JP6828633B2 - Data processing device, its control method, and program - Google Patents

Description

The present invention relates to a data processing device that processes data collected from sensors connected to a network, a control method thereof, and a program.

Sensor processing, which is indispensable for the IoT (Internet of Things) environment in which a large number of sensors are connected to a network, generally aggregates the data acquired by each sensor into a data processing device such as a cloud and processes the data. It is realized by performing (Non-Patent Document 1). Non-Patent Document 2 describes a configuration in which a gateway (fog node or the like) having a cloud function installed between a cloud and a sensor functions as a data processing device. In this way, in an IoT environment in which a data processing device such as a cloud or a gateway collects data from a large number of sensors, as the number of sensors increases, the power consumption of the entire network increases and the processing load on the data processing device increases. The challenge is to become.

In response to such problems, research is underway to reduce the power consumption of the entire IoT network by stopping the operation of some sensors within a predetermined period. Non-Patent Document 3 proposes a technique for starting and stopping the operation of the sensor in accordance with a predetermined communication timing. In Non-Patent Document 4, the timing at which the sensor starts and stops is variably set for each sensor based on the communication quality of the target sensor itself and the communication quality (predicted value) of other sensors around the target sensor. The technology to do is proposed.

A. Al-Fuqaha et al., "Internet of Things: A Survey on Enabling Technologies, Protocols, and Applications," in IEEE Communication Survey & Tutorials, vol. 17, no. 4, 2015. M. Chiang et al., "Fog and IoT: An Overview of Research Opportunities," in IEEE Internet of Things Journal, vol. 3, no. 6, 2016. P. Guo et al., "Sleep scheduling for critical event monitoring in wireless sensor networks," IEEE Trans. Parallel Distrib. Syst., Vol. 23, no. 2, pp. 345-352, Feb. 2012. D. Ye et al., "A Self-Adaptive Sleep / Wake-Up Scheduling Approach for Wireless Sensor Networks," IEEE Trans. Cybern., 2017.

In the above-mentioned prior art, the operating state (operating mode) of the sensor is controlled based on the timing or quality of communication between the communication device such as the sensor and the data processing device, but from the viewpoint of data processing performance in the IoT network. No consideration has been given. Therefore, with the conventional technology, although it can be expected to reduce the power consumption of the entire IoT network, it is difficult to realize high-performance and highly reliable data processing for the data collected from the sensor.

The present invention has been made in view of the above-mentioned problems. An object of the present invention is to provide a technique in which a data processing device controls an operation mode of each sensor in the network so as to reduce the power consumption of the entire network while maintaining the processing accuracy of the data collected from the sensors. There is.

The data processing device according to one aspect of the present invention is a data processing device capable of communicating with each sensor in the network, collecting sensor data from each sensor and performing data processing, and is an area in which the target sensor is arranged. the determination of the situation, a determination unit that performs the data processing for pairs the sensor data collected from sensors operating in a non-power saving mode, outputs the reliability of the determination result and the determination result for the target sensor A power saving means that does not output sensor data among a plurality of sensors in the network and a generation means that generates the determination device by machine learning using sensor data collected in the past from each sensor in the network. The target sensor is a determination means for determining a sensor to be operated in the mode by the machine learning, and the target sensor satisfies a condition indicating that the target sensor is less important in the determination by the determination device in the machine learning. When the target sensor is satisfied, the operation mode of the sensor is determined as a sensor to be operated in the power saving mode, so that the determination means and the sensor determined by the determination means operate in the power saving mode. It is characterized by comprising a control means for controlling the above.

According to the present invention, the data processing device can control the operation mode of each sensor in the network so as to reduce the power consumption of the entire network while maintaining the processing accuracy of the data collected from the sensors.

The figure which shows the network configuration example which concerns on one Embodiment A block diagram showing a configuration example of a gateway device according to an embodiment A flowchart showing a processing procedure of sleep transition control according to one embodiment. The figure which shows the example of the data processing by the determination device which concerns on one Embodiment A flowchart showing a processing procedure of sleep wakeup control according to one embodiment.

Hereinafter, exemplary embodiments of the present invention will be described with reference to the drawings. In each of the following figures, components not necessary for the description of the embodiment will be omitted from the drawings.

<Network configuration>
FIG. 1A is a diagram showing an example of a network configuration including a data processing device according to the present embodiment. In this embodiment, a cloud (server on the cloud) and a plurality of sensors (S1, S2, S3, ...) Are provided. The cloud is communicably connected to each sensor in the network by wire or wirelessly, can collect data (sensor data) output from each sensor, and processes the collected sensor data. An arbitrary number of sensors can be provided under the cloud.

The cloud of FIG. 1A functions as an example of the data processing device of the present invention that collects sensor data from each sensor and performs data processing. In the present invention, for example, as shown in FIG. 1 (B), a gateway device (G1, G2, G3, ...) Is provided between a plurality of subordinate sensors and a cloud (server on the cloud). It can also be applied to other network configurations.

Each gateway device of FIG. 1B has a gateway function of collecting sensor data from a plurality of sensors under the control, performing predetermined processing on the collected data, and transmitting the processing result to the cloud. The gateway device is communicably connected to each of the sensors under its control by wire or wirelessly, and can acquire sensor data output from each sensor. An arbitrary number of sensors can be provided under the gateway device. It should be noted that inter-gateway communication is possible in which one gateway device directly communicates with another gateway device or indirectly communicates via one or more gateway devices, and the gateway devices communicate with each other. It may be possible to acquire data. In the configuration of FIG. 1B, each gateway device functions as an example of the data processing device of the present invention in which sensor data is collected from each sensor and data processing is performed.

Each sensor shown in FIGS. 1A and 1B has a normal mode (non-power saving mode) and a sleep mode (power saving mode) that consumes less power than the normal mode as an operation mode, and data. It can operate in any of the operating modes according to the instructions from the processing device. The normal mode is an operation mode in which the sensor sensing and communication functions operate and sensor data is output. The sleep mode is an operation mode in which the sensor sensing and communication functions are stopped and the sensor data is not output. In sleep mode, the power consumption of the sensor is reduced by stopping the sensing and transmission functions. However, each sensor operates the communication function (intermittently) at predetermined time intervals so that the control signal transmitted from the data processing device can be received even during the operation in the sleep mode.

<Outline of processing by data processing device>
The data processing device of the present embodiment (cloud of FIG. 1A or gateway device of FIG. 1B) collects sensor data and uses a calculation model (judgment device) generated by prior machine learning. It has a function to process the collected data. Specifically, the data processing device performs the following three stages of processing.

● Preprocessing The data processing device first generates a calculation model (judgment device) used for data processing on the collected sensor data by performing machine learning using the sensor data collected in the past.
● Judgment processing Next, the data processing device uses the judgment device generated by the pre-processing to process data on the sensor data collected from the sensors under its own device (and from the surrounding gateway devices as necessary). I do. If the data processing device is a gateway device, the processing result is transmitted to the cloud as needed.
● Feedback processing Furthermore, the data processing device periodically updates the judgment device by machine learning using sensor data newly collected from each sensor in the network in the judgment processing. As a result, the judgment device used for the judgment processing can cope with the daily fluctuation of the sensor data collected from each sensor.

In the present embodiment, the data processing device uses control signals transmitted to each sensor in the network so as to reduce the power consumption of the entire network while maintaining the processing accuracy of the sensor data collected from each sensor. Controls the operating mode of.

Specifically, the data processing device is among a plurality of sensors in the network based on the importance of each sensor related to the judgment (data processing) by the judgment device generated by machine learning in the preprocessing. , Determine the sensor to operate in sleep mode. Further, the data collection process controls the determined sensor to operate in the sleep mode at the execution stage of the determination process. In the above determination, as will be described later, a sensor that satisfies the condition indicating that the importance related to the determination by the corresponding determination device is low is selected as the sensor to be operated in the sleep mode. As a result, it is possible to prevent a decrease in sensor data processing accuracy (judgment accuracy of the judgment device) due to the fact that the sensor data output from the sensor operating in the sleep mode cannot be used for data processing by the corresponding judgment device. ..

Further, the data processing device controls the sensor operating in the sleep mode to return the sensor from the sleep mode to the normal mode when a predetermined return condition described later is satisfied. This return condition deals with a situation in which it is necessary to improve the judgment accuracy of the sensor or the judgment device corresponding to the sensor existing in the vicinity of the sensor, or prevent the judgment accuracy from being lowered. It is stipulated to do.

At the execution stage of the feedback processing, the data processing device may determine the sensor to be operated in the sleep mode by machine learning each time the determination device is updated. As a result, the processing accuracy of the sensor data can be maintained while adapting to the daily fluctuation of the sensor data collected from each sensor (the fluctuation of the situation in the area where the sensor is arranged).

<Functional configuration of data processing device>
FIG. 2 is a block diagram showing a functional configuration example of the data processing device 20 according to the present embodiment. As shown in FIG. 2, the data processing device 20 is roughly divided into a preprocessing unit 21, a determination processing unit 22, a feedback processing unit 23, and a data storage unit 24. The pre-processing unit 21, the determination processing unit 22, and the feedback processing unit 23 are functional units that realize the above-mentioned pre-processing, determination processing, and feedback processing, respectively. The preprocessing unit 21, the determination processing unit 22, and the feedback processing unit 23 can be realized in the data processing device 20 by hardware, software, or a combination thereof. When realized by software, one or more processors (not shown) such as a CPU included in the data processing device 20 read and execute a control program stored in a storage device (not shown) such as a ROM. Can be realized by.

● Pretreatment unit 21
The pre-processing unit 21 generates a determination device that processes sensor data for determination of a target sensor by machine learning using sensor data collected in the past from each sensor in the network. The generated determination device is used by the determination processing unit 22 for data processing on the newly collected sensor data, and outputs the determination result for the target sensor and the reliability of the determination result.

The preprocessing unit 21 can generate a plurality of determination devices by performing machine learning using data sets of different scales (sizes) for sensor data to be machine-learned, for example. In that case, the preprocessing unit 21 confirms the accuracy of the determination result obtained by each generated determination device, and determines that the amount of data required for data processing is the smallest among the determination devices that satisfy a predetermined criterion (for example, the determination that the amount of data required for data processing is the smallest. The device) is determined as a default determination device for determination processing by the determination processing unit 22. The preprocessing unit 21 stores one or more generated determination devices in the data storage unit 24.

The preprocessing unit 21 further determines, among the plurality of sensors in the network, the sensor to be operated in the sleep mode by the above-mentioned machine learning. In machine learning, the pre-processing unit 21 operates the target sensor in sleep mode when the target sensor satisfies the condition indicating that the target sensor is less important for the judgment by the judgment device. To determine as. Based on the result of the determination, the preprocessing unit 21 stores the control information indicating the operation mode of each sensor in the network in the data storage unit 24.

As described above, the case where the condition indicating that the target sensor is of low importance is satisfied corresponds to, for example, the target sensor regardless of whether or not the sensor data output from the target sensor is used. This corresponds to the case where the same judgment result can be obtained by the judgment device. This means that the judgment result does not change even if the data processing by the judgment device is performed by using only the sensor data of the sensors existing around the target sensor without using the sensor data of the target sensor itself. To do. In such a case, even if the target sensor is shifted to the sleep mode, it is possible to maintain the determination accuracy of the corresponding determination device without deteriorating it. Therefore, in the present embodiment, by shifting the sensor satisfying the above conditions to the sleep mode, the power consumption of the entire network can be reduced while maintaining the judgment accuracy of the corresponding judgment device.

● Judgment processing unit 22
The determination processing unit 22 controls the operation mode of the sensor according to the control information stored in the data storage unit 24 so that the sensor determined by the preprocessing unit 21 operates in the sleep mode. The determination processing unit 22 can shift the sensor to the sleep mode by transmitting a control signal including a sleep shift instruction for instructing the sensor to shift to the sleep mode.

The determination processing unit 22 further collects sensor data from each sensor operating in the normal mode, and processes the collected sensor data using the determination device stored in the data storage unit 24. The determination device makes a predetermined determination about the target sensor, which is the sensor corresponding to the determination device, by processing the collected sensor data. In the present embodiment, the determination device determines whether the condition of the area where the target sensor is arranged is normal or abnormal. For example, when each sensor is a temperature sensor that measures temperature (temperature) and outputs a measured value, the determination device outputs the corresponding target sensor from peripheral (nearby) sensors as a result of data processing. If a temperature that is extremely different from the temperature is output, a judgment result indicating an abnormality is output.

The determination processing unit 22 further controls the sensor operating in the sleep mode to return the sensor from the sleep mode to the normal mode when a predetermined return condition described later is satisfied. The determination processing unit 22 returns (shifts) the sensor from the sleep mode to the normal mode by transmitting a control signal including a sleep return instruction for instructing the sensor to return from the sleep mode to the normal mode. It is possible to make it.

● Feedback processing unit 23
The feedback processing unit 23 acquires the sensor data used for the determination processing by the determination processing unit 22, and performs machine learning using the acquired sensor data to periodically determine the determination device used for the determination processing. Update. The feedback processing unit 23 stores the updated determination device in the data storage unit 24, so that the updated determination device can be used for the determination processing by the determination processing unit 22.

Further, each time the determination device is updated, the feedback processing unit 23 determines a sensor to be operated in the sleep mode by machine learning in the same manner as the above-mentioned determination by the preprocessing unit 21. Based on the result of the determination, the feedback processing unit 23 updates the control information stored in the data storage unit 24 indicating the operation mode of each sensor in the network.

Hereinafter, the processing executed by the data processing apparatus 20 will be described more specifically with reference to FIGS. 3 to 5.

<Processing procedure for sleep transition control>
FIG. 3 is a flowchart showing a processing procedure of sleep transition control executed in the data processing device 20. FIG. 4 is a diagram showing an example of data processing by the corresponding determination device when the sensors S1, S2, ... Are the input layers and the sensor S1 is the target sensor, and the neural network is applied to machine learning. An example is shown.

In step S301, the preprocessing unit 21 acquires the sensor data collected in the past output from each sensor in the network. Next, in step S302, the preprocessing unit 21 generates a determination device corresponding to the target sensor by performing machine learning (for example, a support vector machine) using the acquired sensor data. The preprocessing unit 21 stores the generated determination device in the data storage unit 24.

The input of each determination device generated in step S302 is a sensor data string output from the target sensor and surrounding sensors. The determination device determines the target sensor (determines whether the condition of the area where the target sensor is arranged is normal or abnormal) by data processing on the input sensor data string. The output of the judgment device is the judgment result and the reliability of the judgment result. In the present embodiment, the determination result is information indicating whether the condition of the area where the target sensor is arranged is normal or abnormal. The reliability is an index showing the certainty of the determination result.

After that, in step S303, the preprocessing unit 21 uses machine learning to determine whether or not to shift the target sensor to the sleep mode according to the importance of the target sensor, which is related to the determination by the determination device corresponding to the target sensor. decide. Specifically, as described above, the preprocessing unit 21 puts the target sensor into sleep mode when the target sensor satisfies the condition (transition condition) indicating that the target sensor is of low importance. If the transition is not satisfied, it is determined not to shift the target sensor to the sleep mode (maintain the normal mode). More specifically, any one of the following transition conditions 1 to 3 or a combination of one or more is applied to the above transition conditions (criteria).

● Transition condition 1
Transition condition 1 is a case where the sensor data output from the target sensor is used in machine learning, and a case where only the sensor data output from the sensor existing around the target sensor is used without using the sensor data. Therefore, the judgment results by the corresponding judgment devices are the same, and the same reliability is obtained as the reliability of the judgment results. When the transition condition 1 is satisfied, the same determination result can be obtained by the determination device with the same reliability without operating the target sensor in the sleep mode and using the sensor data of the target sensor. That is, it is possible to reduce the power consumption while maintaining the determination accuracy of the determination device.

● Transition condition 2
The transition condition 2 is that the index value obtained in machine learning, which indicates the importance of the target sensor corresponding to the determination device, is lower than a predetermined threshold value. As shown in FIG. 4, for example, the index value is a weight value (w1, w2, w3, w4) applied to the sensor data output from the target sensor S1 in the data processing by the determination device. This index is an index showing how important the sensor data of the target sensor S1 itself is in machine learning.

The lower each weight value in FIG. 4, which is an index value, the less the influence of the sensor data of the target sensor S1 on the determination result by the determination device. This means that the same determination result and reliability can be obtained without the sensor data of the target sensor S1 (that is, the power consumption can be reduced while maintaining the determination accuracy of the determination device). Therefore, according to the transition condition 2, when the index value (each weight value) is lower than the threshold value, it is determined that the sensor data of the target sensor is unnecessary, and it is determined that the target sensor is operated in the sleep mode.

● Transition condition 3
The transition condition 3 is that the probability that a specific determination result is obtained as a determination result based on the sensor data collected in the past by the determination device corresponding to the target sensor is lower than a predetermined threshold value. The specific determination result is, for example, a determination result indicating an abnormality. As described above, when the probability that the determination result indicating the abnormality is obtained is low, it is highly possible that the situation is normal for the area to be determined by the determination device. That is, it is considered that there is little need to make a judgment for detecting an abnormal situation (that is, the importance of the target sensor related to the judgment by the judgment device is low). Therefore, even if the target sensor corresponding to the judgment device is shifted to the sleep mode, it is unlikely that the judgment accuracy of the judgment device is lowered, and it is possible to reduce the power consumption while maintaining the judgment accuracy. ..

The result of the determination in step S303 is stored as control information in the data storage unit 24 by the preprocessing unit 21. After the processing in step S303 is completed, in step S304, the determination processing unit 22 determines whether or not to shift the target sensor to the sleep mode according to the control information stored in the data storage unit 24, and shifts the target sensor. In that case, the process proceeds to step S305, and if the process is not performed, the process ends. In step S305, the determination processing unit 22 transmits a control signal including a sleep transition instruction to the target sensor, and ends the processing. According to the sleep transition instruction, the target sensor shifts to the sleep mode and starts the operation in the sleep mode.

<Procedure for wakeup control>
FIG. 5 is a flowchart showing a processing procedure of sleep wakeup control executed in the data processing device 20.

In step S501, the determination processing unit 22 collects sensor data output from each sensor operating in the normal mode in the network. Next, in step S502, the determination processing unit 22 uses each sensor in the network as the target sensor, and uses the corresponding determination device to perform data processing on the collected sensor data, whereby the target sensor is arranged. Judge whether the condition of the area is normal or abnormal.

After that, in step S503, the determination processing unit 22 determines whether or not the target sensor is operating in the sleep mode, and if it is operating in the normal mode, ends the processing. On the other hand, when the target sensor is operating in the sleep mode, the determination processing unit 22 advances the processing to step S504.

In step S504, the determination processing unit 22 periodically determines whether or not to return (shift) the target sensor from the sleep mode to the normal mode according to a predetermined return condition. This return condition (reference) is for determining that a situation has occurred in which it is necessary to improve the determination accuracy of the sensor or the determination device corresponding to the sensor existing in the vicinity of the sensor, or prevent the determination accuracy from being lowered. It is a condition. More specifically, any one of the following recovery conditions 1 to 3 or a combination of one or more is applied.

● Return condition 1
The return condition 1 is that the reliability of the judgment result of the sensor (target sensor) operating in the sleep mode by the judgment device for the sensor is lower (than before). Alternatively, the return condition 1 may be that the reliability of the determination result is lower than a predetermined threshold value. In this case, in order to improve the determination accuracy of the determination device corresponding to the target sensor, the determination processing unit 22 returns the target sensor to the normal mode so that the sensor data of the target sensor can be used for the determination. ..

● Return condition 2
The return condition 2 is that a specific judgment result (in the present embodiment, a judgment result indicating an abnormality) is obtained as a judgment result by the judgment device for the sensor (target sensor) operating in the sleep mode. is there. In this case, the determination processing unit 22 returns the target sensor to the normal mode and displays the sensor data of the target sensor in order to determine the status of the area to be sensed by the target sensor by the determination device with higher accuracy. Make it available for judgment.

● Return condition 3
The return condition 3 is that it is necessary to improve the determination accuracy of the determination device for the sensor (first sensor) operating in the sleep mode with respect to the sensor (second sensor) existing around the sensor. In this case, in order to perform the determination by the determination device corresponding to the second sensor with higher accuracy, the first sensor is returned to the normal mode so that the sensor data of the first sensor can be used for the determination. As an application example of the return condition 3, the determination processing unit 22 has a specific determination result (as a determination result of the determination device corresponding to the second sensor, in which the second sensor has the above-mentioned importance higher than that of the first sensor. In the present embodiment, the first sensor is returned to the normal mode on condition that the determination result indicating the abnormality) is obtained.

The result of the determination in step S504 is stored as control information in the data storage unit 24 by the preprocessing unit 21. After the processing of step S504 is completed, then in step S505, the determination processing unit 22 determines whether or not to return the target sensor from the sleep mode to the normal mode according to the control information stored in the data storage unit 24. When returning, the process proceeds to step S506, and when not returning, the process ends. In step S506, the determination processing unit 22 transmits a control signal including a sleep wakeup instruction to the target sensor, and ends the processing. According to the sleep wakeup instruction, the target sensor wakes up from the sleep mode and starts the operation in the normal mode.

After the target sensor returns to the normal mode in step S506, the determination processing unit 22 may shift the sensor that does not satisfy the conditions related to the return to the sleep mode again within a predetermined period. As a result, it is possible to reduce the power consumption again without lowering the judgment accuracy of the corresponding judgment device.

As described above, in the present embodiment, the data processing device 20 can communicate with each sensor in the network, collects sensor data from each sensor, and performs data processing. The pre-processing unit 21 is a determination device that processes sensor data for determination of the target sensor, and outputs a determination result of the target sensor and the reliability of the determination result. It is generated by machine learning using sensor data collected in the past from each sensor in the network. The determination processing unit 22 determines, among a plurality of sensors in the network, the sensor to be operated in the sleep mode by machine learning. At that time, the determination processing unit 22 puts the target sensor into sleep mode when the target sensor satisfies the condition indicating that the importance of the target sensor related to the determination by the determination device is low in machine learning. Determined as a sensor to operate with. Further, the determination processing unit 22 controls the operation mode of the determined sensor so that the determined sensor operates in the sleep mode.

According to the present embodiment, the target sensor is shifted to the sleep mode based on the importance of each sensor (target sensor) in the sensor processing and the importance related to the judgment by the corresponding determination device. This makes it possible to reduce the number of sensors operating in the normal mode in the network and reduce the power consumption of the entire network while maintaining the accuracy of data processing by the data processing device at a constant level. Further, by reducing the number of sensors operating in the normal mode, it is possible to reduce the amount of communication between the sensors and the data collection device, and it is possible to reduce the processing load in the data collection device. Further, the processing realized by the present embodiment can be realized by introducing additional settings into a part of the existing system without requiring the introduction of an additional device, and thus the implementation is easy.

The gateway device according to the present embodiment can be realized by a computer program for making a computer function as a gateway device. The computer program is stored in a computer-readable storage medium and can be distributed, or can be distributed via a network.

21: Pre-processing unit 22: Judgment processing unit 23: Feedback processing unit 24: Data storage unit

Claims (15)

It is a data processing device that can communicate with each sensor in the network, collects sensor data from each sensor, and processes the data.
The determination of the status of the area where the target sensor is located, a determination unit that performs the data processing for pairs the sensor data collected from sensors operating in a non-power saving mode, the determination result and the determination of the target sensor A generation means for generating the determination device that outputs the reliability of the result by machine learning using sensor data collected in the past from each sensor in the network.
It is a determination means for determining a sensor to be operated in a power saving mode that does not output sensor data among a plurality of sensors in the network by the machine learning, and is related to the determination by the determination device in the machine learning. When the target sensor satisfies the condition indicating that the target sensor is of low importance, the target sensor is determined as a sensor that operates in the power saving mode.
A control means for controlling the operation mode of the sensor so that the sensor determined by the determination means operates in the power saving mode, and
A data processing device characterized by comprising. The conditions are output from a sensor existing around the target sensor without using the sensor data output from the target sensor in the machine learning when the sensor data output from the target sensor is used. The data processing apparatus according to claim 1, wherein the same reliability is obtained as the reliability of the determination result when only the sensor data is used. The data processing apparatus according to claim 1 or 2, wherein the condition includes that the index value indicating the importance of the target sensor obtained in the machine learning is lower than the threshold value. The data processing apparatus according to claim 3, wherein the index value is a weight value applied to sensor data output from the target sensor in the data processing by the determination device. The condition is any one of claims 1 to 4, wherein the probability that a specific determination result is obtained as the determination result by the determination device based on the sensor data collected in the past is lower than the threshold value. The data processing apparatus according to item 1. Further provided with an update means for periodically updating the determination device generated by the generation means by the machine learning using the sensor data newly collected from each sensor in the network.
The determination means according to any one of claims 1 to 5, further comprising determining a sensor to be operated in the power saving mode by the machine learning each time the determination device is updated. Data processing device. The control means returns the sensor operating in the power saving mode to the non-power saving mode that outputs sensor data, provided that the reliability of the judgment result by the judgment device for the sensor is lowered. The data processing apparatus according to any one of claims 1 to 6, characterized in that. The control means puts the sensor operating in the power saving mode into a non-power saving mode that outputs sensor data on condition that a specific judgment result is obtained as a judgment result by the judgment device for the sensor. The data processing apparatus according to any one of claims 1 to 7, wherein the data processing apparatus is restored. The control means uses the first sensor operating in the power saving mode as a sensor on condition that it is necessary to improve the determination accuracy of the determination device for the second sensor existing in the vicinity of the first sensor. The data processing apparatus according to any one of claims 1 to 8, wherein the data processing apparatus returns to a non-power saving mode for outputting data. The second sensor has a higher importance than the first sensor.
The control means returns the first sensor from the power saving mode to the non-power saving mode on condition that a specific judgment result is obtained as a judgment result by the judgment device for the second sensor. 9. The data processing apparatus according to claim 9. The control means according to any one of claims 7 to 10, wherein the control means periodically determines whether or not to return each sensor operating in the power saving mode to the non-power saving mode. Data processing equipment. 7. The control means is characterized in that, after returning to the non-power saving mode, a sensor that does not satisfy the conditions related to the return is shifted to the power saving mode again within a predetermined period of time. The data processing apparatus according to any one of items 11 to 11. The determination device determines whether the condition of the area where the target sensor is arranged is normal or abnormal by the data processing.
The data processing apparatus according to any one of claims 5, 8 and 10, wherein the specific determination result is a determination result indicating that the situation in the area is abnormal. It is a control method of a data processing device that can communicate with each sensor in the network, collects sensor data from each sensor, and processes the data.
The determination of the status of the area where the target sensor is located, a determination unit that performs the data processing for pairs the sensor data collected from sensors operating in a non-power saving mode, the determination result and the determination of the target sensor A generation process in which the determination device that outputs the reliability of the result is generated by machine learning using sensor data collected in the past from each sensor in the network.
It is a determination step of determining a sensor to be operated in a power saving mode that does not output sensor data among a plurality of sensors in the network by the machine learning, and is related to the determination by the determination device in the machine learning. The determination step of determining the target sensor as a sensor that operates in the power saving mode when the target sensor satisfies the condition indicating that the target sensor is of low importance.
A control step of controlling the operation mode of the sensor so that the sensor determined in the determination step operates in the power saving mode,
A method for controlling a data processing device, which comprises. A program for causing a computer to execute each step of the control method of the data processing device according to claim 14.

Images