JP7803083B2 - Management device and management method - Google Patents

Description

The technology disclosed herein belongs to the technical field of management devices and management methods for grid groups that can be used for grid computing processing.

In recent years, vehicles have been equipped with computing devices with relatively high computing power for electronic control. However, these computing devices have not been effectively utilized when the vehicle is not in use, such as when the vehicle is parked. In response to this situation, grid computing using computing devices installed in multiple vehicles is being considered as a way to effectively utilize the computing devices installed in the vehicles.

For example, Patent Document 1 discloses a grid computing management server that uses a communication device mounted on a vehicle. This management server includes a signal receiving unit that receives a signal from the communication device indicating that it is possible to participate in grid computing, a status determining unit that determines whether the processing capacity of the processing device is insufficient, and a response transmitting unit that sends an instruction to the communication device to participate in grid computing if the processing capacity of the processing device is insufficient.

Japanese Patent Application Laid-Open No. 2020-160661

By the way, there are a variety of jobs that can be processed using grid computing. Therefore, in order to properly complete the processing of a job, it is necessary to properly match the job data with the computing device that will process that job data.

Patent Document 1 discloses creating schedule information indicating periods when a vehicle is not being used for driving from the vehicle's usage history, and determining which vehicles will participate in grid computing based on the schedule information. However, the configuration described in Patent Document 1 simply assigns job data to the computing device of an unused vehicle, which could result in insufficient calculation time to complete the job data calculation process. Furthermore, a vehicle user may suddenly use the vehicle for driving contrary to the schedule information, which also interrupts the calculation process and prevents it from being completed. In grid computing, if some calculation processes cannot be completed, the final calculation results cannot be obtained.

The technology disclosed here was developed in light of these issues, and its purpose is to prevent job processing using grid computing from being interrupted.

To solve the above-mentioned problems, the technology disclosed herein targets a management device for a grid group that can be used for grid computing processing, in which each computing device mounted on a plurality of mobile objects performs computational processing as a computation node, and includes a communication unit capable of communicating with each of the plurality of computing devices, and a control unit. The control unit is configured to execute the following operations: an available time estimation process that estimates available time, which is the time available for the grid computing processing in the computing device, based on the operation history of the computing device; a job completion time estimation process that estimates a job completion time, which is the computation time required for the computing device to complete computational processing of job data to be processed; a job success rate estimation process that calculates a job success rate, which is the probability that the computing device will be able to complete computational processing of the job data, based on the estimated available time and the estimated job completion time; a device extraction process that extracts a specific computing device that is a computing device with a job success rate estimated in the job success rate estimation process that is equal to or greater than a predetermined threshold; and a matching process that assigns the job data to the specific computing device if there is only one specific computing device, or to a specific computing device selected from the plurality of specific computing devices if there are multiple specific computing devices.

With this configuration, the available time during which the computing device can be used to compute job data is estimated from the operating history of the computing device. Once the available time is estimated, the job success rate is estimated by comparing it with the job completion time. Job data is then assigned to a specific computing device whose job success rate is equal to or greater than a predetermined threshold, thereby reducing the possibility of interrupting the computation of job data. This makes it possible to prevent interruptions to job computation processing using grid computing.

In the management device, the control unit may be configured to, in the matching process, select the specific arithmetic unit with the lowest job success rate if there are multiple specific arithmetic units.

This configuration can improve the efficiency of job data computation using grid computing. In other words, if job data requiring a short computation time is assigned to a computing device with a long available time, when job data requiring a long computation time later appears, that job data cannot be assigned to the computing device. With this configuration, the job success rate is calculated based on the estimated available time and the estimated job completion time, so a job success rate close to a predetermined threshold means that the available time and the job completion time are close. Therefore, job data can be assigned to a computing device with an available time close to the job completion time. As a result, the efficiency of job computation using grid computing can be improved.

In the management device, the control unit may be configured to calculate the job completion time as a probability distribution with the job completion time as a variable in the job completion time estimation process.

In other words, job completion times vary depending on the computing power and communication status of the computing device. Therefore, if job completion times are calculated as a probability distribution with the job completion times as variables, the range of job completion times is taken into account, allowing for more accurate calculation of job success rates. This makes it possible to more effectively prevent job processing via grid computing from being interrupted.

In one embodiment of the management device, the control unit terminates the device extraction process when a preset number of specific arithmetic units have been extracted during the device extraction process.

This configuration allows the time spent on matching to be minimized, thereby improving the computational efficiency of jobs using grid computing.

In one embodiment, the control unit may be configured to reduce the set number as the importance of the job data increases.

With this configuration, if the set number is small, the time allocated to matching will be reduced, and available time can be used efficiently for job data calculation processing. This improves the probability of calculation completion, especially for job data with high importance. Therefore, it is possible to more effectively prevent job calculation processing using grid computing from being interrupted.

In another aspect of the technology disclosed herein, a management device for a grid group usable for grid computing processing, in which each computing device mounted on a plurality of mobile bodies performs computational processing as a computation node, is provided with a communication unit capable of communicating with each of the plurality of computing devices, and a control unit. The control unit is configured to perform a group identification process to identify a plurality of computation groups each consisting of at least a portion of the plurality of computing devices and which can become the grid group; an available time estimation process to estimate an available time, which is the time available for the grid computing processing, in the computation group identified in the group identification process; a job completion time estimation process to estimate a job completion time, which is the computation time required to complete computational processing of job data to be processed by grid computing; a job success rate estimation process to calculate a job success rate, which is the probability that the computation group will be able to complete computational processing of the job data, from the available time estimated in the available time estimation process and the job completion time estimated in the job completion time estimation process; and extract specific computation groups, which are computation groups whose job success rate estimated in the job success rate estimation process is equal to or exceeds a predetermined threshold.

With this configuration, the available time during which the operation group can be used to calculate job data is estimated from the operation history of each operation device. Once the available time is estimated, the job success rate for the operation group is estimated by comparing it with the job completion time. Job data is then assigned to a specific operation group whose job success rate is equal to or greater than a predetermined threshold, reducing the possibility that the operation processing of the job data will be interrupted. This makes it possible to prevent interruptions to job processing using grid computing. In particular, with this configuration, the job success rate is estimated for each operation group that can become a grid group, improving the accuracy of matching jobs with grid groups.

The technology disclosed herein also covers a method for managing a group of grids that can be used for grid computing processing, in which each computing device mounted on a plurality of mobile objects performs computational processing as a computation node. Specifically, the management method includes an available time estimation step that estimates available time, which is the time available for the grid computing processing, on the computing device based on the operation history of the computing device; a job completion time estimation step that estimates a job completion time, which is the computation time required for the computing device to complete computational processing of job data to be processed; a job success rate estimation step that calculates a job success rate, which is the probability that the computing device will be able to complete computational processing of the job data, based on the operation history of the computing device and the estimated job completion time; a device extraction step that extracts a specific computing device whose job success rate estimated in the job success rate estimation step is equal to or exceeds a predetermined threshold; and a matching step that assigns the job data to the specific computing device if there is one specific computing device, or to a specific computing device selected from the plurality of specific computing devices if there are multiple specific computing devices.

In this configuration, the job success rate is estimated by comparing the available time with the job completion time. Job data is then assigned to a specific processing device whose job success rate is equal to or greater than a predetermined threshold, reducing the possibility of interrupting the processing of job data. This makes it possible to prevent interruptions to job processing using grid computing.

As described above, the technology disclosed herein can prevent job processing via grid computing from being interrupted.

FIG. 1 is a schematic diagram illustrating the configuration of a system including a management device according to a first exemplary embodiment. FIG. 2 is a conceptual diagram for explaining grid computing. FIG. 3 is a block diagram illustrating the configuration of a vehicle. FIG. 4 is a block diagram illustrating the configuration of a user terminal. FIG. 5 is a block diagram showing the configuration of a client server. FIG. 6 is a block diagram showing the configuration of the management server. FIG. 7 is a flowchart showing the processing operations of the management server from when the job is accepted until the job's arithmetic processing is completed. FIG. 8 is a schematic diagram illustrating an example of an image of the job reception screen. FIG. 9 is a flowchart showing the processing operation of the management server in the matching process of allocating each partial job data to a computing device that constitutes grid computing. FIG. 10 is a flowchart showing the processing operation of the management server when calculating the job success rate. FIG. 11 is a diagram illustrating an example of the operation history of a computing device. FIG. 12 is a schematic diagram for explaining the calculation of the job success rate. FIG. 13 is a table showing an example of the above-threshold list. FIG. 14 is a flowchart illustrating the processing operation of the management server according to the second embodiment in a matching process for allocating job data to grid groups. FIG. 15 is a diagram illustrating an example of the operation history of the computing devices that make up the grid group in the second embodiment. FIG. 16 is a table illustrating an example of an operation group identified by the management server according to the second embodiment.

An exemplary embodiment will be described in detail below with reference to the drawings.

[Embodiment 1]
(System configuration)
1 illustrates an example of the configuration of a system 1 including a vehicle 10 having a computing device 105 according to the first embodiment. The system 1 includes a plurality of vehicles 10, a plurality of user terminals 20, a client server 30, and a management server 50. These components can communicate with each other via a communication network 5. Each of the plurality of vehicles 10 is equipped with a computing device 105.

(Grid Computing)
As shown in FIG. 2 , in the system 1 of the first embodiment, grid computing is configured by the arithmetic devices 105 mounted on each vehicle 10. In grid computing, grid computing processing is performed in which job data is processed by an available arithmetic device 105 among the multiple arithmetic devices 105. In other words, each available arithmetic device 105 corresponds to a grid group available for grid computing processing. Job data refers to data before being divided for distributed processing in grid computing. Hereinafter, the arithmetic devices 105 used for grid computing processing may be collectively referred to as a grid group.

When the vehicle 10 requires the computing power of the arithmetic unit 105, the arithmetic unit 105 enters an operating state and the computing power of the arithmetic unit 105 is utilized. For example, when the vehicle 10 is traveling, the computing power of the arithmetic unit 105 is required to control the traveling of the vehicle 10, and the arithmetic unit 105 enters an operating state.

On the other hand, when the computing power of the arithmetic device 105 is no longer needed in the vehicle 10, the arithmetic device 105 enters a non-operating state, and the computing power of the arithmetic device 105 is no longer used. For example, when the vehicle 10 is stopped and the ignition is turned off or the power is turned off, the computing power of the arithmetic device 105 is no longer needed, and the arithmetic device 105 enters a non-operating state.

Here, when the computing power of the computing device 105 is not needed in the vehicle 10, the computing power of the computing device 105 can be provided for grid computing processing, thereby making effective use of the computing power of the computing device 105. The computing device 105 is basically used as a computing resource for grid computing when the vehicle 10 is stopped, i.e., when the computing power of the computing device 105 is not being used for driving control.

In the following description, the continuous time available for grid computing processing on vehicle 10 will be referred to as "available time."

(Vehicle configuration)
The vehicle 10 is a vehicle owned by a user. The user drives the vehicle 10. In this example, the vehicle 10 is a four-wheeled automobile. The vehicle 10 is also equipped with a battery (not shown). Power from the battery is supplied to on-board devices such as the computing device 105. Examples of such vehicles 10 include electric vehicles and plug-in hybrid vehicles.

As shown in FIG. 3, the vehicle 10 includes an actuator 11, a sensor 12, an input unit 101, an output unit 102, a communication unit 103, a memory unit 104, and a computing device 105.

Actuators 11 include drive system actuators, steering system actuators, braking system actuators, etc. Examples of drive system actuators include the engine, transmission, and motor. Examples of braking system actuators include brakes. Examples of steering system actuators include steering.

Sensor 12 acquires various types of information used to control vehicle 10. Examples of sensors 12 include an exterior camera that captures images outside the vehicle, an interior camera that captures images inside the vehicle, radar that detects objects outside the vehicle, a vehicle speed sensor, an acceleration sensor, a yaw rate sensor, an accelerator position sensor, a steering sensor, a key detection sensor, and an ignition sensor.

The input unit 101 inputs information and data. Examples of the input unit 101 include a navigation system that inputs information according to an operation when operated, a camera that inputs an image showing information, and a microphone that inputs audio showing information. The information and data input to the input unit 101 is sent to the calculation device 105.

The output unit 102 outputs information and data. Examples of the output unit 102 include a display unit that outputs an image representing information, and a speaker that outputs audio representing information.

The communication unit 103 transmits and receives information and data. Information and data received by the communication unit 103 is sent to the calculation device 105. The communication unit 103 is composed of, for example, a wireless communication device.

The memory unit 104 stores information and data.

The computing device 105 has a control unit 106 that controls each part of the vehicle 10. In this example, the control unit 106 controls the actuator 11 in accordance with various information obtained by the sensor 12.

The control unit 106 includes a processor, memory, etc. Examples of processors include a CPU (Central Processing Unit) and a GPU (Graphics Processing Unit). The memory stores programs for operating the processor, information and data indicating the processing results of the processor, etc.

The number of processors constituting the control unit 106 may be one or more. Furthermore, the processor constituting the control unit 106 may be either a CPU or a GPU, or both a CPU and a GPU. In this example, the control unit 106 has both a CPU and a GPU. For example, the control unit 106 is composed of one or more ECUs (Electronic Control Units).

In this example, the memory unit 104 stores vehicle information D11, vehicle state information D12, driving history information D13, computing device information D14, and operation history information D15.

<Vehicle Information>
The vehicle information D11 is information related to the vehicle 10. For example, the vehicle information D11 includes a vehicle ID set for the vehicle 10, a user ID set for the user who owns the vehicle 10, vehicle performance information indicating the performance of the vehicle, etc. The vehicle ID is an example of vehicle identification information that identifies the vehicle 10. The user ID is an example of user identification information that identifies the user.

<Vehicle status information>
The vehicle status information D12 indicates the status of the vehicle 10. For example, the vehicle status information D12 includes vehicle position information, vehicle communication information, vehicle power source information, vehicle battery remaining capacity information, vehicle charging information, etc. The vehicle position information indicates the position (latitude and longitude) of the vehicle 10. The vehicle position information can be acquired, for example, by a GPS (Global Positioning System). The vehicle communication information indicates the communication status of the vehicle 10. The vehicle power source information indicates the power source status of the vehicle 10. For example, the vehicle power source information indicates whether the ignition power is on or off, whether the accessory power is on or off, etc. The vehicle battery remaining capacity information indicates the remaining capacity of a battery (not shown) installed in the vehicle 10. The vehicle charging information indicates whether the vehicle 10 is being charged at a charging facility (not shown).

<Driving history information>
The driving history information D13 is information that indicates the driving history of the vehicle 10. For example, the driving history information D13 indicates the position of the vehicle 10 in association with the date and time.

<Calculation device information>
The arithmetic unit information D14 is information related to the arithmetic unit 105. For example, the arithmetic unit information D14 includes an arithmetic unit ID set in the arithmetic unit 105, a vehicle ID set in the vehicle 10 on which the arithmetic unit 105 is mounted, arithmetic unit performance information indicating the performance of the arithmetic unit 105, etc. The arithmetic unit ID is an example of arithmetic unit identification information for identifying the arithmetic unit 105. The performance of the arithmetic unit 105 indicated in the arithmetic unit performance information includes a computing capacity indicating the computing capacity (specifically, maximum computing capacity) of the arithmetic unit 105, the ratio of the CPU to the GPU in the arithmetic unit 105, etc. The computing capacity of the arithmetic unit 105 is the amount of data that the arithmetic unit 105 can calculate per unit time.

<Operation history information>
The operation history information D15 is information related to the operation state of the arithmetic device 105. For example, the operation history information D15 is information indicating the operation state and the non-operation state in association with time, with the operation state being when the arithmetic device 105 is used for driving control and the non-operation state being when the arithmetic device 105 is not used for driving control. The operation history information D15 may indicate the operation state and the non-operation state of the vehicle 10 separately for weekdays and holidays, or separately for each day of the week. Whether the arithmetic device 105 is in the operation state or the non-operation state may be determined based on, for example, the on/off state of the ignition or the power supply, or the on/off state of the parking lock, as described above.

The operating state of the arithmetic device 105 of the vehicle 10 shown in the operation history information D15 is not limited to the above-mentioned driving control, but also refers to a state in which the arithmetic device 105 is used for other purposes excluding when it is used for calculation processing using grid computing. Examples of other purposes of use include driving control of the vehicle 10, control of various functions of the vehicle 10, etc.

The computing device 105 updates the operation history information D15 as appropriate (e.g., periodically). For example, the computing device 105 monitors the operation rate (computing capacity utilization rate) of the computing device 105 and updates the past operation rate of the computing device 105 shown in the operation history information D15 based on the results of this monitoring.

(Configuration of user terminal)
The user terminal 20 is a terminal device owned by a user. The user operates the user terminal 20 to use various functions. The user can also carry the user terminal 20. Examples of such user terminals 20 include smartphones, tablets, and laptop-type personal computers.

As shown in FIG. 4, the user terminal 20 includes an input unit 201, an output unit 202, a communication unit 203, a memory unit 204, and a control unit 205.

The input unit 201 inputs information and data. Examples of the input unit 201 include an operation unit that is operated to input information corresponding to the operation, a camera that inputs images showing information, and a microphone that inputs audio showing information. For example, a user can operate the operation unit to access the navigation system of the vehicle 10 and thereby register a destination. The information input to the input unit 101 is sent to the calculation device 105.

The output unit 202 outputs information and data. Examples of the output unit 202 include a display unit that outputs an image representing information, and a speaker that outputs audio representing information.

The communication unit 203 sends and receives information and data. Information and data received by the communication unit 303 is sent to the control unit 205.

The memory unit 204 stores information and data.

The control unit 205 controls each unit of the user terminal 20. The control unit 205 has a processor, memory, etc. The memory stores programs for operating the processor, information and data indicating the processing results of the processor, etc.

In this example, the storage unit 204 stores terminal information D21, terminal status information D22, and schedule information D23.

<Device Information>
The terminal information D21 is information related to the user terminal 20. For example, the terminal information D21 includes a user terminal ID set in the user terminal 20, user terminal performance information indicating the performance of the user terminal 20, etc. The user terminal ID is an example of user terminal identification information that identifies the user terminal 20.

<Device status information>
The terminal status information D22 is information indicating the status of the user terminal 20. The terminal status information D22 includes user terminal communication status information indicating the communication status of the user terminal 20, etc.

<Schedule Information>
The schedule information D23 indicates the behavior history and behavior schedule of the user who owns the user terminal 20. The schedule information D23 can be acquired by a schedule function installed in the user terminal 20. Specifically, when a user uses the schedule function to input his or her own behavior history and behavior schedule into the user terminal 20, the schedule information D23 indicating the behavior history and behavior schedule of the user is obtained.

(Client-server configuration)
The client server 30 is owned by a client, who requests the calculation of job data. Examples of such clients include companies, research institutes, and educational institutions.

As shown in FIG. 5, the client server 30 includes an input unit 301, an output unit 302, a communication unit 303, a memory unit 304, and a control unit 305.

The input unit 301 inputs information and data. Examples of the input unit 301 include an operation unit that is operated to input information corresponding to the operation, a camera that inputs an image representing information, and a microphone that inputs audio representing information. The information and data input to the input unit 301 is sent to the control unit 305.

The output unit 302 outputs information and data. Examples of the output unit 302 include a display unit that outputs an image representing information, and a speaker that outputs audio representing information.

The communication unit 303 sends and receives information and data. Information and data received by the communication unit 303 is sent to the control unit 305.

The memory unit 304 stores information and data.

The control unit 305 controls each part of the client server 30. The control unit 305 has a processor, memory, etc. The memory stores programs for operating the processor, information and data indicating the processing results of the processor, etc.

In this example, the storage unit 304 stores client information D31 and job data D1.

<Client Information>
The client information D31 is information about the client. The client information D31 includes a client ID set for the client, a client server ID set for the client server 30 owned by the client, a person in charge's name, address, telephone number, etc. The client ID is an example of client identification information that identifies the client. The client server ID is an example of client server identification information that identifies the client server 30.

<Job Data>
The job data D1 is data corresponding to a job and is processed to execute the job.

Job data D1 can be classified by calculation type. Examples of calculation types include CPU-based calculation types and GPU-based calculation types. Job data D1 of the CPU-based calculation type tends to require complex calculations with many conditional branches, such as simulation calculations. Job data D1 of the GPU-based calculation type tends to require a huge amount of simple calculations, such as image processing and machine learning.

Job data D1 can also be classified by the job's processing conditions. Examples of processing conditions include processing conditions that require constant communication and processing conditions that do not require constant communication. Job data D1 with processing conditions that require constant communication requires that the computational resources (i.e., each computing device 105) be able to communicate at all times in grid computing processing. Job data D1 with processing conditions that do not require constant communication does not require that the computational resources be able to communicate at all times in grid computing processing.

The storage unit 304 may also store job information related to jobs. The job information includes job name information indicating the name of the job, job content information explaining the content of the job, job data information related to the job data corresponding to the job, and job delivery date information indicating the delivery date of the job. The job data information indicates the calculation type, processing conditions, required calculation capacity, and importance of the job data.

(Management Server Configuration)
The management server 50 manages the operation of the system 1, which is configured as a grid computing system. The management server 50 is owned by the operator who operates the system 1.

As shown in FIG. 6, the management server 50 includes an input unit 501, an output unit 502, a communication unit 503, a memory unit 504, and a control unit 505. The configurations of the input unit 501, output unit 502, communication unit 503, memory unit 504, and control unit 505 of the management server 50 are the same as the configurations of the input unit 301, output unit 302, communication unit 303, memory unit 304, and control unit 305 of the client server 30.

In this example, the storage unit 504 stores a user table D51, a computing device table D52, a client table D53, a job table D54, a resource table D55, a matching table D56, an over-threshold list D57, job data D1, and calculation result data D2.

<User table>
The user table D51 is a table for managing users. For each user, the user table D51 registers a user ID set for the user, a vehicle ID set for the vehicle 10 owned by the user, a computing device ID set for the computing device 105 owned by the user, a user terminal ID set for the user terminal 20 owned by the user, and the like.

<Calculation Unit Table>
The arithmetic device table D52 is a table for managing the arithmetic devices 105. In the arithmetic device table D52, for each arithmetic device 105, a arithmetic device ID set in the arithmetic device 105, a user ID set for the user who owns the arithmetic device 105, a vehicle ID set for the vehicle 10 in which the arithmetic device 105 is installed, and the like are registered.

Furthermore, the computing device table D52 registers, for each computing device 105, the performance of each computing device 105 (such as computing capacity and CPU/GPU ratio), the operating status of each computing device 105 (operating history and operating schedule), and the like. In other words, the computing device table D52 includes operating status information D5 indicating the operating status of each of the multiple computing devices 105, and performance information D6 indicating the performance of each of the multiple computing devices 105. The performance information D6 includes computing capacity information D7 indicating the computing capacity of each of the multiple computing devices 105.

The management server 50 communicates with each vehicle 10 to periodically obtain information about these calculation devices 105 and update the calculation device table D52.

<Client Table>
The client table D53 is a table for managing clients. For each client, the client table D53 registers a client ID set for that client, a client server ID set for the client server 30 owned by the client, the name, address, telephone number, etc. of the person in charge of the client. The client table D53 records the usage history of grid computing for each client.

<Job Table>
The job table D54 is a table for managing jobs requested by clients. For each job, the job table D54 registers the reception number assigned to that job, the client ID assigned to the client that requested the job, the name and content of the job, etc. The job table D54 also registers for each job the calculation type and processing conditions of the job data corresponding to that job, the required calculation capacity that is the calculation capacity required to calculate the job data, the deadline assigned to that job, the importance of that job, etc.

<Resource Table>
The resource table D55 is a table for managing the computing capacity in grid computing processing. In the resource table D55, for each computing device 105, the computing device ID set for that computing device 105 is registered.

Matching Table
The matching table D56 is a table for managing the results of the matching process between jobs and grid groups. For each job, the matching table D56 registers the reception number set for the job, the job data corresponding to the job, the arithmetic device ID set for each arithmetic device 105 that constitutes the grid group assigned to the job data by the matching process, and the like.

<Over-threshold list>
The over-threshold list D57 is a list recording specific arithmetic devices whose job success rate Ps, which is the probability that the arithmetic device 105 can complete the arithmetic processing of each partial job data obtained by dividing the job data D1, is equal to or exceeds a predetermined threshold when the arithmetic device 105 calculates the job success rate Ps for each partial job data. The over-threshold list D57 is created for each partial job data. The method of calculating the job success rate Ps will be described later.

<Job Data>
The job data D1 stored in the storage unit 504 is the accepted job data D1. This job data D1 is the job data before being divided into partial job data for distributed processing. In the first embodiment, the job data D1 is divided into partial job data after matching the job data D1 with a grid group (step S2, which will be described later).

<Calculation result data>
The calculation result data D2 stored in the storage unit 504 is calculation result information calculated by grid computing processing, and indicates the results of the calculation.

(Grid computing processing)
Next, the grid computing process will be described with reference to FIG.

First, in step S1, the control unit 505 accepts a job. Specifically, in response to an operation by a client's staff member, the client server 30 sends a job request application to the management server 50. In response to that application, the control unit 505 of the management server 50 performs the following processing.

The control unit 505 requests the client server 30 to send the information required to accept the job (specifically, client information about the client requesting the job and job information about the job). In this example, the control unit 505 sends image data of the job acceptance screen to the client server 30. The control unit 305 of the client server 30 reproduces the image of the job acceptance screen from the image data and outputs (displays) the image on the output unit 302 (display unit).

As shown in FIG. 8, the job reception screen is a screen for entering information required to accept a job. The job reception screen is provided with a client name input field R101 for entering the client name, a person in charge name input field R102 for entering the client person in charge name, an address input field R104 for entering the client address, a job name input field R111 for entering the job name, a job content input field R112 for entering a description of the job content, a calculation type input field R113 for entering the calculation type of the job data corresponding to the job, a processing condition input field R114 for entering the processing conditions of the job data, a required calculation amount input field R115 for entering the required calculation amount of the job data, a delivery date input field R116 for entering the delivery date of the job, and a register button B100.

The client staff operates the input unit 301 of the client server 30 to input the necessary information into the job reception screen. This inputs client information about the client requesting the job and job information about the job. Once this information has been input, the client staff operates the input unit 301 of the client server 30 to press the registration button B100 on the job reception screen. When the registration button B100 is pressed, the control unit 305 of the client server 30 sends the information entered into the job reception screen (client information and job information) to the management server 50. The control unit 505 of the management server 50 receives the client information and job information and updates the client table D53 and job table D54.

Next, in step S2, the control unit 505 matches each grid group with the job. The control unit 505 compares the job data information and causes each computing device 105 that constitutes a grid group suitable for executing the job to execute the job. The control unit 505 updates the matching table D56 and stores the results of this matching.

Next, in step S3, the control unit 505 assigns partial job data obtained by dividing the job data D1 to each arithmetic device 105 that constitutes the grid group. The assignment of partial job data to the arithmetic devices 105 will be described later.

Next, in step S4, the control unit 505 causes each assigned computing device 105 to perform computational processing of the partial job data.

Next, in step S5, the control unit 505 determines whether all of the arithmetic units 105 to which the job data D1 has been distributed have completed calculations. If all of the arithmetic units 105 have completed calculations, the control unit 505 proceeds to step S6. If at least some of the arithmetic units 105 have not completed calculations, the control unit 505 returns to step S4 and continues the calculation of the job.

Then, in step S6, the operator of system 1 awards a reward to the user who provided the computing power of the computing device 105 for the grid computing process. Examples of rewards awarded to users include points that can be used in system 1, virtual currency, and product discount benefits. For example, the control unit 505 of the management server 50 performs processing to award a reward to the user who provided the computing power of the computing device 105 for the grid computing process. Examples of processing to award a reward include processing to associate the "user ID" set for the user with "points" (or virtual currency) that can be used in system 1 and register them in user table D51, and processing to send information indicating product discount benefits to the user terminal 20 owned by the user.

Furthermore, a reward may be granted by the client to a user who provides the computing power of the computing device 105 for grid computing processing. For example, the control unit 305 of the client server 30 may execute processing to grant a reward to a user who provides the computing power of the computing device 105 for grid computing processing.

(Partial job data allocation)
As described above, grid computing performs distributed processing in which the arithmetic processing of partial job data obtained by dividing job data D1 is assigned to each arithmetic device 105 and processed in parallel. In this distributed processing, the arithmetic processing of job data D1 itself is not completed until the arithmetic processing of all partial job data is completed. In other words, if some arithmetic devices 105 transition to an operating state and the arithmetic processing of the partial job data in that arithmetic device 105 is interrupted, the arithmetic processing of job data D1 is interrupted. This results in the calculation results performed by the other arithmetic devices 105 being wasted. In particular, in this embodiment, the arithmetic devices 105 forming the grid group are arithmetic devices mounted on the vehicle 10. Therefore, they may be suddenly used for driving control, which may interrupt the arithmetic processing of partial job data. To avoid this, it is possible to estimate the continuous available time of each arithmetic device 105 in advance based on the operation history information D15 and assign partial job data to the arithmetic device with the longest available time. However, with this method, partial job data with a short calculation time will be assigned to a computing device 105 with a long available time, and there is a risk that the computing device 105 will become unavailable if partial job data requiring a long calculation time arises later.

In this first embodiment, the method of allocating partial job data is devised so that partial job data is allocated to a computing device 105 that has a high probability of completing the computational processing, while preventing the computing device 105 from becoming over-specified for the partial job data. Specifically, a job success rate, which is the probability that computational processing of partial job data can be completed within the available time, is estimated for each computing device 105 that makes up the grid group, and partial job data is allocated to each computing device 105 based on this job success rate.

The allocation of partial job data to the arithmetic device 105 will be described in detail below with reference to the flowcharts in Figures 9 and 10, and Figures 11 to 13. Note that the allocation of partial job data described below is performed by hardware constituting the control unit 505 of the management server 50 executing a program for matching partial job data with the arithmetic device 105. Note that the following description will be given assuming that there is at least one specific arithmetic device, which will be described later.

Figure 9 is a flowchart showing the processing operation of the control unit 505 when allocating partial job data to each computing device 105. The control unit 505 executes this flowchart for each piece of job data D1.

First, in step S31, the control unit 505 divides the job data D1 into partial job data. The number of divisions may be determined based on the number of computing devices 105 that make up the grid group, or may be determined by dividing the job data D1 into parts that are easy to divide based on the content of the job data D1.

Next, in step S32, the control unit 505 selects the partial job data to be assigned.

Next, in step S33, the control unit 505 estimates the job success rate for the selected partial job data for each computing device 105. The control unit 505 estimates the job success rate based on the flowchart shown in FIG. 10.

When calculating the job success rate, the control unit 505 first estimates the continuous available time of the target computing device 105 from the operation history information D15 of the computing device 105 in step S331. In detail, the control unit 505 estimates a first probability distribution P1(T) for the computing device 105, with the available time T as a variable.

Figure 11 shows an example of the operation history of the target computing device 105. In Figure 11, the horizontal axis represents time, which runs from midnight to 24:00. ON in Figure 11 indicates that the computing device 105 is in an operating state, and OFF indicates that the computing device 105 is in a non-operating state. The operation history shown in Figure 11 is the operation history for weekdays. The control unit 505 may extract only the operation history for the same day as the current day from the operation history information D15. In particular, when estimating available hours on holidays, it is preferable to extract only the operation history for holidays. Note that although Figure 11 also identifies the locations where the devices are not operating, the locations do not need to be specified as long as the times at which the devices are not operating can be identified.

As shown in Figure 11, consecutive non-operating times fluctuate daily depending on the user's behavior. For this reason, available time is not uniquely determined, but has a certain range. Therefore, estimating it as a probability distribution better reflects the actual available time. The control unit 505 applies the current time and estimates the available time as the period from the current time to the end of the non-operating time to which that time belongs. For example, in the operation history shown in the upper diagram of Figure 11, if the current time is 1:00, the end of the non-operating time is 7:30, so the available time is 6 hours and 30 minutes from 1:00 to 7:30. The available time is then calculated for multiple samples to calculate the first probability distribution P1(T) as shown in Figure 12.

Next, in step S332, the control unit 505 estimates the job completion time of the partial job data. Specifically, the control unit 505 calculates a second probability distribution P2(t) with the job completion time t as a variable. That is, the job completion time varies depending on various conditions, such as limitations due to the communication status and charging status of the computing device 105. For this reason, the job completion time is not uniquely determined but has a certain range. Therefore, estimating it as a probability distribution better reflects the actual job completion time. The control unit 505 calculates the second probability distribution P2(t) based on the required calculation amount R115 input at the time of reception, the calculation capacity of the computing device 105 (calculation capacity recorded in the computing device table D52), the calculation time required for similar partial job data calculations performed in the past, and other factors. For example, when the calculation capacity of the current computing device 105 is higher than the calculation capacity of a computing device that previously performed calculations on similar partial job data, the control unit 505 estimates that the current job completion time will be shorter than the past calculation.

Next, in step S333, the control unit 505 extracts N samples with different available times T from the first probability distribution P1(T). Here, six samples, T1 to T6, are extracted, as shown in FIG. 12. The number of samples may be changed depending on the importance of the job. For example, the number of samples may be increased as the importance of the job increases. Samples may be extracted randomly, or may be more likely to be extracted at times with higher probability. The importance of the job may be input by a person in charge of the client server 30, or may be set by the control unit 505 based on the required calculation amount and delivery date of the job.

Next, in step S334, the control unit 505 extracts M samples with different job completion times t from the second probability distribution P2(t). Here, five samples, t1 to t5, are extracted as shown in FIG. 12. The number of samples may be changed depending on the importance of the job. For example, the number of samples may be increased as the importance of the job increases. Samples may be extracted randomly, or may be extracted more easily at times with higher probability.

Next, in step S335, the extracted samples are compared to calculate the number K of combinations in which the available time T is equal to or greater than the job completion time t. As shown in Figure 12, the only time in which the available time is less than the job completion time is when the available time is T1 and the job completion time is t5. There are 6 x 5 = 30 combinations of the available time T and the job completion time t, so the number of combinations in which the available time T is equal to or greater than the job completion time t is 29.

Then, in step S336, the number K calculated in step S335 is divided by the total number of combinations (M x N), and the result multiplied by 100 is estimated as the job success rate. Here, the job success rate is as follows:

Job success rate = {29/(6 x 5)} x 100 = 96.7 (%)
After step S336, the control unit 505 returns and proceeds to the next step S34.

Returning to FIG. 9, after calculating the job success rate, in step S34, the control unit 505 extracts the computing device 105 that calculated the job success rate as a specific computing device if the job success rate is equal to or greater than a predetermined threshold. Here, the extracted specific computing device is recorded in the above-threshold list D57 together with the job success rate, as shown in FIG. 13. Vehicle A, vehicle C, and vehicle D shown in the specific computing device column in FIG. 13 refer to vehicles equipped with specific computing devices. The specific computing device may be indicated by the computing device ID of the computing device 105 rather than by the vehicle. This step S34 is executed each time calculation of the job success rate for one computing device 105 is completed.

Next, in step S35, the control unit 505 determines whether the number of specific arithmetic units has reached a preset number. This preset number is set according to the importance of the job. For example, the higher the importance of the job, the smaller the preset number. If the control unit 505 determines that the number of specific arithmetic units has reached the preset number (YES), the control unit 505 proceeds to step S37. On the other hand, if the control unit 505 determines that the number of specific arithmetic units is less than the preset number (NO), the control unit 505 proceeds to step S36. Note that if the importance of a job is high, the importance of the partial job data of that job will naturally be high, so the aforementioned "job importance" may be interpreted as "importance of partial job data."

In step S36, the control unit 505 determines whether the comparison of the job success rate with the threshold has been completed for all arithmetic devices 105. If the comparison has been completed for all arithmetic devices 105 (YES), the control unit 505 proceeds to step S37. On the other hand, if the comparison has not been completed for some arithmetic devices 105 (NO), the control unit 505 returns to step S33 and estimates the job success rate for another arithmetic device 105.

In step S37, the extraction of specific arithmetic units is completed. By completing the extraction of specific arithmetic units when the number of specific arithmetic units reaches the set number, the time allocated to allocating partial job data can be minimized, thereby improving the efficiency of the arithmetic processing of job data D1.

In the next step S38, the control unit 505 assigns the partial job data to the specific processing device with the lowest job success rate among the specific processing devices. In this embodiment 1, the job success rate is calculated based on the estimated available time and the estimated job completion time, so a job success rate close to a predetermined threshold means that the available time is close to the job completion time. Therefore, the partial job data can be assigned to a processing device 105 with a high job success rate and an available time close to the job completion time. This increases the probability that the processing of the partial job data will be completed, while preventing the processing device 105 from being over-specified for the partial job data. Note that when only one specific processing device is extracted in step S34, the specific processing device has the lowest job success rate, and therefore the partial job data is assigned to that specific processing device.

Then, in step S39, the control unit 505 determines whether allocation of all the divided partial job data has been completed. If the answer is YES, meaning that allocation of all the partial job data has been completed, the control unit 505 returns and proceeds to step S4. On the other hand, if the answer is NO, meaning that allocation of some of the partial job data has not been completed, the control unit 505 proceeds to step S40, selects another partial job data, and returns to step S33.

When the control unit 505 calculates the job success rate for multiple partial job data and finds that the specific arithmetic unit with the lowest job success rate is the same arithmetic unit 105, it assigns each partial job data to a different arithmetic unit 105 based on the over-threshold list. At this time, the control unit 505 preferentially assigns partial job data to the specific arithmetic unit with the lowest job success rate. Note that the control unit 505 may not subsequently calculate the job success rate for other partial job data for the arithmetic unit 105 to which partial job data has been assigned.

Although the example shown here is one in which there is at least one specific arithmetic unit, if there are no specific arithmetic units, i.e., no arithmetic units 105 with a job success rate exceeding the threshold, the control unit 505 may assign the partial job data to the arithmetic unit 105 with the highest job success rate, or may reset the threshold and extract a specific arithmetic unit again. Also, the allocation of job data to the currently selected grid group may be stopped, and another grid group may be newly selected, and job success rate estimation and specific arithmetic unit extraction may be performed for the arithmetic units 105 that make up that other grid group.

Furthermore, the loop of steps S33 to S36 is executed for each computing device 105 until the set number of specific computing devices have been extracted, or the job success rate has been calculated for all computing devices 105 and a determination has been made as to whether they are specific computing devices. Therefore, when the number of specific computing devices reaches the set number after job success rate calculation has been completed for only some of the computing devices 105 that make up the grid group, job success rate calculation is not performed for the remaining computing devices 105 in the grid group.

Furthermore, when there are a large number of arithmetic devices 105 constituting the grid group matched with the job in step S2, the control unit 505 may allocate the partial job data to only some of the arithmetic devices 105, rather than all of the arithmetic devices 105 being candidates for allocation. In this case, the control unit 505 sets the number of arithmetic devices 105 that are candidates for allocation to a number equal to or greater than the number of divisions of the job data D1, i.e., the number of partial job data. The control unit 505 may increase the number of arithmetic devices 105 that are candidates for allocation as the importance of the job increases.

Therefore, the management server 50 according to the first embodiment includes a communication unit 503 capable of communicating with each of the plurality of arithmetic units 105, and a control unit 505. The control unit 505 performs an available time estimation process (step S331) that estimates the continuous available time of the arithmetic unit 105 from the operation history of each arithmetic unit 105, a job completion time estimation process (step S332) that estimates the job completion time, which is the calculation time required for the arithmetic unit 105 to complete the calculation processing of the partial job data to be processed, and a job completion time estimation process (step S333) that estimates the time required for the arithmetic unit 105 to complete the calculation processing of the partial job data to be processed based on the estimated available time and the estimated job completion time. The system is configured to execute a job success rate estimation process (step S33, steps S333-S336) that calculates a job success rate, which is the probability that the computational processing of job data can be completed; a device extraction process (steps S337, S338) that extracts a specific computing device, which is a computing device 105 whose job success rate estimated in the job success rate estimation process is equal to or exceeds a predetermined threshold; and a matching process (step S37) that assigns partial job data to the specific computing device if there is only one specific computing device, or to a specific computing device selected from the multiple specific computing devices if there are multiple specific computing devices. This assigns partial job data to a computing device 105 whose job success rate is equal to or exceeds the predetermined threshold, thereby reducing the likelihood that computational processing of the partial job data will be interrupted. As a result, interruptions to job computational processing using grid computing can be reduced.

In particular, in this embodiment 1, when there are multiple specific arithmetic units in the matching process, the control unit 505 selects the specific arithmetic unit with the lowest job success rate. This allows partial job data to be assigned to a arithmetic unit 105 with a high job success rate and an available time close to the job completion time. This increases the probability that arithmetic processing of partial job data will be completed, while preventing the arithmetic unit 105 from becoming over-specified for the partial job data. As a result, the efficiency of job calculations using grid computing can be improved.

In addition, in this embodiment 1, in the job completion time estimation process, the control unit 505 calculates the job completion time as a second probability distribution P2(t) with the job completion time as a variable. This takes into account fluctuations in job completion time, making it possible to calculate the job success rate with greater accuracy. As a result, it is possible to more effectively prevent interruptions to job calculation processing using grid computing.

In addition, in this embodiment 1, the control unit 505 also calculates the available time of each computing device 105 as a first probability distribution P1(T) using the available time as a variable. This allows for consideration of fluctuations in available time, making it possible to calculate the job success rate with greater accuracy. As a result, it is possible to more effectively prevent interruptions to job computation processing using grid computing.

In addition, in this embodiment 1, the control unit 505 terminates the device extraction process when a preset number of specific arithmetic devices have been extracted. This makes it possible to minimize the time spent matching arithmetic devices 105 with partial job data. As a result, it is possible to improve the efficiency of job calculations using grid computing.

In particular, in this embodiment 1, the control unit 505 performs device extraction processing each time the job success rate estimation processing is completed for one computing device 105, and determines whether a set number of specific computing devices have been extracted. As a result, when the job success rate calculation has been estimated for only some of the computing devices 105 that make up a grid group, if a set number of specific computing devices have been extracted, it is no longer necessary to calculate the job success rate for the remaining computing devices 105 in the grid group. As a result, the time spent matching computing devices 105 with partial job data can be minimized, thereby improving the efficiency of job calculations using grid computing.

In addition, in this embodiment 1, the control unit 505 sets a smaller set number as the job becomes more important. In other words, if the set number is smaller, less time is allocated to matching, and available time can be used efficiently for arithmetic processing of partial job data. This can improve the probability of completing arithmetic, particularly for partial job data of highly important jobs. Furthermore, if the set number is small, the number of specific arithmetic devices extracted decreases, increasing the likelihood that only arithmetic devices 105 with a job success rate higher than the threshold will be extracted as specific arithmetic devices. This increases the likelihood that partial job data of highly important jobs will be assigned to arithmetic devices 105 with a high job success rate. Therefore, it is possible to more effectively prevent job arithmetic processing using grid computing from being interrupted.

[Embodiment 2]
Hereinafter, the second embodiment will be described in detail with reference to the drawings. In the following description, parts common to the first embodiment will be denoted by the same reference numerals, and detailed description thereof will be omitted.

In the above-described first embodiment, partial job data obtained by dividing job data D1 is assigned to the computing device 105 based on the job success rate. This second embodiment differs from the above-described first embodiment in that the job success rate of the grid group is taken into consideration when matching the job with the grid group.

The matching between jobs and grid groups will be explained in detail below with reference to the flowchart in Figure 14 and Figures 15 and 16. Note that the matching process described below is performed by hardware constituting the control unit 505 of the management server 50 executing a program for matching jobs and grid groups.

Figure 14 is a flowchart showing the processing operations of the control unit 505 when matching a job with a grid group. The control unit 505 executes this flowchart for each job data D1.

First, in step S201, the control unit 505 identifies a computation group consisting of multiple computation devices 105. Each of these computation groups is a computation resource for grid computing, i.e., a group that can become a grid group. The control unit 505 uses the operation history of each computation device 105 to identify downtime periods during which the computation device 105 is continuously inactive for a predetermined period of time or more, and the location of the computation device 105 during the downtime periods (i.e., the vehicle's parked location). The control unit 505 identifies a combination of multiple computation devices whose downtime periods overlap and whose locations are close to each other as a computation group.

Figure 15 shows an example of the operation history of the calculation device 105 on a weekday. Vehicles A to D are different vehicles, each equipped with a calculation device 105. The apartment complexes are the same apartment complex, and company A is the same company. Company A and company B are different companies. As shown in Figure 15, the non-operational times of multiple calculation devices 105 overlap in the apartment complex and company A. Note that while Figure 15 shows the operation history for a specific day, the control unit 505 also takes into account the operation history for other weekdays. Also, while the grid group is composed of four vehicles 10 each equipped with a calculation device 105, the number of vehicles 10 each equipped with a calculation device 105 that constitute the grid group may be less than three or five or more. In either case, the control unit 505 identifies the calculation group by taking into account the operation history of all calculation devices 105. Furthermore, in the case of a holiday, the control unit 505 identifies the calculation group based on the operation history for the holiday.

The control unit 505 identifies multiple calculation groups, allowing for overlapping calculation devices 105 between groups. This is because the available time varies depending on the combination of calculation devices 105. The control unit 505 identifies multiple calculation groups and creates a table such as that shown in FIG. 16. In the groups illustrated in FIG. 16, group A is the group to which the calculation devices 105 of all vehicles 10, vehicles A to D, belong; group B is the group to which the calculation devices 105 of vehicles A, C, and D belong; group C is the group to which the calculation devices 105 of vehicles A and C belong; and group D is the group to which the calculation devices 105 of vehicles A, B, and D belong. In this way, the control unit 505 identifies multiple groups with different numbers and combinations of calculation devices 105.

Next, in step S202, the control unit 505 calculates a third probability distribution in which the available time of the calculation group is used as a variable, taking into account fluctuations in the length of overlapping non-operational time from the operation history of each calculation device 105 that makes up the calculation group.

Next, in step S203, the control unit 505 calculates a fourth probability distribution in which the job completion time of job data D1 is used as a variable, based on the past calculation times of similar job data D1 and the calculation capabilities of each calculation device 105 that make up the calculation group.

Next, in step S204, the control unit 505 calculates a job success rate, which is the probability that the calculation processing of job data D1 can be completed, for each calculation group. The method of calculating the job success rate is the same as the method of calculating the job success rate described in embodiment 1 above, except that the available time is changed from the available time of the calculation device 105 to the available time of the calculation group, and the job completion time of the partial job data is changed to the available time of job data D1. The control unit 505 calculates the job success rate based on the third probability distribution calculated in step S203 and the fourth probability distribution calculated in step S204.

Next, in step S205, the control unit 505 extracts operation groups whose job success rates are equal to or greater than a predetermined threshold as specific operation groups. This step S205 is executed each time the job success rate of an operation group is calculated.

Next, in step S206, the control unit 505 determines whether the number of specific operation groups has reached a preset number. This preset number is set according to the importance of the job. For example, the higher the importance of the job, the smaller the preset number. If the control unit 505 determines that the number of specific operation groups has reached the preset number (YES), the control unit 505 proceeds to step S208. On the other hand, if the control unit 505 determines that the number of specific operation groups is less than the preset number (NO), the control unit 505 proceeds to step S207.

In step S207, the control unit 505 determines whether the comparison of the job success rate with the threshold value has been completed for all operation groups. If the result is YES, meaning that the comparison has been completed for all operation groups, the control unit 505 proceeds to step S208. On the other hand, if the result is NO, meaning that the comparison has not been completed for some operation groups, the control unit 505 returns to step S202 and estimates the job success rate for another operation group.

In step S208, extraction of the specific operation group is completed.

In the next step S209, the control unit 505 assigns job data D1 to the specific operation group with the lowest job success rate among the specific operation groups. Note that if only one specific operation group is extracted in step S203, that specific operation group becomes the specific operation device with the lowest job success rate, and therefore job data D1 is assigned to that specific operation group.

In this second embodiment, the loop of steps S202 to S207 is also executed for each operation group until the set number of specific operation groups have been extracted, or the job success rate has been calculated for all operation groups and a determination has been made as to whether they are specific operation groups. Therefore, when the number of specific operation groups reaches the set number after the job success rate has been calculated for only some of the operation groups that are candidates for assignment of job data D1, the job success rate will not be calculated for the remaining operation groups.

Therefore, in this second embodiment, the control unit 505 of the management server 50 performs a group identification process (step S201) for identifying a plurality of operation groups each consisting of at least a portion of a plurality of operation devices 105 and capable of becoming a grid group for grid computing processing, an available time estimation process (step S202) for estimating the continuous available time for the operation group identified in the group identification process, a job completion time estimation process (step S203) for estimating the job completion time, which is the calculation time required to complete the operation processing of job data D1 to be processed by grid computing, and a calculation process (step S204) for estimating the available time estimated in the available time estimation process. The system is configured to execute a job success rate estimation process (step S204) that calculates a job success rate, which is the probability that a computation group can complete a job, from the job completion time estimated by the job completion time estimation process; a group extraction process (step S205) that extracts specific computation groups, which are computation groups whose job success rates estimated by the job success rate estimation process are equal to or greater than a predetermined threshold; and a matching process (step S209) that assigns job data D1 to the specific computation group if there is only one, or to a specific computation group selected from the multiple specific computation groups if there are multiple specific computation groups. This assigns job data D1 to a specific computation group whose job success rate is equal to or greater than the predetermined threshold, thereby reducing the likelihood that computation processing of job data D1 will be interrupted. This reduces the likelihood of job computation processing using grid computing being interrupted. In particular, this configuration estimates the job success rate for each computation group that can become a grid group, thereby improving the accuracy of matching between jobs and grid groups.

(Other embodiments)
The technology disclosed herein is not limited to the above-described embodiments, and can be substituted within the scope of the claims.

For example, in the above-described first and second embodiments, the computational resource is an arithmetic device 105 mounted on a vehicle 10 as a mobile body. However, this is not limiting, and the computational device 105 may be mounted on a mobile body other than a vehicle 10. Examples of such mobile bodies include transportation machinery and mobile information terminals. Examples of transportation machinery include motorcycles, railway vehicles, ships, aircraft, and drones. Examples of mobile information terminals include notebook personal computers, tablets, and smartphones. In this case, the operating state of the computational device of the mobile information terminal can be determined, for example, based on whether a specific app is running.

Furthermore, in the above-described embodiment, the memory unit 504 of the management server 50 may be configured from a single storage device or from multiple storage devices. The multiple storage devices may be consolidated into a single management server 50, or may be distributed across multiple management servers 50 (not shown) that communicate with each other via the communication network 5.

Furthermore, in the above-described embodiment, the control unit 505 of the management server 50 may be configured as a single control unit, or may be configured as multiple control units. The multiple control units may be consolidated into a single management server 50, or may be distributed across multiple management servers 50 that communicate with each other via the communication network 5.

The above-described embodiments are merely examples and should not be interpreted as limiting the scope of the present disclosure. The scope of the present disclosure is defined by the claims, and all modifications and variations that fall within the equivalent range of the claims are within the scope of the present disclosure.

The technology disclosed herein is useful for managing a group of grids that can be used for grid computing, in which each computing device mounted on multiple mobile objects performs computational processing as a computation node.

10. Vehicles (moving objects)
50 Management server (management device)
503 Communication unit 505 Control unit

Claims (6)

A management device for a grid group that can be used for grid computing processing in which each computing device mounted on a plurality of mobile objects performs computing processing as a computing node,
a communication unit capable of communicating with each of the plurality of arithmetic units;
a control unit,
The control unit
an available time estimation process for estimating an available time, which is a time available for the grid computing process in the computing device, from an operation history of the computing device;
a job completion time estimation process for estimating a job completion time, which is a calculation time required for the arithmetic device to complete arithmetic processing of job data to be processed;
a job success rate estimation process for calculating a job success rate, which is a probability that the arithmetic device can complete arithmetic processing of the job data, based on the estimated available time and the estimated job completion time;
a device extraction process for extracting a specific arithmetic device that is a arithmetic device whose job success rate estimated in the job success rate estimation process is equal to or greater than a predetermined threshold;
a matching process of allocating the job data to the specific arithmetic unit when there is one specific arithmetic unit, or to a specific arithmetic unit selected from the specific arithmetic units when there are multiple specific arithmetic units;
A management device configured to be able to execute the above. 2. The management device according to claim 1,
The management device is characterized in that, in the matching process, when there are a plurality of the specific arithmetic units, the control unit selects the specific arithmetic unit with the lowest job success rate. 3. The management device according to claim 1,
The management device is characterized in that, in the job completion time estimation process, the control unit calculates the job completion time as a probability distribution using the job completion time as a variable. The management device according to any one of claims 1 to 3,
The management device according to claim 1, wherein the control unit terminates the device extraction process when a preset number of specific arithmetic units have been extracted in the device extraction process. A management device for a grid group that can be used for grid computing processing in which each computing device mounted on a plurality of mobile objects performs computing processing as a computing node,
a communication unit capable of communicating with each of the plurality of arithmetic units;
a control unit,
The control unit
a group identification process for identifying a plurality of operation groups each of which is composed of at least a portion of the plurality of operation devices and which can become the grid group;
an available time estimation process for estimating an available time that is a time that can be used for the grid computing process in the computing group identified in the group identification process;
a job completion time estimation process for estimating a job completion time, which is a calculation time required to complete arithmetic processing of the job data to be processed by grid computing;
a job success rate estimation process that calculates a job success rate, which is the probability that the operation group can complete the operation processing of the job data, from the available time estimated in the available time estimation process and the job completion time estimated in the job completion time estimation process;
a group extraction process for extracting specific operation groups, which are operation groups whose job success rates estimated in the job success rate estimation process are equal to or greater than a predetermined threshold;
a matching process of allocating the job data to the specific operation group if there is one specific operation group, or to a specific operation group selected from the multiple specific operation groups if there are multiple specific operation groups;
A management device configured to be able to execute the above. A management method for managing, by a management device, a group of grids available for grid computing processing in which each computing device mounted on a plurality of mobile objects performs computing processing as a computing node, the method comprising:
an available time estimation step of estimating an available time, which is a time available for the grid computing process in the computing device, from an operation history of the computing device;
a job completion time estimation step of estimating a job completion time, which is a calculation time required for the arithmetic device to complete arithmetic processing of job data to be processed;
a job success rate estimation step of calculating a job success rate, which is a probability that the arithmetic device can complete the arithmetic processing of the job data, based on the operation history of the arithmetic device and the estimated job completion time;
a device extraction step of extracting a specific arithmetic device that is a arithmetic device whose job success rate estimated in the job success rate estimation step is equal to or greater than a predetermined threshold;
a matching step of assigning the job data to the specific arithmetic device if there is one specific arithmetic device, or to a specific arithmetic device selected from the multiple specific arithmetic devices if there are multiple specific arithmetic devices .