WO2012081718A1

WO2012081718A1 - Performance prediction device, performance prediction method and performance prediction program

Info

Publication number: WO2012081718A1
Application number: PCT/JP2011/079276
Authority: WO
Inventors: 圭介梅津
Original assignee: 日本電気株式会社
Priority date: 2010-12-15
Filing date: 2011-12-13
Publication date: 2012-06-21
Also published as: JPWO2012081718A1

Abstract

An information provision server cannot output information about communication speed if no measurement information under specified communication conditions is stored therein. The performance prediction device is connected to a log storage means in which a plurality of records of parameters of a plurality of types, each containing one of a plurality of values which a parameter of an applicable type can take and a performance value, are stored. The performance prediction device determines a relaxed condition, which represents the most reliable combination determined on the basis of the number of applicable records in the log storage means, that match a given combination out of combinations of the values of the parameters of the plurality of types; computes a relaxed-condition performance value from the performance values contained in the records that match the applicable relaxed condition; and upon receiving a prediction condition, which is one of the plurality of values which a parameter of an applicable type can take, for each of the parameters of the plurality of types, outputs a value obtained by multiplying a coefficient, which is determined according to the difference between the applicable prediction condition and the relaxed condition, by the relaxed-condition performance value as a predicted performance value

Description

Performance prediction apparatus, performance prediction method, and performance prediction program

The present invention relates to a performance prediction apparatus, a performance prediction method, and a performance prediction program.

In recent years, the number of mobile terminals used by users at destinations has increased. For example, a small terminal such as a mobile phone, a PHS (Personal Handy-phone System), a PDA (Personal Digital Assistant), or an in-vehicle device attached to a car is an example of a mobile terminal.
These mobile terminals generally have a connection function with a wireless communication line such as a 3G (3rd Generation) line, a PHS line, and a wireless LAN (Local Area Network). With this function, the mobile terminal can communicate with other information devices and access to the WWW (World Wide Web). However, when a mobile terminal is used, a situation such as a decrease in throughput or an offline state often occurs. The reason is that the communication environment such as the radio wave intensity changes as the mobile terminal moves with the user.
When the mobile terminal is in an offline state, the user's convenience is reduced, for example, data on the Internet cannot be referred to. Therefore, in order to eliminate such a decrease in user convenience, there is an increasing number of cases where a prefetch technique for prefetching data that will be required in the future in advance to a mobile terminal is employed.
By the way, in order to effectively use the prefetch technique, it is necessary to know in advance an accurate predicted value of the throughput of the communication channel that can be used at a certain point in time. This is because the storage capacity of the cache is finite, and it is desirable to prefetch and store only necessary data in accordance with a future drop in throughput or communication disconnection.
The information providing server of Patent Document 1 can be used for communication speed prediction. The information providing server acquires information from the mobile terminal and stores measurement information including a position, a communication speed, and a communication condition. When the information providing server receives information specifying a predetermined position and communication condition from the mobile terminal, the information providing server includes one or more communication speeds measured around the predetermined position and under the communication condition from the stored measurement information. The measurement information is extracted and returned to the mobile terminal.
The wireless communication terminal device of Patent Document 2 measures and displays the electric field strength of signals from other wireless communication terminals. This wireless communication terminal apparatus calculates a weighted average of electric field strengths at time T (n) and time T (n−1) that is ΔT backward from the time.

JP 2010-177945 A JP 2001-339355 A

If the information providing server of Patent Document 1 does not store measurement information under specified communication conditions, it cannot output communication speed information. Further, the information providing server cannot output reliable communication speed information when the number of measurement information under the designated communication condition is small.
An object of this invention is to enable the performance prediction which solves the said subject.

A performance prediction apparatus according to an embodiment of the present invention includes a log storage unit that stores, for each of a plurality of types of parameters, any one of a plurality of values that can be taken by the parameters of the type, and a plurality of records including the performance values. Among the combinations of the parameter values of the plurality of types, the relaxation condition that is the combination having the highest reliability determined based on the number of the records of the log storage unit that matches the combination is determined, A relaxation condition performance value is calculated from a performance value included in the record that matches the relaxation condition, and for each of the plurality of types of parameters, a prediction condition that is one of a plurality of values that the parameter of the type can take Input a correction relaxation condition property that is a multiplication value of a coefficient determined according to the difference between the prediction condition and the relaxation condition and the relaxation condition performance value It comprises performance prediction means for outputting the value as the predicted performance values.
The performance prediction program according to an embodiment of the present invention includes a log storage unit that stores, for each of a plurality of types of parameters, any one of a plurality of values that can be taken by the type of parameters, and a plurality of records including the performance values. Access and determine the relaxation condition that is the combination with the highest reliability determined based on the number of the records of the log storage unit that matches the combination among the values of the parameters of the plurality of types, A prediction condition that calculates a relaxation condition performance value from a performance value included in the record that matches the relaxation condition, and is one of a plurality of values that can be taken by the parameter of the type for each of the plurality of types of parameters. And a correction that is a product of the coefficient determined according to the difference between the prediction condition and the relaxation condition and the relaxation condition performance value To execute a performance prediction processing to output a sum condition performance values as predicted performance value to the computer.
The performance prediction method according to an embodiment of the present invention includes a log storage unit that stores, for each of a plurality of types of parameters, one of a plurality of values that can be taken by the parameter of the type and a plurality of records including the performance value. Access and determine the relaxation condition that is the combination with the highest reliability determined based on the number of the records of the log storage unit that matches the combination among the values of the parameters of the plurality of types, A prediction condition that calculates a relaxation condition performance value from a performance value included in the record that matches the relaxation condition, and is one of a plurality of values that can be taken by the parameter of the type for each of the plurality of types of parameters. And the corrected relaxation condition, which is a product of the coefficient determined according to the difference between the prediction condition and the relaxation condition, and the relaxation condition performance value. And it outputs the performance value as predicted performance values.

According to the present invention, even if there is no communication record in a specified situation or there are only a few, it is possible to output a predicted performance value under the situation.

The structure of the performance prediction apparatus 10 of this Embodiment is shown. The structure different from FIG. 1 of the performance prediction apparatus 10 of this Embodiment is shown. The measurement log 15 memorize | stored in the log memory | storage part 11 is shown. An example of the type of parameter 17 (parameter type) and possible values of parameter 17 are shown below. An example of a weight is shown. 3 is an operation flowchart of the performance prediction apparatus 10. It is a detailed operation | movement flowchart of the performance estimation part 14 in FIG.6 S103. The performance prediction part 14 shows a mode that the position on a map is divided | segmented into the division divided into the vertical and horizontal, and is managed. A calculation method of the predicted performance value of the device at a future time, which is executed by the performance prediction unit 14, is expressed by a mathematical expression. The structure of the performance prediction system 40 including the performance prediction apparatus 10 of 2nd Embodiment is shown. The structure different from FIG. 10 of the performance prediction apparatus 10 of 2nd Embodiment is shown. The structure of the performance prediction system 40 including the performance prediction apparatus 10 of 3rd Embodiment is shown. The structure of the performance prediction apparatus 10 of 4th Embodiment is shown.

(First embodiment)
A first embodiment of the present invention will be described with reference to the drawings. The following is a description of prediction of the throughput of a device that is an example of performance. However, the performance in the present invention is not limited to the throughput of the device. The performance may be, for example, the cutting occurrence rate per unit time.
FIG. 1 shows a configuration of a performance prediction apparatus 10 according to the present embodiment. The performance prediction apparatus 10 predicts the communication performance of a certain device at a future time. Here, the device is a moving communication device, and is typically a communication device mounted on a mobile phone, a smartphone, a PDA, a ship, a vehicle, or the like. The performance prediction apparatus 10 may be built in a device to be predicted, or may exist in a location different from the device to be predicted.
The performance prediction apparatus 10 includes a log storage unit 11, a performance ratio calculation unit 12, a condition setting unit 13, and a performance prediction unit 14. The log storage unit 11 stores, as a measurement log 15, data obtained by recording the performance of a device measured at a certain time point together with the situation where the device is placed at that time point.
FIG. 3 shows the measurement log 15 stored in the log storage unit 11. FIG. 3 shows three measurement logs 15. Each measurement log 15 stores, for example, five basic data 16 of measurement start date and time, measurement period, performance measurement result, latitude and longitude, and a plurality of types of parameters 17 such as model, weather, and temperature.
FIG. 4 shows examples of parameter 17 types (parameter types) and possible values of the parameter 17. The information in FIG. 4 is stored in a memory or the like of the performance prediction apparatus 10.
Each measurement log 15 stores, for example, specific values (parameter values) of model A and model B corresponding to the parameter 17 of the parameter type “model”. The parameter value of the parameter type that was unknown at the time of performance measurement is not stored in the measurement log 15.
The performance ratio calculation unit 12 calculates a performance ratio for each of two parameter values having different parameter types from the measurement log 15 stored in the log storage unit 11. At this time, the performance ratio calculation unit 12 ignores the parameter 17 and the basic data 16 other than the one parameter type.
For example, the performance ratio calculation unit 12 calculates the ratio of the performance of the model A and the model B that are different parameter values of the parameter type “device”.
At this time, the performance ratio calculation unit 12 extracts the measurement log 15 in which the parameter value of the parameter 17 of the type “device” is the model A, and calculates the average performance of the model A from the extracted measurement log 15. When the measurement log 15 is extracted, the performance ratio calculation unit 12 ignores the values of the parameter 17 and the basic data 16 other than the parameter value of the parameter type “device”.
Similarly, the performance ratio calculation unit 12 extracts the measurement log 15 in which the parameter value of the parameter 17 of the type “equipment” is model B, calculates the average performance of the model B from the extracted measurement log 15, Calculate two ratios with the numerator and the other as the denominator.
That is, the performance ratio calculated here is the performance ratio in the situation where the values of the parameter 17 of the parameter 17 of the type other than “model” are mixed.
The performance calculated by the performance ratio calculation unit 12 is a weighted average performance. The weighted average performance is obtained by adding the performance measurement results of the measurement log 15 to be calculated to the weights that change depending on the measurement start date and taking the average.
FIG. 5 shows an example of the weight. The information in FIG. 5 is stored in a memory or the like of the performance prediction apparatus 10.
The weight is a value that increases as the time difference between the measurement start time and the current time in the measurement log 15 decreases. That is, the value becomes larger as the measurement log 15 is newer. By multiplying the weights by the measurement start date and time, the new measurement log 15 has a greater influence on the average performance value calculated than the old measurement log 15.
Incidentally, measured values of communication performance, for example, communication throughput, change in the medium to long term due to an increase in the number of base stations and an increase in the number of contracted terminals even at the same place and at the same time. Therefore, the performance ratio calculation unit 12 increases the weight of the new measurement log 15 and calculates the average value of the communication throughput. Thereby, the performance prediction apparatus 10 can estimate the communication throughput reflecting the medium- to long-term change of the communication throughput.
Here, an example of calculating the performance ratio of the model A and the model B, which is the parameter value of the parameter 17 of the type “model”, will be described. First, the performance ratio calculation unit 12 calculates the weighted average performance of the measurement log 15 having the parameter values of the model A and the model B, respectively. At this time, the performance ratio calculation unit 12 ignores the parameter 17 of a type other than the model and calculates the weighted average performance.
As a result, the influence of the parameter 17 of a type other than the model on the weighted average performance is statistically reduced. Therefore, the performance ratio calculation unit 12 can obtain the weighted average performance that does not depend on the parameter 17 of a type other than the model.
The performance ratio calculation unit 12 uses the ratio of the performance of the model A to the model B (weighted average performance of the model A / weighted average performance of the model B), and the ratio of the performance of the model B to the model A (model B). Weighted average performance / model A weighted average performance).
For example, when the weighted average throughput of model A is 1600 bps and the weighted average throughput of model B is 1000 bps, the performance ratio of model A to model B is 1600 ÷ 1000 = 1.6. That is, the model A has a throughput 1.6 times larger than the model B.
Similarly, the performance ratio calculation unit 12 calculates and stores a ratio of two performances for every two combinations of values that can be taken by the parameter 17 of the type for all parameter types. The performance ratio calculation unit 12 calculates the ratio of the performance between a sunny day and a rainy day that is a parameter value of the parameter type “weather”, and when walking (indoors) that is the parameter value of the parameter type “reception environment”. The ratio of performance while moving a car is also calculated.
When the performance ratio is known in advance, the performance ratio calculation unit 12 may not be provided.
The condition setting unit 13 determines a set of parameter values at a future time for which a predicted performance value is to be calculated as a prediction condition. For example, when predicting the throughput after 10 minutes from now, the condition setting unit 13 determines a parameter value of a situation where it is predicted that a terminal or the like to be predicted will be placed after 10 minutes.
The condition setting unit 13 acquires a prediction condition from the user via, for example, a connected input / output unit. The condition setting unit 13 may store parameter values such as the parameter type “model” and “NW type” in advance.
In addition, the condition setting unit 13 may receive time or position from the user, and may acquire part or all of the prediction conditions from a weather forecast server, a terminal specification information presentation server, or the like based on the information. Furthermore, the condition setting unit 13 may calculate some or all of the prediction conditions from certain data. For example, the condition setting unit 13 may estimate a parameter value such as the parameter type “reception environment” from the position information. The condition setting unit 13 determines parameter values that cannot be acquired or estimated as invalid values.
When the prediction condition is fixed, the condition setting unit 13 may not be provided.
The performance prediction unit 14 calculates a predicted performance value based on the position input by the user or the like and the prediction condition determined by the condition setting unit 13. The performance prediction unit 14 calculates a predicted performance value using the measurement log 15 stored in the log storage unit 11 and the performance ratio calculated by the performance ratio calculation unit 12.
In addition, when the place used as prediction object is fixed, the performance prediction part 14 does not need to input a position from the outside.
FIG. 2 shows a configuration different from that of FIG. 1 of the performance prediction apparatus 10 of the present embodiment. The performance prediction apparatus 10 may be a computer that includes a CPU 21 (Central Processing Unit) and a storage device 22 and executes a performance prediction program 23 stored in the storage device 22. In this case, the CPU 21 that executes the performance prediction program 23 functions as the performance ratio calculation unit 12, the condition setting unit 13, and the performance prediction unit 14. For example, the storage device 22 functions as the log storage unit 11.
Next, an outline of the operation of the present embodiment will be described with reference to FIG. FIG. 6 is an operation flowchart of the performance prediction apparatus 10.
The performance ratio calculation unit 12 calculates a performance ratio from the measurement log 15 stored in the log storage unit 11 (step S101). As described above, the performance ratio calculation unit 12 calculates and stores a ratio of two performances for every two combinations of values that the parameter 17 of the type can take for all parameter types. The performance ratio calculated here is a weighted average performance ratio.
The condition setting unit 13 sets a prediction condition at a future time (step S102).
The performance prediction unit 14 calculates a predicted performance value of the device under the prediction condition (step S103).
An example of a method for calculating the predicted performance value will be described in detail with reference to FIG. FIG. 7 is a detailed operation flowchart of the performance prediction unit 14 in step S103 of FIG. The performance predicting unit 14 manages the position on the map by dividing it into sections that are divided vertically and horizontally as shown in FIG.
The performance prediction unit 14 first reads the latitude and longitude of the prediction conditions determined by the condition setting unit 13. Next, it is determined to which section on the map the point represented by the read latitude and longitude belongs (step S201).
After the partition determination, the performance prediction unit 14 lists all the conditions combining the parameter values of the parameter types that are not invalid values as the relaxation conditions from the prediction conditions determined by the condition setting unit 13 (step S202).
This operation will be described based on the following specific example.
1) The defined parameter types are “device”, “weather”, and “NW congestion degree”.
2) Possible values of the parameter 17 of the type “device” are {A, B, C}.
3) Possible values of the parameter 17 of the type “weather” are {clear, cloudy, rain}.
4) Possible values of the parameter 17 of the type “NW congestion degree” are {low, medium, high}.
5) The prediction condition is (A, cloudiness, invalid value).
In this example, the performance prediction unit 14 lists the following combinations as relaxation conditions.
(A, clear), (A, cloudy), (A, rain)
(B, clear), (B, cloudy), (B, rain)
(C, clear), (C, cloudy), (C, rain)
Next, the performance prediction unit 14 calculates the prediction reliability for each listed relaxation condition (step S203).
The prediction reliability is the total weight of the measurement log 15 included in the section that satisfies the relaxation condition and is determined in step S201. As shown in FIG. 5, the weight is a value determined by the measurement start time. That is, the predicted reliability increases as the number of measurement logs 15 near the current time increases.
For example, the performance prediction unit 14 calculates the prediction reliability of the relaxation condition (B, clear) as follows. First, the performance prediction unit 14 is included in the section determined in step S201, and the parameter value of the parameter 17 of the type “model” is B and the parameter value of the parameter 17 of the type “weather” is clear. The log 15 is extracted from the log storage unit 11. At this time, the performance prediction unit 14 ignores other types of parameter values and values of the basic data 16 other than “latitude” and “longitude”.
Next, the performance prediction unit 14 determines the weight of each measurement log 15 with reference to the definition information as shown in FIG. Calculate as
The performance prediction unit 14 selects one relaxation condition having the highest predicted reliability from the listed relaxation conditions, and sets the predicted reliability of the relaxed condition as the predicted reliability in the relaxed condition (step S204).
The magnitude of the prediction reliability represents a tendency that the number of measurement logs 15 matching the condition is large and the freshness is high. Therefore, the performance prediction unit 14 can increase the reliability of the prediction by predicting the performance with respect to the relaxation condition having a high prediction reliability.
Next, the performance prediction unit 14 calculates the prediction reliability of the prediction condition by the same method as in step S203 (step S205).
The performance prediction unit 14 calculates the prediction reliability of the prediction condition using the measurement log 15 that satisfies the prediction condition and is included in the section determined in step S201.
After step S205, the performance prediction unit 14 calculates a relaxation reliability ratio α and a prediction reliability ratio β that are ratios between the prediction reliability under the relaxation condition and the prediction reliability under the prediction condition (step S206).
The performance prediction unit 14 calculates the relaxation reliability ratio α and the prediction reliability ratio β as follows.
1) Relaxation reliability ratio α = prediction reliability under relaxation conditions ÷ (prediction reliability under relaxation conditions + prediction reliability under prediction conditions)
2) Prediction reliability ratio β = prediction reliability under prediction conditions / (prediction reliability under relaxation conditions + prediction reliability under prediction conditions)
After step S206, the performance prediction unit 14 compares the relaxation condition selected in step S204 with the prediction condition, and extracts a combination of parameter values having different values. The invalid value parameter 17 is ignored. Then, the performance prediction unit 14 selects the ratio of the performance of “parameter value specified by the prediction condition” to “parameter value specified by the relaxation condition” for each combination, and calculates the ratio (step S207). ).
For example, when the relaxation condition selected in step S204 is (C, clear) and the prediction condition is (A, cloudy, invalid value), the performance prediction unit 14 determines (C, A) and (clear, cloudy). A combination is extracted, the following two ratios are selected, and both are multiplied to calculate the ratio.
1) Performance ratio of model A to C (A / C)
2) Ratio of performance to cloudy weather (cloudy / sunny)
Next, the performance prediction unit 14 obtains the weighted average performance A under the relaxation condition selected in Step 204 (Step S208).
The performance prediction unit 14 calculates using the measurement log 15 included in the section determined in step S201 while satisfying the relaxation condition. The weighted average performance A under the relaxation condition is an average value obtained by multiplying the performance measurement result of the measurement log 15 satisfying the relaxation condition by the weight of each measurement log 15. The average value is a value obtained by dividing the total value of values obtained by multiplying the performance measurement result of the measurement log 15 by the weight of each measurement log 15 by the total value of the weights.
For example, the performance prediction unit 14 calculates the weighted average performance A of the relaxation condition (C, clear) selected in step S204 as follows. First, the performance prediction unit 14 is included in the section determined in step S201, and the parameter value of the parameter 17 of the type “model” is C and the parameter value of the parameter 17 of the type “weather” is clear. The log 15 is extracted from the log storage unit 11. At this time, the performance prediction unit 14 ignores other types of parameter values and values of the basic data 16 other than “latitude” and “longitude”.
Next, the performance prediction unit 14 refers to the definition information as shown in FIG. 5 and the “measurement start date and time” as the basic data 16, determines the weight for each of the extracted measurement logs 15, and determines the basic data 16. Multiplication with “performance measurement result” is obtained to obtain a multiplication value. The performance prediction unit 14 obtains the sum of the multiplication values and the sum of the weights for the extracted measurement log 15 and divides the former by the latter to obtain the average performance A.
In addition, the performance prediction unit 14 calculates the weighted average performance B under the prediction condition by the same method as in step S208 (step S209).
Finally, the performance prediction unit 14 multiplies the weighted average performance A under the relaxation condition by the relaxation reliability ratio α and the performance ratio ratio, and the weighted average performance B under the prediction condition by the prediction reliability ratio β. Are added together and output as a predicted performance value of the device (step S210).
FIG. 9 shows, by mathematical formulas, a method for calculating a predicted performance value of a device at a future time, which is executed by the performance prediction unit 14.
Next, the effect of this embodiment will be described.
The performance prediction apparatus 10 of the present embodiment can output the predicted performance value of the device even when there are no measurement logs 15 that satisfy the prediction condition or there are only a few. The reason is that the performance prediction unit 14 calculates a predicted performance value based on the measurement log 15 that satisfies the relaxation condition.
The performance prediction apparatus 10 of the present embodiment can calculate a highly reliable predicted performance value. The reason is that the performance prediction device 10 calculates a predicted performance value based on a performance ratio obtained by neutralizing various factors and a relaxation condition with high prediction reliability. This is because the performance prediction device 10 calculates the predicted performance value so as to strongly reflect the value of the new measurement log 15.
That is, the performance prediction apparatus 10 of the present embodiment can accurately predict the future performance of the device in consideration of the transition of measured performance values over the medium to long term and the situation of the user at that time.
(Second Embodiment) Next, a second embodiment of the present invention will be described with reference to the drawings.
FIG. 10 shows a configuration of a performance prediction system 40 that includes the performance prediction apparatus 10 of the present embodiment. The performance prediction system 40 is configured by connecting the performance prediction device 10 and an external device 90 via a network 41.
The performance prediction apparatus 10 is added with a result storage unit 19 as compared with the first embodiment. The performance prediction apparatus 10 according to the present embodiment also includes a log storage unit 11, a performance ratio calculation unit 12, a condition setting unit 13, and a performance prediction unit 14. However, the log storage unit 11 does not store the initial measurement log 15. In addition, the performance prediction unit 14 can communicate with the outside.
The external device 90 includes an original log storage unit 95 and a prediction condition storage unit 96. The original log storage unit 95 stores the measurement log 15 shown in FIG. The prediction condition storage unit 96 stores prediction conditions.
In the present embodiment, the result storage unit 19 acquires the measurement log 15 from the original log storage unit 95 of the external device 90 and stores it in the log storage unit 11 prior to predicting the communication performance of the device. Further, the condition setting unit 13 acquires a prediction condition from the prediction condition storage unit 96 of the external device 90. Further, the performance prediction unit 14 outputs the calculated predicted performance value to the external device 90. Other points are the same as in the first embodiment.
FIG. 11 shows a configuration different from that of FIG. 10 of the performance prediction apparatus 10 of the present embodiment. The performance prediction apparatus 10 may be a computer that includes a CPU 21, a storage device 22, and a communication device 24 and executes a performance prediction program 23 stored in the storage device 22.
The performance prediction apparatus 10 according to the present embodiment is connected to an external device 90 via the communication device 24. The external device 90 is a computer or the like that includes a storage device 91.
In this case, the CPU 21 that executes the performance prediction program 23 functions as the performance ratio calculation unit 12, the condition setting unit 13, the performance prediction unit 14, and the result storage unit 19. In addition, the storage device 22 functions as the log storage unit 11.
Further, the storage device 91 of the external device 90 functions as the original log storage unit 95 and the prediction condition storage unit 96.
Next, the effect of this embodiment will be described.
The performance prediction apparatus 10 of this embodiment calculates the performance value of the device at the future time of the external device 90, and the external device 90 can utilize the predicted performance value of the device at the future time. The reason is that the performance prediction device 10 acquires the measurement log 15 and the prediction condition from the external device 90 and outputs the predicted performance value to the external device 90.
(Third Embodiment) Next, a third embodiment of the present invention will be described with reference to the drawings.
FIG. 12 shows a configuration of a performance prediction system 40 that includes the performance prediction apparatus 10 of the present embodiment. The performance prediction system 40 is configured by connecting a mobile device 30, an external device 90 and an information providing device 80 via a network 41. The network 41 includes a mobile phone communication network, a wired and wireless Internet, and the like.
The mobile device 30 is a mobile communication device including the performance prediction device 10 and the communication / processing device 31 according to the present invention, such as a mobile phone, a smartphone, a communication device mounted on a ship, a vehicle, or the like. The communication / processing device 31 performs the communication and information processing functions of the mobile device 30. When the mobile device 30 is, for example, a smartphone, the communication / processing device 31 performs all communication and information processing functions as a smartphone, such as calling, Web server access, and e-mail. In addition, the communication / processing device 31 acquires an instruction from the user or presents information to the user via the connected input / output device 33.
The performance prediction apparatus 10 in this embodiment predicts the communication performance of the communication / processing device 31 and outputs the predicted performance value to the communication / processing device 31. Based on the acquired predicted performance value, the communication / processing device 31 prefetches data required in the future from the information providing device 80 or the like into the cache 32 or the like. The communication / processing device 31 performs prefetch using a known technique.
The performance prediction device 10 can communicate with the external device 90, the information providing device 80, and the like via the communication / processing device 31. The performance prediction apparatus 10 according to the present embodiment includes a performance ratio calculation unit 12, a condition setting unit 13, and a performance prediction unit 14, but does not include the log storage unit 11.
The log storage unit 11 is provided in the external device 90. The external device 90 is, for example, a collection device for the measurement log 15 operated by a mobile phone company or a communication company.
The performance ratio calculation unit 12, the condition setting unit 13, and the performance prediction unit 14 of the performance prediction device 10 of the present exemplary embodiment access the log storage unit 11 of the external device 90 via the communication / processing device 31.
The information providing device 80 is a Web server or the like, and the condition setting unit 13 accesses the information providing device 80 via the communication / processing device 31.
The condition setting unit 13 acquires necessary data from the information providing device 80 and the communication / processing device 31 when determining the prediction condition. For example, the condition setting unit 13 acquires parameter values of the parameter type “model” and “NW type” from the communication / processing device 31. In addition, the condition setting unit 13 acquires parameter values such as parameter type “weather” and “temperature” corresponding to the latitude and longitude input from the user from the weather information providing server. The condition setting unit 13 acquires parameter values such as the parameter type “congestion degree” corresponding to the latitude and longitude input from the user from the area information providing server.
In other respects, the performance prediction apparatus 10 of this embodiment is the same as that of the first embodiment.
Next, the effect of this embodiment will be described.
The performance prediction apparatus 10 of the present embodiment makes it easy to use the measurement log 15 and determine measurement conditions. The reason is that data necessary for prediction is acquired from the external device 90 operated by a communication company or the like and the information providing device 80 operated by a third party.
(Fourth Embodiment) Next, a fourth embodiment of the present invention will be described with reference to the drawings. FIG. 13 shows the configuration of the performance prediction apparatus 10 of the present embodiment.
The performance prediction apparatus 10 includes, for each of a plurality of types of parameters 17, a log storage unit 11 that stores any one of a plurality of values that can be taken by the type of parameter 17 and a plurality of measurement logs 15 that include performance measurement results. A connected performance prediction unit 14 is provided.
The performance prediction unit 14 selects a combination (relaxation condition) having the highest reliability determined based on the number of measurement logs 15 in the log storage unit 11 that matches the combination among the values of the parameters 17 of a plurality of types. The relaxation condition performance value is calculated from the performance measurement result included in the measurement log 15 that is determined and matches the relaxation condition.
Further, the performance prediction unit 14 inputs any one of a plurality of values (prediction conditions) that can be taken by the parameter 17 of each type for each of the plurality of types of parameters 17, and sets the prediction condition and the relaxation condition. A multiplication value of a coefficient determined according to the difference and the relaxation condition performance value is output as a predicted performance value.
The performance prediction apparatus 10 of the present embodiment can output a predicted performance value under the conditions even when there are no measurement logs 15 under the prediction conditions or there are only a few. The reason is that the performance prediction unit 14 calculates a predicted performance value based on the measurement log 15 that satisfies the relaxation condition.
While the present invention has been described with reference to the embodiments (and examples), the present invention is not limited to the above embodiments (and examples). Various changes that can be understood by those skilled in the art can be made to the configuration and details of the present invention within the scope of the present invention.
This application claims the priority on the basis of Japanese application Japanese Patent Application No. 2010-279018 for which it applied on December 15, 2010, and takes in those the indications of all here.

DESCRIPTION OF SYMBOLS 10 Performance prediction apparatus 11 Log memory | storage part 12 Performance ratio calculation part 13 Condition setting part 14 Performance prediction part 15 Measurement log 16 Basic data 17 Parameter 19 Result storage part 21 CPU
22 storage device 23 performance prediction program 24 communication device 30 mobile device 31 communication / processing device 32 cache 33 input / output device 40 performance prediction system 41 network 80 information providing device 90 external device 91 storage device 95 original log storage unit 96 prediction condition storage unit

Claims

For each of a plurality of types of parameters, any one of a plurality of values that can be taken by the parameters of that type, and a log storage unit that stores a plurality of records including performance values, and a combination of the values of the plurality of types of parameters Among them, a relaxation condition that is a combination having the highest reliability determined based on the number of the records of the log storage unit that matches the combination is determined, and the performance value included in the record that matches the relaxation condition Calculate the relaxation condition performance value from
For each of the plurality of types of parameters, a prediction condition that is one of a plurality of values that can be taken by the type of parameter is input, and a coefficient determined according to a difference between the prediction condition and the relaxation condition And a performance prediction means for outputting a corrected relaxation condition performance value that is a product of the relaxation condition performance value as a predicted performance value,
A performance prediction apparatus comprising:
The performance prediction means extracts different parameter values of the same type from the prediction conditions and the relaxation conditions, and calculates performance values calculated from all the records including the parameter values extracted from the prediction conditions, The performance prediction apparatus according to claim 1, wherein a ratio divided by a performance value calculated from all the records including a parameter value extracted from a relaxation condition is used as the coefficient.
The record contains a date and time;
The performance prediction apparatus according to claim 1, wherein the performance prediction unit determines the reliability higher as the number of records near the current time increases.
The performance prediction means calculates the reliability A that is the reliability from the record that matches the relaxation condition, and the reliability B that is the reliability from the record that matches the prediction condition, and matches the prediction condition The predicted condition performance value is calculated from the performance value of the record to be calculated, and the calculated value by the formula “α × correction relaxation condition performance value + β × predicted condition performance value” (where α + β = 1, α: β = A: B) Is output as the predicted performance value. The performance prediction apparatus according to claim 1.
Condition setting for determining a model as the prediction condition and further determining a parameter value of one or more types of parameters of communication method, weather, temperature, humidity, communication environment, and network congestion based on the input position The performance prediction apparatus according to claim 1, comprising means.
An external device comprising the log storage means, and
6. The performance prediction device according to claim 1, and inputs / outputs the log storage device of the external device connected via a network according to an instruction from the performance prediction device, and the performance prediction device A performance prediction system including a mobile terminal including a communication / processing device that determines data to be stored in a cache based on the predicted performance value acquired from
For each of a plurality of types of parameters, access one of a plurality of values that can be taken by the parameter of the type and a log storage unit that stores a plurality of records including performance values, and Among the combinations, the relaxation condition that is the combination having the highest reliability determined based on the number of the records of the log storage unit that matches the combination is determined, and the performance included in the record that matches the relaxation condition Calculate the relaxation condition performance value from the value,
For each of the plurality of types of parameters, a prediction condition that is one of a plurality of values that can be taken by the type of parameter is input, and a coefficient determined according to a difference between the prediction condition and the relaxation condition A performance prediction program that causes a computer to execute a performance prediction process that outputs a corrected relaxation condition performance value that is a product of the relaxation condition performance value and the relaxation condition performance value as a predicted performance value.
Different parameter values of the same type are extracted from the prediction condition and the relaxation condition, and performance values calculated from all the records including the parameter value extracted from the prediction condition are extracted from the relaxation condition 8. The performance prediction program according to claim 7, which causes a computer to execute the performance prediction process using a ratio obtained by dividing the performance value calculated from all the records including a parameter value as the coefficient.
For each of a plurality of types of parameters, access one of a plurality of values that can be taken by the parameter of the type and a log storage unit that stores a plurality of records including performance values, and During the combination, a relaxation condition that is a combination having the highest reliability determined based on the number of the records of the log storage unit that matches the combination is determined,
Calculate the relaxation condition performance value from the performance value included in the record that matches the relaxation condition,
For each of the plurality of types of parameters, a prediction condition that is one of a plurality of values that can be taken by the type of parameter is input, and a coefficient determined according to a difference between the prediction condition and the relaxation condition A performance prediction method for outputting a corrected relaxation condition performance value that is a product of the relaxation condition performance value and the relaxation condition performance value as a predicted performance value.
Different parameter values of the same type are extracted from the prediction condition and the relaxation condition, and performance values calculated from all the records including the parameter value extracted from the prediction condition are extracted from the relaxation condition The performance prediction method according to claim 9, wherein a ratio divided by a performance value calculated from all the records including a parameter value is used as the coefficient.