy”).
The reasons to combine Garcia into Sharma are the same as articulated in the rejection of claim 1 above.
While Sharma teaches performing the ML task to analyze the input data and making an inference on the input data about the one or more aspects of the industrial system ([0013]: “the machine learning model operates on the first sensor data stream and produces a stream of first inferences about a first network device in real-time”), Sharma and Garcia do not explicitly teach “wherein at least one of the first inferences or the second inferences are based on input data buffered according to timer, interrupt, and data availability events.”
STMicroelectronics teaches wherein at least one of the first inferences or the second inferences are based on input data buffered according to timer, interrupt, and data availability events (Page 33, Section 6.5: “If a change in the ODR or BDR (batch data rate) configuration is performed, the application can correctly reconstruct the timestamp and know exactly when the change was applied without disabling FIFO batching”; Page 35, Section 6.5.7: “In addition, it is possible to configure a counter of the batch events of accelerometer or gyroscope sensors. The flag COUNTER_BDR_IA in FIFO_STATUS2 (3Bh) alerts that the counter reaches a selectable threshold (CNT_BDR_TH_[10:0] field in COUNTER_BDR_REG1 (0Bh) and COUNTER_BDR_REG2 (0Ch)). This allows triggering the reading of FIFO with the desired latency of one single sensor”; Page 33, Section 6.5: “Writing external sensor data in FIFO can be triggered by the accelerometer data-ready signal or by an external sensor interrupt”; Page 6, Section 2.2: “The ASM330LHHX embeds a dedicated core for machine learning processing that provides system flexibility, allowing some algorithms run in the application processor to be moved to the MEMS sensor with the advantage of consistent reduction in power consumption. Machine learning core logic allows identifying if a data pattern (for example motion, pressure, temperature, magnetic data, and so forth) matches a user-defined set of classes. Typical examples of applications could be activity detection like running, walking, driving, and so forth. The ASM330LHHX machine learning core works on data patterns coming from the accelerometer and gyroscope sensors, but it is also possible to connect and process external sensor data (like magnetometer) by using the sensor hub feature (mode 2)”).
The reasons to combine STMicroelectronics into Sharma in view of Garcia are the same as articulated in the rejection of claim 1 above.
Regarding claim 17, Sharma in view of Garcia and STMicroelectronics teaches the method of claim 16.
Sharma further teaches wherein the first inferences are received ([0020]: “one or more of the first inferences produced by the edge-converted model and one or more data points from the first sensor data stream may be transmitted to the remote cloud network for evaluation”), the method further comprising aggregating the first inferences over time ([0113]: “The data publisher can transfer raw sensor data as well as ingested and enriched pre-processed sensor data and intelligence information to the local time-series database 576 and to remote cloud storage 573. Further, the data publisher can be used to retrieve aggregated data stored in the local time-series database and to transfer the data to the remote cloud storage, for example to facilitate developing machine learning models for deployment to the edge, as well as evaluating and updating machine learning models already deployed on the edge platform”; [0235]: “the transferred predictions, inferences, data and analytics results, and other information can be aggregated in cloud storage 573”) and/or over a plurality of APL edge devices of the at least one APL edge devices ([0012]: “Multiple edge-based models communicate and are chained using a topic-based publish-subscribe infrastructure”).
Regarding claim 18, Sharma in view of Garcia and STMicroelectronics teaches the method of claim 17.
Sharma further teaches further comprising: generating control data based on a result of the aggregation ([0159]: “The converted and optimized models are thus able to execute very rapidly in real-time on the streaming sensor data received at the edge platform and to provide immediate outputs at rates sufficient for edge-based machine learning applications to trigger immediate actions”),
wherein the control data is configured to control a field device of the industrial system ([0069]: “The edge intelligence provides important information, for example information that can be used for managing, maintaining, and optimizing the performance of industrial machines and other industrial components in an IIoT environment”); and
outputting the control data to the at least one ([0159]: “The converted and optimized models are thus able to execute very rapidly in real-time on the streaming sensor data received at the edge platform and to provide immediate outputs at rates sufficient for edge-based machine learning applications to trigger immediate actions”).
Regarding claim 19, Sharma in view of Garcia and STMicroelectronics teaches the method of claim 16.
Sharma further teaches wherein the second inferences are configured to be used for generating control signals to control a field device of the industrial system ([0069]: “The edge intelligence provides important information, for example information that can be used for managing, maintaining, and optimizing the performance of industrial machines and other industrial components in an IIoT environment”), and
the data transmitted to the second ML model is configured to be processed via the second ML model for making the second inferences ([0020]: “one or more of the first inferences produced by the edge-converted model and one or more data points from the first sensor data stream may be transmitted to the remote cloud network for evaluation. At the remote cloud network, the inferences may be evaluated for accuracy using a remote version of the edge-converted machine learning model that is trained and adapted to operate on a stored aggregated set of the first sensor data. The remote version of the model may be applied to one or more data points from the first sensor data stream to produce one or more second inferences”).
Regarding claim 20, Sharma in view of Garcia and STMicroelectronics teaches the method of claim 11.
While Sharma teaches that data streams can be processed in parallel ([0099]: “A processing unit may require inputs from multiple data streams simultaneously and may produce one or many valuable output streams”), Sharma and Garcia do not explicitly teach “wherein the timer, interrupt, and data availability events run in parallel.”
STMicroelectronics further teaches wherein the timer, interrupt, and data availability events run in parallel (Page 5, Section 2.1: “All 16 finite state machines are independent: each one has its dedicated memory area and it is independently executed. An interrupt is generated when the end state is reached or when some specific command is performed”; Page 6, Section 2.2: “The ASM330LHHX can be configured to run up to 8 flows simultaneously and independently and every flow can generate up to 256 results”).
Conclusion
Applicant's amendment necessitated the new ground(s) of rejection presented in this Office action. Accordingly, THIS ACTION IS MADE FINAL. See MPEP § 706.07(a). Applicant is reminded of the extension of time policy as set forth in 37 CFR 1.136(a).
A shortened statutory period for reply to this final action is set to expire THREE MONTHS from the mailing date of this action. In the event a first reply is filed within TWO MONTHS of the mailing date of this final action and the advisory action is not mailed until after the end of the THREE-MONTH shortened statutory period, then the shortened statutory period will expire on the date the advisory action is mailed, and any nonprovisional extension fee (37 CFR 1.17(a)) pursuant to 37 CFR 1.136(a) will be calculated from the mailing date of the advisory action. In no event, however, will the statutory period for reply expire later than SIX MONTHS from the mailing date of this final action.
Any inquiry concerning this communication or earlier communications from the examiner should be directed to Magdalena Kossek whose telephone number is (571)272-5603. The examiner can normally be reached Mon-Fri 8:00-5:00 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, Robert Fennema can be reached on (571)272-2748. 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.
/M.I.K./Examiner, Art Unit 2117
/ROBERT E FENNEMA/Supervisory Patent Examiner, Art Unit 2117
Prosecution Insights