CN103475790A - Intelligent mobile terminal power consumption management method - Google Patents
Intelligent mobile terminal power consumption management method Download PDFInfo
- Publication number
- CN103475790A CN103475790A CN2013104033442A CN201310403344A CN103475790A CN 103475790 A CN103475790 A CN 103475790A CN 2013104033442 A CN2013104033442 A CN 2013104033442A CN 201310403344 A CN201310403344 A CN 201310403344A CN 103475790 A CN103475790 A CN 103475790A
- Authority
- CN
- China
- Prior art keywords
- frequency
- application
- cpu
- terminal
- mobile terminal
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Mobile Radio Communication Systems (AREA)
- Telephone Function (AREA)
Abstract
The invention provides an intelligent mobile terminal power consumption management method. The method comprises the steps that 11), when an application is operated in a terminal, the saturation frequency of the application is obtained, and 12) dynamic adjustment is carried out on the frequency of a CPU based on the ONDEMAND algorithm, wherein for each application, the saturation frequency obtained in the step 11) is used as the highest frequency of the ONDEMAND algorithm. The saturation frequency of the application is obtained through prediction of a prediction model based on the active character of the terminal when the application is operated. By means of the method, the problem that user experience can not be correspondingly improved when excessive computing resources are distributed is avoided, the energy conservation optimization effect is achieved, and power consumption can be precisely lowered according to user experience.
Description
Technical field
The present invention relates to data processing technique and terminal field of energy-saving technology, specifically, the present invention relates to a kind of intelligent mobile terminal power consumption management method.
Background technology
In recent years, the smart mobile phone of diverse in function more and more is subject to market and welcomes, but meanwhile, and the too short problem of smart mobile phone stand-by time is also more and more outstanding, to the user, causes inconvenience.At present, the development speed of battery technology has been difficult to catch up with the user for the craving for of longer stand-by time, and therefore by more optimal smart mobile phone power managed strategy, comes prolongs standby time to become one of more and more important research direction.
CPU is the core component of smart mobile phone, and it calculates support for mobile phone application of all shapes and colors provides, and such as game, browser, tool software etc., these are applied spent electric weight and have surpassed the electric weight that traditional signal post expends.In order to meet the demand of these application, cpu performance constantly promotes, and its power consumption also grows with each passing day, and effectively reduces the CPU power consumption to reducing smart mobile phone overall power important in inhibiting.
At present, two of the CPU power managed algorithms commonly used are Linux PERFORMANCE algorithm and Linux POWERSAVING algorithm.It is preferential that Linux PERFORMANCE algorithm is intended to performance, and it can be located at highest level to cpu frequency always, and Linux POWERSAVING algorithm is energy conservation priority, and it is located at floor level to frequency always.These two kinds of algorithms are not all considered dynamically regulating frequency, cause or have wasted resource, or having damaged user's experience.
Linux ONDEMAND algorithm is to carry out a kind of CPU power managed algorithm of regulating frequency according to real-time cpu busy percentage.In typical Linux ONDEMAND algorithm, when cpu busy percentage surpasses a upper limit, frequency can be set in highest frequency; When utilance, lower than a lower limit, frequency will be set in the frequency that can keep than the utilance of the upper limit low 10%.With respect to Linux PERFORMANCE algorithm, Linux ONDEMAND algorithm can effectively reduce the power consumption of CPU.Yet this power managed algorithm does not directly consider that the user experiences, and increasing practice shows, in a lot of situations, Linux ONDEMAND algorithm easily distributes too much computational resource but can't correspondingly improve user's experience, therefore, on the basis of Linux ONDEMAND algorithm, also there is the space that further reduces power consumption.
In sum, current in the urgent need under a kind of prerequisite that can experience the assurance user, accurately reduce the solution of intelligent mobile terminal power consumption.
Summary of the invention
The purpose of this invention is to provide a kind of can experience under the prerequisite do not reduced the assurance user, accurately reduce the solution of intelligent mobile terminal power consumption.
For achieving the above object, the invention provides a kind of model training method of predicted application saturation frequency, comprise the following steps:
1) set up forecast model, the terminal operating characteristic vector that when this model will be applied operation, the active characteristics of terminal forms is as input, using the Candidate Frequency lid as output; When wherein, described application moves, the active characteristics of terminal comprises: the busy time of CPU accounts for ratio or the percentage of running time; Ratio or the percentage of the running time of CPU on each frequency; And CPU continues average and the variance of busy time span;
2) train described forecast model based on sample set, the active characteristics of terminal while having gathered the application operation in this sample set, and corresponding user experiences and cpu frequency.
When wherein, described application moves, the active characteristics of terminal also comprises: touch screen causes average and the variance of system break number of times; The average of touch screen time span; The average of interval time and variance between touch screen; And the gravity average of component and variance on the y direction of principal axis in the terminal local Coordinate System.
Wherein, described forecast model is neural network model.
Wherein, described step 2), in, when the ratio of the average user experience value of a large number of users and maximum user experience value surpasses predetermined threshold, regard the frequency that reaches capacity as.
The present invention also provides a kind of method of predicted application saturation frequency, comprises the following steps:
Start application, gather the terminal operating characteristic vector of mobile terminal in this application running, the forecast model that collected terminal operating characteristic vector input is trained by the described method of any one in claim 1～4, the frequency that this forecast model is exported is this and applies corresponding saturation frequency.
The present invention also provides a kind of intelligent mobile terminal power consumption management method, comprises the following steps:
11) utilize the method for aforementioned predicted application saturation frequency to obtain the saturation frequency of applying;
12) based on the ONDEMAND algorithm, the frequency of CPU is carried out to dynamic adjustments, wherein the resulting saturation frequency of step 11) is as the highest frequency in the ONDEMAND algorithm.
Wherein, in described step 11), the model of predicted application saturation frequency completes training at server end, then, server sends the parameter of described model to mobile terminal, at mobile terminal, sets up the saturation frequency that history lists records each application, when each application moves for the first time, obtain corresponding saturation frequency and be recorded in described history lists by the end activity characteristic vector prediction collected, each this application of startup, check that history lists obtains corresponding saturation frequency later.
Wherein, in described step 11), for the unseen newtype application of described forecast model, mobile terminal is accepted frequency and the user experience value that the user manually arranges, and the two and corresponding terminal operating characteristic vector are sent to server, described server adds the described sample set for model training by it.
Compared with prior art, the present invention has following technique effect:
1, the present invention can prevent from distributing too much computational resource but can't correspondingly improve user's experience, thereby reaches the effect of energy saving optimizing.
2, the present invention can reach the effect of experiencing accurate reduction power consumption according to the user.
The accompanying drawing explanation
Fig. 1 shows the distribution situation of user's experience of lower six different application of different CPU frequency;
Fig. 2 shows the flow chart of one embodiment of the invention;
Fig. 3 shows and respectively is applied in the contrast of running time on each frequency of CPU under traditional ONDEMAND algorithm and the ONDEMAND algorithm based on one embodiment of the invention;
Fig. 4 shows in experiment the time dependent curve of cpu busy percentage when different application is moved; Wherein the X coordinate means tactic time percentage sheet from small to large;
Fig. 5 shows in experiment the time dependent curve of the caused system break number of touch screen when different application is moved;
Fig. 6 shows 27 motion feature scatter diagrams that are applied in while moving in experiment;
Fig. 7 shows the schematic diagram of mobile phone coordinate system;
Fig. 8 shows the precision that the embodiment that uses the different characteristic vector in the present invention predicts saturation frequency;
CPU power consumption contrast when Fig. 9 shows difference under algorithms of different and should move;
Overall Power Consumption contrast when Figure 10 shows difference under algorithms of different and should move.
Embodiment
For the ease of understanding, the research that paper the present inventor does for the relation between cpu frequency and user's experience.The inventor has done an investigation that relates to 20 users and 6 application (comprising: Talking Tom, Snow Pro, Storm, Fruit Ninja, UC Browser and QQ), in order to the characteristics of digging user, application.In investigation, when the user uses application, the frequency of random adjustment CPU, and inquiry user's experience.The user experiences and is divided into three grades: good, in, poor.So just, obtain the user of user's experience on each CPU frequency and experience distribution.Fig. 1 shows the distribution situation of user's experience of lower six different application of different CPU frequency.With reference to figure 1, wherein each bar shaped means the corresponding number of users of this user's experience level.For some application, Talking Tom for example, Snow Pro, Storm, along with the raising of cpu frequency, comment the number of users of " good " to increase, and comments the number of users of " poor " to reduce.But, also there are other application, such as Fruit Ninja and UC Browser, when cpu frequency brings up to 800 from 600, the user does not experience and significantly promotes.And, for QQ, when cpu frequency brings up to 600 from 300, the user experiences also and does not significantly promote.Can find out, cpu frequency and user experience not always positive correlation, and there is a saturation frequency in some application, and cpu frequency is during higher than saturation frequency, and the user experiences almost and do not promote.By screening the saturation frequency of different application, saturation frequency is set to the frequency lid of this application, make the dynamic adjustments scope of CUP frequency not exceed this frequency lid, can prevent from distributing too much computational resource but can't correspondingly improve the phenomenon appearance that the user experiences, thereby reach the effect of energy saving optimizing.Further, in order to determine exactly the frequency lid, a kind of scheme of the predict frequency lid based on neural network model has been proposed in the following embodiments, a series of active characteristics when it moves by being applied in mobile terminal, predict the frequency lid of each user for each application, thereby reach the effect of experiencing accurate reduction power consumption according to the user.
Below in conjunction with embodiment, the present invention will be further described.
A kind of method of saturation frequency of predicted application is provided according to one embodiment of present invention.
The saturation frequency of application is experienced relevant to the user, can collect the sample of some by experiment, in sample, recording user is experienced the relation changed with the CPU frequency, and a series of active characteristics while being applied in accordingly in mobile terminal operation, for convenience of describing, the vector hereinafter these described a series of active characteristics formed is called the terminal operating characteristic vector.Sample based on collected, the neural network training model, in this neural network model, the neuron using each active characteristics in the terminal operating characteristic vector as the neural network model input layer, using candidate's frequency lid respectively as a neuron of neural network model output layer.Utilize above-mentioned trained neural network model, i.e. the saturation frequency of measurable each application also arranges corresponding frequency lid.
In this step, the user is experienced to UX and quantizes to be defined as:
Wherein, W
1, W
2, W
3mean the scoring that the user experiences, be respectively 1,2,3, the evaluation of representative " poor, in, good ".N
1, N
2, N
3expression is to the number of corresponding scoring.Easily find out, under this definition, the maximum that the user experiences UX is 3.In the present embodiment, UX is thought to the user who reaches the highest experiences, and now regards as and can't significantly promote the situation that the user experiences again in the situation of (being that the UX maximum is more than 90%) more than 2.7.More generally, can first set a threshold value (for example 90%), when the ratio of the average user experience value of a large number of users and maximum user experience value surpasses this threshold value, can regard as and can't significantly promote again the situation that the user experiences, regard the frequency that reached capacity as.
Prior art is only adjusted cpu frequency by cpu busy percentage, and unlike the prior art, the active characteristics be applied in mobile terminal while moving that it utilizes comprises a series of features relevant to CPU to the present embodiment.These features relevant to CPU comprise:
The busy time of a, CPU accounts for the ratio (R of running time
bt),
Percentage (the P of b, CPU running time on each candidate's frequency (be Candidate Frequency lid frequency)
f),
Average and the variance (μ (Lcb), δ (Lcb)) of c, lasting busy time span.
Particularly, R
btcpu busy percentage at running time of 100% and the ratio of total run time.R
btthis application of larger explanation is just larger to the demand of amount of calculation.Such as, Fig. 4 shows in experiment the time dependent curve of cpu busy percentage when different application is moved; Wherein the X coordinate means tactic time percentage sheet from small to large.In Fig. 4, the R of Storm
btup to 65%, and its corresponding frequency lid is 800MHz.On the contrary, the R of QQ and UC
btvalue only has 30%, and finally their frequency lid is lower 600MHz.
But, only use R
btinadequate, because R
btdo not consider cpu frequency.Such as different application may have close R
bt, but operate in different cpu frequencies.Like this, affirm more consumption calculations resource in that application of upper frequency.The present embodiment is introduced P
fthis feature.Fig. 3 shows and respectively is applied in the contrast of running time on each frequency of CPU under traditional ONDEMAND algorithm and the ONDEMAND algorithm based on one embodiment of the invention, and as shown in Figure 3, in ONDEMAND algorithm (the w/o cap in figure), application has P separately
fvalue.
The continuous busy time (CBsession) refers to that cpu busy percentage keeps for 100% time continuously.Usually, a long-term Cbsession is caused by larger load, brings higher μ (Lcb).On the contrary, less load can only cause shorter Cbsession and lower μ (Lcb).An application that has higher μ (Lcb) should operate on higher cpu frequency, its real time reaction of guarantee like this.The present embodiment is further used δ (Lcb) to be classified to different application.Fig. 4 shows 6 average and variances that are applied in the continuous busy time span of CPU while moving in experiment.As shown in Figure 4, MPC has higher δ (Lcb), and this means that it needs higher frequency lid, namely 800MHz.By contrast, the fruit person of bearing has lower μ (Lcb) and δ (Lcb), illustrates that it only needs lower frequency lid.
The feature that above-mentioned CPU is relevant can finely be portrayed the demand of application to cpu performance in most situation, has higher precision of prediction.Yet for some special applications, for example application of game class and video class, the relevant feature of CPU is for obtaining respectively 70% and 50% precision of prediction, as shown in Figure 8.Given this, in a preferred embodiment, can also increase other two category features, touch screen feature and motion feature, classified to application.
Dissimilar application has own unique touch screen feature usually, in the system break information that this will cause at touch screen, embodies to some extent.Fig. 5 shows in experiment the time dependent curve of the caused system break number of touch screen when different application is moved.Can see that UC Browser has touch screen more frequently than Gofishing and Dragon, and Storm almost do not have touch screen.In addition, the time of the touch screen of different application also is not quite similar.Such as, such as some application need to drag the page more, and dragging the page, it is long that its touch screen time is generally touched a page widgets than point.In addition, the time interval between twice touch screen of different application changes also very greatly.Consider these situations, adopted the touch screen feature of three aspects in the preferred embodiment, comprise: touch screen causes average and the variance (μ (Nint) of system break number of times (Nint), δ (Nint)), sample frequency is 5Hz, the average of touch screen time span (μ (Ltouch)), the average of interval time between touch screen (Linterval) and variance (μ (Linterval), δ (Linterval)).
Below introduce again the feature of mobile phone motion aspect.The feature of these motion aspects be from the acquisition of information of the gravity sensor of embedded in mobile phone to.Fig. 6 shows 27 motion feature scatter diagrams that are applied in while moving in experiment.Wherein, the x axle means the average of gravity component on the mobile phone vertical direction, with μ (y), means.The y axle is corresponding standard deviation, i.e. δ (y).For clarity sake, application is noted as game class, browser class and video class.In figure, plane has been divided into four zones (Domain), and Domain1 has comprised the horizontal machine of shaking hands of needs, and the application of frequently rocking mobile phone, such as the game of the class of competing for speed.Domain2 is mainly that some need the horizontal machine of shaking hands, but the application of almost not rocking.Some perpendicular machines of shaking hands, do not need the application of rocking, and such as browser, generally is distributed in Domain3.In addition, do not apply and appear in Domain4 in experiment.
Fig. 8 has shown the precision of prediction that the feature of use different sets is brought.Only use cpu character can make the application of browser class that precision of prediction is preferably arranged, but can be only that precision of prediction reaches 70% and 50% for the application of game class and video class.When using above-mentioned all features, can bring up to 91% and 85% to corresponding pre-survey precision.Use result to confirm aforementioned analysis.
Based on aforementioned analysis, for one embodiment of the present of invention, provide a kind of method of training the saturation frequency forecast model, comprise step 101～103:
Step 101: set up a three-layer neural network model, its input layer has 12 neurons, and hidden layer has 18 neurons, and output layer has 4 neurons.
Wherein, the output layer of three-layer neural network model is the vector of 4 elements, and each element represents respectively 300MHz, 600MHz, 800MHz and HIGHER.The meaning of HIGHER is, apply too high to the demand of calculated performance, to such an extent as to 800MHz can not meet its demand.
12 neurons of the input layer of three-layer neural network model are exactly 12 features introducing previously, and these 12 features form the terminal operating characteristic vector.12 features are respectively:
1, the busy time of CPU accounts for ratio or the percentage (busy referring to, the utilance of CPU is 100%) of running time;
2, ratio or the percentage of the running time of CPU on each frequency, in the present embodiment, by CPU ratio or the percentage of running time on 300MHz, and the ratio of CPU running time on 600MHz or percentage is respectively as two characteristic values, 300MHz, 600MHz are less two of numerical value in the Candidate Frequency lid of final output;
3, continue average and the variance (the lasting busy time refers to, the lasting maintenance 100% of the utilance of CPU) of busy time span, wherein average and variance are separately as a characteristic value;
4, touch screen causes average and the variance of system break number of times, and wherein average and variance are separately as a characteristic value;
5, the average of touch screen time span, press the time of unclamping from finger;
6, the average of interval time and variance between touch screen, wherein average and variance are separately as a characteristic value;
7, average and the variance of gravity upper component of y direction of principal axis (longitudinal direction that refers to mobile phone, Fig. 7 shows the schematic diagram of mobile phone coordinate system) in the mobile phone local Coordinate System, wherein average and variance are separately as a characteristic value.
In above-mentioned 12 features, 1 to 3 is the CPU correlated characteristic, and 4 to 5 is the touch screen feature, the 7 physical motion features that are terminal.4 to 7 can be referred to as non-CPU correlated characteristic.With reference in above to the analysis of each category feature, can find out, selecting whole 12 features is only a preferred version as the input layer of forecast model, in other embodiment, also can select the input layer of 12 parts in feature as forecast model, for example can only use the CPU correlated characteristic and not use the physical motion feature of touch screen feature and terminal, this forecast model can predict comparatively exactly the saturation frequency of the great majority application (for example application of browser class) of non-video class and non-game class.
Step 102: the collecting sample data are as training set, the three-layer neural network model of setting up with this training set training step 101.In different application runnings, the active characteristics of 12 mobile terminals described in collection above, thus form the terminal operating characteristic vector.In the present embodiment, the data of 20 users and 27 application have been gathered as training set.
Step 103: after the user starts an application, gather the terminal operating characteristic vector of mobile terminal in this application running, three-layer neural network model by collected terminal operating characteristic vector input through step 102 training, obtain in 300MHz, 600MHz, 800MHz, HIGHER at output layer, resulting frequency is this and applies corresponding saturation frequency (when Output rusults is HIGHER, can think that saturation frequency is too high, now can be considered the frequency lid for empty, the frequency lid is not set in addition).
The inventor has implemented the model that 4 retransposing proof procedures are verified the present embodiment.
Experiment adopts the ME525 of Motorola mobile phone.Its application processor adopts TIOMAP3610, integrated ARM Cortex A8 kernel, and it has 300MHz, 600MHz, tri-frequencies that can regulate of 800MHz.Adopt the integrated display chip of PowerVR SGX530 simultaneously.Almost can be competent at the 3D game of all existing main flows.Have 512M RAM, and the ten thousand pixel TFT screens that have been equipped with 3.7 cun 1600.Adopt android2.2 operating system, can well meet user's experience under high performance mode.
The inventor has developed a background service and has carried out acquisition system action message (being the end activity characteristic vector), and its sample frequency is for being 5Hz.Gathered system activity information when 20 users use 27 application in experiment, and the user experiences feedback.27 application testing are on Google Play, to be all popular, comprising angry bird, QQ, UC browser etc.Each test process duration 3 minutes, total acquisition time has reached 20 hours.After carrying out feature extraction, just obtain 540 training examples, and labelled to each training examples by corresponding saturation frequency (experiencing feedback by the user obtains).
Training set is divided into to 4 groups randomly, then chooses arbitrarily 3 and merge into training set, another one is as the checking collection.Like this, the average training set precision of prediction obtained is 92%, and what checking was gathered is 80%.The inventor has also used 5 both not occur in training set, and the new application also do not occurred at the checking collection, as test set, is tested the model trained.The experiment demonstration, this model can obtain 82% precision of prediction.
Respectively apply the scheme of saturation frequency based on above-mentioned prediction, can do the power consumption management to mobile terminal further, thereby accurately reduce the power consumption of mobile terminal.
According to another embodiment of the invention, provide a kind of mobile terminal power consumption management method, comprised the following steps:
Step 1 a: during application of terminal operating, obtain the saturation frequency of this application.The method of prediction saturation frequency, referring to embodiment above, repeats no more herein.
In the present embodiment, the framework of whole system can adopt the form of client-server.The model of prediction saturation frequency completes training at high performance server end.Then, server sends the parameter of forecast model to client, namely intelligent mobile terminal.In client, set up a history lists and recorded the frequency lid (being saturation frequency) of each application.When each application moves for the first time, obtain corresponding frequency by the end activity characteristic vector prediction collected and cover and be recorded in described history lists.Like this, each this application of startup later, only need to check that history lists can obtain corresponding frequency lid.If do not find in history lists, system will start forecasting process, and the new application by this-frequency lid record is added in history lists.
In addition, consider that when forecast model was not met the application of newtype and occurred, forecast model need to improve self adaptively.So the present embodiment provides user interface, the frequency lid that makes the user manually to arrange.Information (terminal operating characteristic vector) when simultaneously, the background program of terminal is by the value of this frequency lid and system operation is sent to server.These data will become the new training sample of machine learning, thus continuous prediction correcting model.
Step 2: based on the ONDEMAND algorithm, the frequency of CPU is carried out to dynamic adjustments, wherein, for each application, respectively using the highest frequency of the resulting saturation frequency of step 1 in the ONDEMAND algorithm.
The power consumption management method of the present embodiment is based upon on the ONDEMAND algorithm, original Linux ONDEMAND algorithm is: when cpu busy percentage (cpu_util) is greater than a upper limit (UP_THRESHOLD), by the set of frequency of CPU at " highest frequency " (highest_frequency); When cpu busy percentage is less than a lower limit (DOWN_THRESHOLD), frequency setting can maintained at least level of (UP_THRESHOLD-10%) of cpu busy percentage.The correlative code of ONDEMAND algorithm is as follows:
In the present embodiment, in the method for taking dynamic restriction " highest frequency ", namely according to different application, give different " highest_frequency ", this " highest frequency " can be described as the frequency lid.Adopt this method can effectively prevent that cpu frequency from excessively heightening.And, experiment showed, that the frequency lid is set can't cause obviously reduce the running time respectively be applied on low frequency.As shown in Figure 3, the method of more original ONDEMAND algorithm and the present embodiment, be applied in the not significant change of time distribution on this low frequency of 300MHz, that is to say, eliminate the remarkable minimizing that excessively heightening of frequency can't be brought running time on low frequency, therefore can reduce the power consumption of CPU on the whole.
In order to verify the validity of the present embodiment, the inventor has studied frequency covers the matching rate predicted the outcome with user's actual selection.Whether the precision of prediction that frequency is covered has determined to arrange the frequency lid and can avoid the user to experience decline.Table 1 has provided the matching rate predicted the outcome with user's actual selection.
Table 1
Numerical value sum on diagonal has reached 87.4%, means that precision of prediction reaches 87.4%.The numerical value sum of upper Delta Region means the ratio that false positive is shared, and this situation does not reach the energy-saving effect of expection, but can't damage user's experience.The frequency lid that the numerical value sum representative of lower triangle is predicted is on the low side, may have influence on the user and experience, but this situation only accounts for 7.6% generally.That is to say, in the situation that 92.4%, the decline that the algorithm of the present embodiment can not bring the user to experience, so, it can be used safely.
And application processor energy-saving effect aspect: use Power Monitor to measure the actual power loss of mobile phone.CPU power consumption contrast when Fig. 9 shows difference under algorithms of different and should move.As shown in Figure 9, the precision of prediction of the present embodiment is close to optimum Oracle strategy.The experimental result demonstration, for CPU, the scheme of the present embodiment can make power-dissipation-reduced 11% to 84%.
Usually, the little application for computation requirement, as Strom:Lin and Chinese Chess, the effect of the present embodiment is clearly, but the larger application for the computation requirement amount, as Strom:Run and Gofishing, also has certain effect.It should be noted that, amount of calculation is not the reliable foundation of assessment energy-saving effect.Such as, this game of Need for Speed, computational requirements is very large in fact for it, but it does not but need high frequency lid.The scheme of the present embodiment also can identify this situation.This shows, the scheme of the present embodiment goes for many situations.
The electric weight of saving on application processor, also can see effect in the level of complete machine usually.Figure 10 shows the comparison of Overall Power Consumption of the strategy of linuxe ONDEMAND DFS strategy and the present embodiment.Result shows, on the complete machine level the energy-conservation 10mW of the scheme of the present embodiment to 466mW, average out to 82mW.For specific application, such as Need for Speed and Storm:Dragon, there is respectively the power consumption of 466mW and 256mW to save.For UC Browser and the such application of QQ, although the relatively energy-conservation of application processor is not very high, finally still have that 93mW and 73mW's is definitely energy-conservation.This is because the power consumption of application processor is the main power consumption of complete machine at this moment.
The system-level energy-conservation of this tens milliwatt is significant for smart phone user, because it can very significant prolongs standby time.The power consumption of smart mobile phone when standby is 25mW, and the strategy of the present embodiment is 82mW at the average energy-conservation of when operation application.This just means, as long as the mobile phone active state (non-holding state) of accumulation reaches 5 hours, just can prolongs standby time reach 16 hours.
Finally it should be noted that, above embodiment is only in order to describe technical scheme of the present invention rather than the present technique method is limited, the present invention can extend to other modification, variation, application and embodiment in application, and therefore thinks that all such modifications, variation, application, embodiment are in spirit of the present invention and teachings.
Claims (8)
1. the model training method of a predicted application saturation frequency, comprise the following steps:
1) set up forecast model, the terminal operating characteristic vector that when this model will be applied operation, the active characteristics of terminal forms is as input, using the Candidate Frequency lid as output; When wherein, described application moves, the active characteristics of terminal comprises: the busy time of CPU accounts for ratio or the percentage of running time; Ratio or the percentage of the running time of CPU on each frequency; And CPU continues average and the variance of busy time span;
2) train described forecast model based on sample set, the active characteristics of terminal while having gathered the application operation in this sample set, and corresponding user experiences and cpu frequency.
2. the model training method of predicted application saturation frequency according to claim 1, is characterized in that, during described application operation, the active characteristics of terminal also comprises: touch screen causes average and the variance of system break number of times; The average of touch screen time span; The average of interval time and variance between touch screen; And the gravity average of component and variance on the y direction of principal axis in the terminal local Coordinate System.
3. the model training method of predicted application saturation frequency according to claim 1, is characterized in that, described forecast model is neural network model.
4. the model training method of predicted application saturation frequency according to claim 1, it is characterized in that, described step 2), in, when the ratio of the average user experience value of a large number of users and maximum user experience value surpasses predetermined threshold, regard the frequency that reaches capacity as.
5. the method for a predicted application saturation frequency, comprise the following steps:
Start application, gather the terminal operating characteristic vector of mobile terminal in this application running, the forecast model that collected terminal operating characteristic vector input is trained by the described method of any one in claim 1～4, the frequency that this forecast model is exported is this and applies corresponding saturation frequency.
6. an intelligent mobile terminal power consumption management method, comprise the following steps:
11) utilize method claimed in claim 5 to obtain the saturation frequency of application;
12) based on the ONDEMAND algorithm, the frequency of CPU is carried out to dynamic adjustments, wherein the resulting saturation frequency of step 11) is as the highest frequency in the ONDEMAND algorithm.
7. intelligent mobile terminal power consumption management method according to claim 6, it is characterized in that, in described step 11), the model of predicted application saturation frequency completes training at server end, then, server sends the parameter of described model to mobile terminal, set up at mobile terminal the saturation frequency that history lists records each application, when each application moves for the first time, obtain corresponding saturation frequency and be recorded in described history lists by the end activity characteristic vector prediction collected, each this application of startup, check that history lists obtains corresponding saturation frequency later.
8. intelligent mobile terminal power consumption management method according to claim 7, it is characterized in that, in described step 11), for the unseen newtype application of described forecast model, mobile terminal is accepted frequency and the user experience value that the user manually arranges, and the two and corresponding terminal operating characteristic vector are sent to server, described server adds the described sample set for model training by it.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310403344.2A CN103475790B (en) | 2013-09-06 | 2013-09-06 | A kind of intelligent mobile terminal power consumption management |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310403344.2A CN103475790B (en) | 2013-09-06 | 2013-09-06 | A kind of intelligent mobile terminal power consumption management |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103475790A true CN103475790A (en) | 2013-12-25 |
| CN103475790B CN103475790B (en) | 2016-03-30 |
Family
ID=49800479
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310403344.2A Active CN103475790B (en) | 2013-09-06 | 2013-09-06 | A kind of intelligent mobile terminal power consumption management |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103475790B (en) |
Cited By (20)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104699224A (en) * | 2015-02-26 | 2015-06-10 | 深圳市金立通信设备有限公司 | Power-saving control method |
| WO2015172367A1 (en) * | 2014-05-16 | 2015-11-19 | 华为技术有限公司 | Method and apparatus for resource configuration |
| CN105867521A (en) * | 2015-02-10 | 2016-08-17 | 马维尔国际有限公司 | Method and apparatus for adjusting the clock frequency of hardware in a computing device, and user equipment |
| CN106055070A (en) * | 2016-05-24 | 2016-10-26 | 青岛海信移动通信技术股份有限公司 | Terminal power consumption control method and apparatus |
| CN106063337A (en) * | 2014-02-20 | 2016-10-26 | 高通股份有限公司 | Wake lock management through application monitoring |
| WO2016179977A1 (en) * | 2015-05-12 | 2016-11-17 | AMD Products (China) Co., Ltd. | Temporal thermal coupling aware power budgeting method |
| CN106484077A (en) * | 2016-10-19 | 2017-03-08 | 上海青橙实业有限公司 | Mobile terminal and its electricity saving method based on application software classification |
| CN106648023A (en) * | 2016-10-02 | 2017-05-10 | 上海青橙实业有限公司 | Mobile terminal and power-saving method of mobile terminal based on neural network |
| WO2017113696A1 (en) * | 2015-12-31 | 2017-07-06 | 乐视控股（北京）有限公司 | Frequency modulation parameter adjusting method, device, frequency modulator and mobile terminal |
| CN107168859A (en) * | 2017-05-09 | 2017-09-15 | 中国科学院计算技术研究所 | Energy consumption analysis method for Android device |
| WO2018024071A1 (en) * | 2016-08-01 | 2018-02-08 | Huawei Technologies Co., Ltd. | Application-specific, performance-aware energy optimization |
| CN107992183A (en) * | 2017-12-18 | 2018-05-04 | 联想(北京)有限公司 | A kind of optimization method and electronic equipment of CPU power consumption limit value |
| CN108228414A (en) * | 2017-04-06 | 2018-06-29 | 珠海市魅族科技有限公司 | Processor process parameter monitoring method and monitoring system based on Linux |
| WO2018157785A1 (en) * | 2017-03-02 | 2018-09-07 | Huawei Technologies Co., Ltd. | Apparatus and method for reducing energy consumption of hardware based on application state |
| CN108829598A (en) * | 2018-06-13 | 2018-11-16 | 深圳大学 | A kind of consumption detection method and system of mobile application |
| CN109062715A (en) * | 2018-07-05 | 2018-12-21 | Oppo（重庆）智能科技有限公司 | The determination method, apparatus and terminal of memory clock frequency |
| CN109151895A (en) * | 2017-06-16 | 2019-01-04 | 中国移动通信有限公司研究院 | A kind of data transmission method, device, server and Network Central Node |
| CN109324902A (en) * | 2018-09-21 | 2019-02-12 | 深圳市中科明望通信软件有限公司 | A kind of method, mobile terminal and storage medium adjusting mobile terminal working frequency |
| CN114217688A (en) * | 2021-12-24 | 2022-03-22 | 中国科学技术大学 | NPU power consumption optimization system and method based on neural network structure |
| CN117130769A (en) * | 2023-02-25 | 2023-11-28 | 荣耀终端有限公司 | Frequency modulation method, training method of frequency adjustment neural network and electronic equipment |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080065342A1 (en) * | 2006-09-11 | 2008-03-13 | Mainnet Communications Ltd. | Apparatus and method to provide power grid diagnostics |
| US20120166861A1 (en) * | 2010-12-28 | 2012-06-28 | Kuo-Cheng Lu | Method for adjusting clock frequency of a processing unit of a computer system and related device |
| CN102929383A (en) * | 2012-11-06 | 2013-02-13 | 山东大学 | Method for reducing embedded system power consumption through central processing unit (CPU) dynamic frequency modulation technology |
-
2013
- 2013-09-06 CN CN201310403344.2A patent/CN103475790B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080065342A1 (en) * | 2006-09-11 | 2008-03-13 | Mainnet Communications Ltd. | Apparatus and method to provide power grid diagnostics |
| US20120166861A1 (en) * | 2010-12-28 | 2012-06-28 | Kuo-Cheng Lu | Method for adjusting clock frequency of a processing unit of a computer system and related device |
| CN102929383A (en) * | 2012-11-06 | 2013-02-13 | 山东大学 | Method for reducing embedded system power consumption through central processing unit (CPU) dynamic frequency modulation technology |
Non-Patent Citations (1)
| Title |
|---|
| 张立 等: "一种基于Android系统网络模块功耗的评估和分析", 《计算机科学》 * |
Cited By (30)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106063337B (en) * | 2014-02-20 | 2019-10-18 | 高通股份有限公司 | Pass through the wake-up lock tube reason of application monitoring |
| CN106063337A (en) * | 2014-02-20 | 2016-10-26 | 高通股份有限公司 | Wake lock management through application monitoring |
| WO2015172367A1 (en) * | 2014-05-16 | 2015-11-19 | 华为技术有限公司 | Method and apparatus for resource configuration |
| CN105308941A (en) * | 2014-05-16 | 2016-02-03 | 华为技术有限公司 | Method and apparatus for resource configuration |
| CN105867521A (en) * | 2015-02-10 | 2016-08-17 | 马维尔国际有限公司 | Method and apparatus for adjusting the clock frequency of hardware in a computing device, and user equipment |
| CN104699224A (en) * | 2015-02-26 | 2015-06-10 | 深圳市金立通信设备有限公司 | Power-saving control method |
| WO2016179977A1 (en) * | 2015-05-12 | 2016-11-17 | AMD Products (China) Co., Ltd. | Temporal thermal coupling aware power budgeting method |
| WO2017113696A1 (en) * | 2015-12-31 | 2017-07-06 | 乐视控股（北京）有限公司 | Frequency modulation parameter adjusting method, device, frequency modulator and mobile terminal |
| CN106055070A (en) * | 2016-05-24 | 2016-10-26 | 青岛海信移动通信技术股份有限公司 | Terminal power consumption control method and apparatus |
| WO2018024071A1 (en) * | 2016-08-01 | 2018-02-08 | Huawei Technologies Co., Ltd. | Application-specific, performance-aware energy optimization |
| CN109564458A (en) * | 2016-08-01 | 2019-04-02 | 华为技术有限公司 | Application program is specific, performance aware it is energy-optimised |
| CN106648023A (en) * | 2016-10-02 | 2017-05-10 | 上海青橙实业有限公司 | Mobile terminal and power-saving method of mobile terminal based on neural network |
| CN106484077A (en) * | 2016-10-19 | 2017-03-08 | 上海青橙实业有限公司 | Mobile terminal and its electricity saving method based on application software classification |
| WO2018157785A1 (en) * | 2017-03-02 | 2018-09-07 | Huawei Technologies Co., Ltd. | Apparatus and method for reducing energy consumption of hardware based on application state |
| US10732703B2 (en) | 2017-03-02 | 2020-08-04 | Futurewei Technologies, Inc. | Apparatus and method for reducing an energy consumption of hardware based on an application state |
| CN108228414A (en) * | 2017-04-06 | 2018-06-29 | 珠海市魅族科技有限公司 | Processor process parameter monitoring method and monitoring system based on Linux |
| CN108228414B (en) * | 2017-04-06 | 2021-03-16 | 珠海市魅族科技有限公司 | Linux-based processor process parameter monitoring method and monitoring system |
| CN107168859A (en) * | 2017-05-09 | 2017-09-15 | 中国科学院计算技术研究所 | Energy consumption analysis method for Android device |
| CN109151895B (en) * | 2017-06-16 | 2022-07-22 | 中国移动通信有限公司研究院 | Data transmission method, device, server and network center node |
| CN109151895A (en) * | 2017-06-16 | 2019-01-04 | 中国移动通信有限公司研究院 | A kind of data transmission method, device, server and Network Central Node |
| CN107992183B (en) * | 2017-12-18 | 2022-05-31 | 联想(北京)有限公司 | CPU power consumption limit value optimization method and electronic equipment |
| CN107992183A (en) * | 2017-12-18 | 2018-05-04 | 联想(北京)有限公司 | A kind of optimization method and electronic equipment of CPU power consumption limit value |
| CN108829598B (en) * | 2018-06-13 | 2019-11-29 | 深圳大学 | A kind of consumption detection method and system of mobile application |
| CN108829598A (en) * | 2018-06-13 | 2018-11-16 | 深圳大学 | A kind of consumption detection method and system of mobile application |
| CN109062715A (en) * | 2018-07-05 | 2018-12-21 | Oppo（重庆）智能科技有限公司 | The determination method, apparatus and terminal of memory clock frequency |
| CN109324902A (en) * | 2018-09-21 | 2019-02-12 | 深圳市中科明望通信软件有限公司 | A kind of method, mobile terminal and storage medium adjusting mobile terminal working frequency |
| CN109324902B (en) * | 2018-09-21 | 2021-03-26 | 深圳市中科明望通信软件有限公司 | Method for adjusting working frequency of mobile terminal, mobile terminal and storage medium |
| CN114217688A (en) * | 2021-12-24 | 2022-03-22 | 中国科学技术大学 | NPU power consumption optimization system and method based on neural network structure |
| CN114217688B (en) * | 2021-12-24 | 2024-05-10 | 中国科学技术大学 | NPU power consumption optimization system and method based on neural network structure |
| CN117130769A (en) * | 2023-02-25 | 2023-11-28 | 荣耀终端有限公司 | Frequency modulation method, training method of frequency adjustment neural network and electronic equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103475790B (en) | 2016-03-30 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103475790A (en) | Intelligent mobile terminal power consumption management method | |
| CN108090652A (en) | The method for building the electricity transaction index system based on big data technology | |
| CN103092699A (en) | Cloud computing resource pre-distribution achievement method | |
| CN109766234A (en) | Disk storage capacity prediction technique based on time series models | |
| CN103024762A (en) | Service feature based communication service forecasting method | |
| Li et al. | Smartcap: User experience-oriented power adaptation for smartphone's application processor | |
| CN102185731B (en) | Network health degree testing method and system | |
| Peters et al. | Web browser workload characterization for power management on HMP platforms | |
| CN105786711A (en) | Data analysis method and device | |
| CN110458594A (en) | A kind of network surveying questionnaire generation method and device based on user's weight | |
| CN111737078A (en) | Load type-based adaptive cloud server energy consumption measuring and calculating method, system and equipment | |
| CN109242169A (en) | A kind of short-term load forecasting method towards large power customers | |
| CN109947558A (en) | Host resource utilization rate calculation method and resource regulating method | |
| CN112308270A (en) | Long-term electricity load prediction method and device and computer implementation system | |
| CN114781766B (en) | Hydrological information prediction method, device, equipment and storage medium for hydrological site | |
| CN112954707A (en) | Energy saving method and device for base station, base station and computer readable storage medium | |
| CN112433813A (en) | Resource scheduling method, system, equipment and computer readable storage medium | |
| CN105338186A (en) | Context awareness-based Android mobile terminal power management method | |
| CN111404974B (en) | Cloud computing efficiency evaluation method and device and evaluation equipment | |
| CN118193072A (en) | Method and device for controlling server dormancy, storage medium and electronic equipment | |
| KR20150061914A (en) | Skin texture predicting method and apparatus | |
| CN102546999B (en) | Method, control device and system for reducing device power consumption based on business model | |
| CN117350366A (en) | Network model construction method and related equipment | |
| Cheng et al. | Behavior-aware integrated CPU-GPU power management for mobile games | |
| CN103217897B (en) | Method and system for optimizing screen refresh rate of touch-screen smart mobile terminal |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C53 | Correction of patent of invention or patent application | ||
| CB03 | Change of inventor or designer information |
Inventor after: Yan Guihai Inventor after: Han Yinhe Inventor after: Li Xueliang Inventor after: Li Xiaowei Inventor before: Li Xueliang Inventor before: Yan Guihai Inventor before: Han Yinhe Inventor before: Li Xiaowei |
|
| COR | Change of bibliographic data |
Free format text: CORRECT: INVENTOR; FROM: LI XUELIANG YAN GUIHAI HAN YINHE LI XIAOWEI TO: YAN GUIHAI HAN YINHE LI XUELIANG LI XIAOWEI |
|
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| EE01 | Entry into force of recordation of patent licensing contract |
Application publication date: 20131225 Assignee: Zhongke Yuanshu (Beijing) Technology Co., Ltd. Assignor: Institute of Computing Technology, Chinese Academy of Sciences Contract record no.: X2019990000091 Denomination of invention: Intelligent mobile terminal power consumption management method Granted publication date: 20160330 License type: Common License Record date: 20190919 |
|
| EE01 | Entry into force of recordation of patent licensing contract |