CN109905868A - A kind of intelligence wearable device Bluetooth communication prediction technique and system - Google Patents
A kind of intelligence wearable device Bluetooth communication prediction technique and system Download PDFInfo
- Publication number
- CN109905868A CN109905868A CN201910148244.7A CN201910148244A CN109905868A CN 109905868 A CN109905868 A CN 109905868A CN 201910148244 A CN201910148244 A CN 201910148244A CN 109905868 A CN109905868 A CN 109905868A
- Authority
- CN
- China
- Prior art keywords
- wearable device
- rssi value
- network model
- neural network
- server
- 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
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
- 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 discloses a kind of intelligent wearable device Bluetooth communication prediction technique and systems, method includes training process, prediction process and communication process: the training process includes: the RSSI value that server-side samples intelligent wearable device, generate training sample, neural network model is trained, trained neural network model is obtained；The prediction process includes: the RSSI value that server-side samples intelligent wearable device, generates analysis sample, is analyzed by the neural network model the analysis sample, predict the peak value moment of next RSSI value；The communication process includes: the intelligent wearable device according to the peak value moment and server-side progress Bluetooth communication.Have many advantages, such as to can effectively reduce energy consumption of intelligent wearable device during Bluetooth communication prediction, forecasting accuracy is good, can effectively improve Bluetooth communication quality.
Description
Technical field
The present invention relates to technical field of bluetooth communication more particularly to a kind of intelligent wearable device Bluetooth communication prediction techniques
And system.
Background technique
How in the existing method for reducing wearable device Bluetooth communication energy consumption, periodic motion is reduced most possibly
The path loss of middle radio-frequency module transmission signal, signal minimal path loss point of the pre- measurement equipment in periodic motion is therein
One of important technology.The technology of signal minimal path loss point in predetermined period movement is a kind of opportunistic, real-time
It is required that relatively high Predicting Technique, can realize wearable under the premise of telecommunication service quality under meeting periodic motion scene
The minimum of equipment communication energy consumption, the technology will communicate in conjunction with prediction technique, take full advantage of intelligence under edge calculations scene
It can mobile phone and the extra computing resource of edge calculations node.In recent years, flourishing with wearable device, it is energy-saving
Predicting Technique be widely used in various wireless communication systems.
A typical scene of the wearable device in Bluetooth communication are as follows: wearable Intelligent bracelet/smartwatch is worn on
The left/right wrist of human body monitors the health indicators such as movement, the heart rate of human body, while by these health indicator data via bluetooth
Radio-frequency module is sent in smart phone, and the energy mobile phone for being stored in left/right pocket of trousers is responsible for data as the receiving end of bluetooth equipment
Storage and calculating.Wearable device movement approximately periodic with human body for a long time, such as walk, running, bluetooth transceiver
Between distance periodically can increase and reduce, the path attenuation of signal in the air also can in periodically variation in, even
When body is periodically blocked between transceiver, the path loss of signal is up to maximum.To guarantee by body check
When the radio frequency signal decayed can guarantee between bluetooth equipment can reliable communication, the transmission power of Bluetooth RF module is often
It is brought to most high-grade.This is undoubtedly a kind of most conservative approach for coping with worst case, for situations most in movement, radio frequency
Maximum path loss is often not achieved in the decaying of signal, if sending data with maximum transmission power, energy is all waste.
So for the communication energy consumption of bluetooth equipment is effectively reduced, currently used method is the minimal path in radiofrequency signal
Diameter loss point executes communication task.Wearable device indicates to receive signal matter using received signal strength indicator (i.e. RSSI value)
Amount, when RSSI value reaches wave crest, the path loss of wireless signal is up to minimum value.Currently, when being arrived using prediction RSSI value
Carry out the minimal path loss point of prediction signal up to the method for wave crest, wearable device only executes communication in minimal path loss point and appoints
Business, the setting of communication equipment transmission power only need to guarantee there is reliable communication quality, this side at the minimal path loss point
Method can greatly reduce the communication energy consumption of wearable device.
The prediction wearable device RSSI value method that when reaches wave crest mainly has: a kind of method is smart phone periodicity
Wearable device is acquired in the RSSI value of transmission phase, and the RSSI value is sent back into wearable device, wearable device makes again
The fundamental frequency f of RSSI signal sequence is extracted with fast Fourier method (FFT transform)0, with fundamental frequency f0InverseAs next
The Approximate prediction when a RSSI value wave crest arrives.It is fairly simple at the time of this method prediction RSSI value arrival wave crest, in a short time
Can obtain preferable effect, but the disadvantage is that: the movement of human body is not stringent periodic motion, when the period of motion with tie up before
There is little deviation in the period of motion held, and prediction result can gradually be accumulated into biggish offset error；When offset error is larger,
The fundamental frequency f that FFT transform is looked for novelty is re-started again0, frequently calculate also larger to the energy consumption of wearable device.
Another method is the RSSI value of smart phone periodically acquisition wearable device, is equally that send back the value can
Wearable device, wearable device acquire local three axis accelerometer signal simultaneously, while handling RSSI value and accelerometer letter
Number, by both data fusions, the deviation of acceleration value and RSSI value adjacent peaks is obtained, cluster behaviour is carried out to all deviations
Make, using the mean value of maximum cluster as the deflection forecast of the next wave crest of RSSI value and the current wave crest of acceleration value, with acceleration
When supposition RSSI value reaches wave crest based on counting current wave crest.This method acceleration of motion meter wave crest moves Clustering,
The accuracy of prediction is improved to a certain extent.The disadvantage is that: it is required that wearable device has accelerometer, cost is excessively high；In addition,
Clustering is not enough to predict the not stringent periodic motion of human body, generates drift when the period of motion, prediction error also can gradually increase
Greatly, it can be solved by correction although error increases, frequent correct operation, also by the electricity of great consumption wearable device
Amount.
Which kind of Predicting Technique is either used, error caused by coping with human motion period migration is all difficult to, requires
Predicted value is calculated on wearable device.
Summary of the invention
The technical problem to be solved in the present invention is that, for technical problem of the existing technology, the present invention provides one
It kind can effectively reduce energy consumption of intelligent wearable device during Bluetooth communication prediction, forecasting accuracy is good, can be effective
Improve the intelligent wearable device Bluetooth communication prediction technique and system of Bluetooth communication quality.
In order to solve the above technical problems, technical solution proposed by the present invention are as follows: a kind of intelligence wearable device Bluetooth communication
Prediction technique, including training process, prediction process and communication process:
The training process includes: the RSSI value that server-side samples intelligent wearable device, training sample is generated, to nerve
Network model is trained, and obtains trained neural network model；
The prediction process includes: the RSSI value that server-side samples intelligent wearable device, generates analysis sample, passes through institute
It states neural network model to analyze the analysis sample, predicts the peak value moment of next RSSI value；
The communication process includes: that the intelligent wearable device is blue according to the peak value moment and server-side progress
Tooth communication.
Further, the communication process includes: that server-side actively issues communication application, intelligence to intelligent wearable device
Wearable device opens the Bluetooth communication between the server-side according to the communication application；
Or:
Intelligent wearable device carries out time supervision by receiving the peak value moment, reaches in the peak value moment
Bluetooth communication between Shi Kaiqi and the server-side.
Further, in the training process, further include the validity that server-side analyzes the RSSI value variation, work as institute
When stating validity and meeting preset constraint condition, training sample is generated according to the RSSI value.
Further, it includes: to carry out in advance to the RSSI value that sampling obtains that training sample is generated in the training process
Processing, obtain the position of the wave crest in the RSSI value, and calculate the alternate position spike of adjacent peaks, using the alternate position spike information as
Training sample.
Further, it includes: to analyze the RSSI value that analysis sample is generated during the prediction, is obtained described
The crest location of RSSI value, and the alternate position spike of two neighboring wave crest is calculated, and using alternate position spike described in one group as analysis sample.
Further, the server-side includes first service end and second service end；In the training process: described
One server-side samples the RSSI value of intelligent wearable device, and generates training sample；The second service end is according to the training
Sample is trained neural network model, and trained neural network model is passed to the first service end.
Further, further include verifying adjustment process: server-side monitors the RSSI value of the intelligent wearable device, and sentences
Deviation between the disconnected peak value moment predicted and true peak, when the deviation is unsatisfactory for preset decision condition, into one
Step is trained the neural network model.
Further, the preset decision condition includes:
P1: when the deviation is less than predetermined deviation threshold value, determine that deviation is met the requirements；
P2: when recurring the number for being unsatisfactory for P1 greater than preset threshold value, judgement is unsatisfactory for decision condition.
A kind of intelligence wearable device Bluetooth communication forecasting system, including server-side and intelligent wearable device；The intelligence
Energy wearable device is communicated to connect by blueteeth network and the server-side；
The server-side is used to sample the RSSI value of intelligent wearable device, training sample is generated, to neural network model
It is trained, obtains trained neural network model；And the RSSI value of intelligent wearable device is sampled, analysis sample is generated,
The analysis sample is analyzed by the neural network model, predicts the peak value moment of next RSSI value；
The intelligence wearable device carries out Bluetooth communication according to the peak value moment and the server-side.
Further, the server-side includes first service end and second service end；
The first service end samples the RSSI value of intelligent wearable device, and generates training sample；The second service
End is trained neural network model according to the training sample, and trained neural network model is passed to described the
One server-side.
Further, the server-side is also used to: the RSSI value of the monitoring intelligent wearable device, and judges to predict
Peak value moment and true peak between deviation, when the deviation is unsatisfactory for preset decision condition, further to described
Neural network model is trained.
Compared with the prior art, the advantages of the present invention are as follows:
1, the present invention is by server-side monitoring from the received signal strength indicator of intelligent wearable device Bluetooth signal
It is worth (RSSI value) Lai Xunlian neural network model, and by monitoring the RSSI value, is divided by trained neural network model
The peak value moment of next RSSI value is predicted in analysis, to open between intelligent wearable device and server-side in the peak value moment
Bluetooth communication does not carry out at intelligent wearable device end during neural network model training and prediction, does not need to consume
The electricity of intelligent wearable device, moreover, server-side is set by the way that passively monitoring intelligence is wearable in trained and monitoring process
Standby RSSI value can be realized as, and does not need intelligent wearable device and sends additional data to server-side, does not also need intelligence
There is additional sensor on wearable device to support (such as acceleration transducer), for intelligent wearable device, energy
It consumes small, the stand-by time of intelligent wearable device can be effectively ensured.
2, server-side actively issues communication application, intelligence to intelligent wearable device when peak value moment reaches in the present invention
Wearable device opens the Bluetooth communication with server-side according to this application, so that can also be in server-side to the monitoring of peak value moment
It carries out, it can be achieved that intelligent wearable device end only needs to consume energy required for sending data, further reduced can
Dress the energy consumption of smart machine.
3, the present invention is by neural network model, when analyzing prediction peak value especially by wavelet-neural network model
It carves, can sufficiently learn, analyze the local feature of signal in the time domain, the period is small in Accurate Prediction human cyclin motion process
Offset, predictablity rate are high.
Detailed description of the invention
Fig. 1 is the flow diagram of the specific embodiment of the invention.
Fig. 2 is the specific implementation step schematic diagram of the specific embodiment of the invention.
Fig. 3 is the system composed structure schematic diagram of the specific embodiment of the invention.
Specific embodiment
Below in conjunction with Figure of description and specific preferred embodiment, the invention will be further described, but not therefore and
It limits the scope of the invention.
In the present embodiment, intelligent wearable device Bluetooth communication prediction technique, including training process, prediction process and logical
Letter process: training process includes: the RSSI value that server-side samples intelligent wearable device, training sample is generated, to neural network
Model is trained, and obtains trained neural network model；Prediction process includes: that server-side samples intelligent wearable device
RSSI value generates analysis sample, is analyzed by neural network model analysis sample, predicts the peak of next RSSI value
It is worth the moment；Communication process includes: intelligent wearable device according to peak value moment and server-side progress Bluetooth communication.
In the present embodiment, communication process includes: that server-side actively issues communication application, intelligence to intelligent wearable device
Wearable device opens the Bluetooth communication between server-side according to communication application；Or: intelligent wearable device passes through reception
Peak value moment, and time supervision is carried out, the Bluetooth communication between server-side is opened when peak value moment reaches.
In the present embodiment, it is preferable that it further include the validity of server-side analysis RSSI value variation in the training process, when
When validity meets preset constraint condition, training sample is generated according to RSSI value.
In the present embodiment, training sample is generated in training process includes: to analyze the RSSI value, is obtained described
The crest location of RSSI value, and the alternate position spike of two neighboring wave crest is calculated, and using alternate position spike described in one group as analysis sample.
In the present embodiment, it includes: that the RSSI value obtained to sampling pre-processes that analysis sample is generated during prediction,
The position of the wave crest in RSSI value is obtained, and calculates the alternate position spike of adjacent peaks, using alternate position spike information as analysis sample.
In the present embodiment, it is preferable that server-side includes first service end and second service end；In the training process: first
Server-side samples the RSSI value of intelligent wearable device, and generates training sample；Second service end is according to training sample to nerve
Network model is trained, and trained neural network model is passed to first service end.
In the present embodiment, it is preferable that further include verifying adjustment process: server-side monitors the RSSI of intelligent wearable device
Value, and judge the deviation between the peak value moment predicted and true peak, when deviation is unsatisfactory for preset decision condition, into
One step is trained neural network model.Preset decision condition includes: P1: when deviation is less than predetermined deviation threshold value,
Determine that deviation is met the requirements；P2: when recurring the number for being unsatisfactory for P1 greater than preset threshold value, judgement is unsatisfactory for determining item
Part.
The intelligent wearable device Bluetooth communication forecasting system of the present embodiment, including server-side and intelligent wearable device；
Intelligent wearable device is communicated to connect by blueteeth network and server-side；Server-side is used to sample the RSSI of intelligent wearable device
Value generates training sample, is trained to neural network model, obtains trained neural network model；And sample intelligence can
The RSSI value of wearable device generates analysis sample, is analyzed by neural network model analysis sample, is predicted next
The peak value moment of RSSI value；Intelligent wearable device carries out Bluetooth communication according to peak value moment and server-side.
In the present embodiment, server-side includes first service end and second service end；First service end sampling intelligence can be worn
The RSSI value of equipment is worn, and generates training sample；Second service end is trained neural network model according to training sample, and
Trained neural network model is passed into first service end.
In the present embodiment, server-side is also used to: being monitored the RSSI value of intelligent wearable device, and is judged the peak predicted
The deviation being worth between moment and true peak, when deviation is unsatisfactory for preset decision condition, further to neural network model
It is trained.
In the present embodiment, the above method of the invention and system are carried out specifically by a specific application scenarios
It is bright.As shown in Figures 2 and 3, in this application scene, including intelligent wearable device (smartwatch), as first service end
Smart phone and marginal end (or cloud or background server end etc.) as second service end.Intelligent wearable device
It is connected between first service end by bluetooth approach, passes through wireless network connection between first service end and second service end.
Smart phone, which can be monitored directly, determines the signal strength indication (RSSI value) from wearable smart machine bluetooth.It is wearable in intelligence
In the case that the bluetooth power of equipment is constant, variation with intelligent wearable device relative to the position of smart phone can draw
Play the variation for the signal strength indication that smart phone can receive.As shown in Figure 2, the bluetooth letter that intelligent wearable device end issues
Number intensity power is constant (waveform signal with essentially identical crest height), but in mobile phone end, received bluetooth
Signal then has apparent wave crest and trough, because intelligent the distance between wearable device and smart phone change (such as
The wearable arm with user of intelligence carries out regular swing) or centre have the resistance of barrier (such as user's body)
Every so that the attenuation degree difference of Bluetooth signal, therefore, produces apparent wave crest and trough.
It further include that server-side analyzes the RSSI value change to more save the energy consumption of smart phone in this application scene
The validity of change generates training sample according to the RSSI value when the validity meets preset constraint condition.It crosses herein
Cheng Zhong judges that the preset constraint condition of validity includes: the variation whether RSSI value is greater than predetermined threshold level and RSSI value
Whether preset variation threshold value is greater than.Smart phone judges whether the RSSI value from intelligent wearable device is greater than pre- gating
Whether the variation of threshold value and RSSI value is greater than preset variation threshold value, and intelligent wearable device is judged if all meeting
In effective exercise.If smart phone samples RSSI value with the frequency of 100Hz, a period (such as 10s) is obtained
Interior sampled value, obtained sampled value are stored in array, are denoted as RSSIdevice[n], then judging at intelligent wearable device
Mode in effective exercise is represented by shown in following formula: { max (RSSIdevice[n])≥-70,n∈[1,1000]}∩{max
(RSSIdevice[n])-min(RSSIdevice[n]) >=30, n ∈ [1,1000] }, in this scene, predetermined threshold level is set as 70,
Variation threshold value is set as 30.When judging that intelligent wearable device is in effective exercise, continue subsequent process, otherwise, then
It is repeated after waiting for a period of time and carries out above-mentioned deterministic process.As carried out movement inspection in mobile phone end in training process in Fig. 2
It surveys, after determining that intelligence wearable device is in normal effective exercise, then carries out subsequent step.
In this application scene, further, judge that the preset constraint condition of validity can also include: described in judgement
The period 1 property of RSSI value peak value judges at intelligent wearable device when period 1 property meets preset periodic condition
When in effective exercise, continue subsequent process, otherwise, is then repeated after waiting for a period of time and carry out above-mentioned deterministic process.Specifically
Ground is pre-processed by carrying out low-pass filtering etc. to RSSI value, then uses general wave crest lookup algorithm, first obtains the pole in RSSI value
Value removes valley value therein and local crest value, obtains real crest value T in RSSI valuepeak, wave crest lookup algorithm may
Existing algorithm is realized.The spacing value T of each adjacent wave peak value time of occurrence is calculated againgap(alternate position spike i.e. between adjacent peaks),
And further acquire the average value of spacing valueWhen the average bits of the spacing value are when preset first interval, then intelligence is judged
When energy wearable device is in effective exercise, continue subsequent process, otherwise, then it is above-mentioned to repeat progress after waiting for a period of time
Deterministic process.The average value that period 1 property passes through spacing valueTo characterize.In this application scene, which is preferably
[0,5Hz], i.e.,When, judge that intelligent wearable device is to wear on the user's body and in effective exercise,
Otherwise it is assumed that the frequency of proper motion of the motion frequency beyond human body, and think intelligent wearable device and be in lost motion
In, it is repeated after needing to wait for a period of time and carries out above-mentioned deterministic process.
In the present embodiment, further, the preset constraint condition for judging validity can also include: described in judgement
The Secondary periodicity of RSSI value peak value judges at intelligent wearable device when Secondary periodicity meets preset periodic condition
When in effective exercise, continue subsequent process, otherwise, is then repeated after waiting for a period of time and carry out above-mentioned deterministic process.Specifically
RSSI value including being obtained by Fast Fourier Transform (FFT) to sampling is handled, and extracts fundamental frequency f therein0, then ask reciprocalAnd according to T0Determine that preset second interval is [aT0,bT0], a, b are preset coefficient.If above-mentioned average value
When being still located on preset second interval, i.e.,Judge that intelligent wearable device is to wear on the user's body
And in effective exercise, otherwise, is then repeated after waiting for a period of time and carry out above-mentioned deterministic process.It is excellent in this application scene
Select a=0.9, b=1.1.
In the present embodiment, after the validity for determining RSSI value variation by above-mentioned judgement, smart phone can will be adopted
The effective RSSI value that sample obtains, or the RSSI value obtained to the sampling for re-starting a period of time are handled, and generate mind
Training sample through network model.It is to re-start the sampling of a period of time, then after judging that RSSI value is effective with this in Fig. 2
Section time sampling obtained RSSI value generates training sample.The process for generating training sample includes: the RSSI obtained to sampling
Value carries out low-pass filtering, obtains the crest value of RSSI using method identical with the validity of RSSI value variation is judged, and calculate
The alternate position spike of adjacent peaks, and using the position difference as training sample.The alternate position spike deposit array T being calculatedgap。
It in the present embodiment, can be directly in mobile phone end after smart phone obtains training sample by analysis RSSI value
Neural network model is trained by training sample, second service end can also be sent for training sample by network
(marginal end) is trained neural network model by second service end, and joins after the completion of training, then by neural network model
Number passes to smart phone, and trained neural network model is constructed on smart phone.By sending side for training sample
Acies is trained by marginal end, can be reduced requirement of the smart phone to calculated performance, be mitigated the calculating task of smart phone
Amount, reduces the energy consumption of smart phone.
In the present embodiment, neural network model is preferably wavelet-neural network model.It is preferred that wavelet neural network is defeated
Enter layer with 10 nodes, it includes 128 nodes that middle layer, which is 1 layer of hidden layer, and output layer is 1 node.It is trained for the first time
When, select preceding 11 continuous spacing value TgapAs 1 training sample, wherein the 1st to 10 spacing value TgapAs small echo mind
Input through network, the 11st spacing value TgapAs the output target value of wavelet neural network, after primary training, after
A sample is moved, that is, selects the 2nd to 11 spacing value TgapAs input, the 12nd spacing value T is selectedgapAs output target
Value.It repeats the above process, until training is completed.By using wavelet-neural network model, be multiplexed small echo length has the present embodiment
Limit, the waveform that average value is 0, the inner product of any small echo and constant function all level off to 0, the integral of morther wavelet in one cycle
Level off to 0 characteristic.This characteristic keeps the conversion ratio Fourier transformation of wavelet transformation reply signal even better, the reason is that,
Wavelet transformation allows to describe the separation of more accurate local feature and signal characteristic.Wavelet basis function is first put down in the time domain
Stretching after shifting, it can sufficiently learn, analyze the local feature of signal in the time domain, and this conversion process is referred to as small echo change
It changes, wavelet neural network, therefore Wavelet Neural Network is referred to as using the neural network of wavelet basis function as hidden layer transmission function
Network prediction model can preferably cope with the minor shifts in human motion period.
After training neural network model, smart phone continues sampling and obtains RSSI value, and is generated and divided according to RSSI value
Analyse sample.Since the input terminal of the wavelet-neural network model of the present embodiment uses 10 nodes, sample is analyzed generating
When, smart phone obtains 10 spacing value T by analyzing RSSI valuegap, then 10 spacing value TgapAs 1 analysis sample,
The analysis sample is input to trained wavelet-neural network model, can be calculated next wave crest of prediction relative to
The spacing value T of a nearest wave crestgap, the time of occurrence that can also predict to obtain next wave crest is a nearest wave crest
Time of occurrence+Tgap.Smart phone sampling obtains RSSI value and analyzes determining crest location, and calculates the position of adjacent peaks
Difference is all real-time perfoming.Specific process includes the RSSI value of smart phone real-time monitoring intelligence wearable device, and is passed through
The modes such as low filtering pre-process sampled value；Sampled value is detected in real time, its medium wave peak is gone out with real-time detection
Position.Existing method can be used in the method for real-time detection wave crest.Simplest mode such as predefines some threshold value, works as RSSI value
When by being less than the changes of threshold for greater than the threshold value, then there is wave crest in judgement.It is, of course, also possible in other manners, such as: false
It is located in most freshly harvested multiple RSSI values, inverse n-thp+ 1 RSSI value is denoted as RSSIphone[n], with its time interval npIt is a
The sampled value of sequence is denoted as RSSI respectivelyphone[n-np]、RSSIphone[n+np], if RSSIphone[n] meets { RSSIphone[n]>
RSSIth}∩{RSSIphone[n]≥max(RSSIphone[n-np],RSSIphone[n+np]), then the RSSI value is global wave
The peak value at peak, recording the crest location is tcurr-peak；Wherein, npFor the fault-tolerant factor, n is takenp=5, RSSIthFor array by greatly to
The RSSI value of float the 25%th.Smart phone is detected by the RSSI value to intelligent wearable device, is obtaining 10 companies
When spacing value between continuous adjacent peaks, then in this, as an analysis sample, smart phone passes through trained wavelet neural
Network is analyzed, i.e., the predictable appearance position for obtaining next wave crest is relative to the spacing value between a nearest wave crest
Tpre, then the current moment that goes out of next wave crest is tpre-peak=tbase+Tpre, tbaseGo out current moment for a nearest wave crest.Under
One wave crest goes out before current moment arrives, and intelligent wearable device can enter dormant state, until the moment arrive just into
Row wakes up, and carries out data communication.
In the present embodiment, it is predicting to obtain the spacing value T of next wave crest by wavelet-neural network modelpreAfterwards, i.e.,
The time that next wave crest occurs can be calculated.At this point it is possible to by spacing value TpreIt is sent to intelligent wearable device, or
The time that next wave crest occurs is sent to intelligent wearable device, when monitoring this by intelligent wearable device this whether to
It reaches, and opens Bluetooth communication when reaching at the moment, send the data to smart phone.And the moment reach before and data
After being sent completely, the bluetooth equipment of intelligent wearable device can enter dormant state.To only be needed in intelligent wearable device
In the case where wanting lesser bluetooth to send power, so that it may which data stable are sent to by high quality from intelligent wearable device
Smart phone.In the present embodiment, it at the time of next wave crest can also being monitored by smart phone and occur, and is reached at the moment
When, Bluetooth communication application is actively issued to intelligent wearable device from smart phone, intelligent wearable device is logical according to the bluetooth
Bluetooth communication is opened in letter application, sends the data to smart phone.Such mode can further reduce intelligently wearable set
Standby energy consumption.
In the present embodiment, next wave crest that smart phone also detects that prediction obtains goes out current moment and actual wave crest occurs
Deviation between moment, when deviation, which occurs, in connection is greater than preset threshold value greater than the number of predetermined deviation threshold value, then it is assumed that
Prediction result is not accurate enough, needs again to be trained neural network model.Occur by next wave crest that prediction obtains
Spacing value be Tpre, and be T by monitoring identified actual interval valuegap′.The difference of the two of so continuous 3 prediction is greater than
Actual interval value is Tgap' 10% when, then it is assumed that prediction result is not accurate enough, needs again to instruct neural network model
Practice.
Above-mentioned only presently preferred embodiments of the present invention, is not intended to limit the present invention in any form.Although of the invention
It has been disclosed in a preferred embodiment above, however, it is not intended to limit the invention.Therefore, all without departing from technical solution of the present invention
Content, technical spirit any simple modifications, equivalents, and modifications made to the above embodiment, should all fall according to the present invention
In the range of technical solution of the present invention protection.
Claims (10)
1. a kind of intelligence wearable device Bluetooth communication prediction technique, which is characterized in that including training process, prediction process and lead to
Letter process:
The training process includes: the RSSI value that server-side samples intelligent wearable device, training sample is generated, to neural network
Model is trained, and obtains trained neural network model；
The prediction process includes: the RSSI value that server-side samples intelligent wearable device, generates analysis sample, passes through the mind
The analysis sample is analyzed through network model, predicts the peak value moment of next RSSI value；
The communication process includes: that the intelligent wearable device is logical according to the peak value moment and server-side progress bluetooth
Letter.
2. intelligence wearable device Bluetooth communication prediction technique according to claim 1, it is characterised in that: in the training
It in the process, further include that server-side analyzes the validity that the RSSI value changes, when the validity meets preset constraint condition
When, training sample is generated according to the RSSI value.
3. intelligence wearable device Bluetooth communication prediction technique according to claim 2, it is characterised in that: described to train
It includes: that the RSSI value obtained to sampling pre-processes that training sample is generated in journey, obtains the wave crest in the RSSI value
Position, and the alternate position spike of adjacent peaks is calculated, using the alternate position spike information as training sample.
4. intelligence wearable device Bluetooth communication prediction technique according to claim 3, it is characterised in that: described to predict
It includes: to analyze the RSSI value that analysis sample is generated in journey, obtains the crest location of the RSSI value, and calculate phase
The alternate position spike of adjacent two wave crests, and using alternate position spike described in one group as analysis sample.
5. intelligence wearable device Bluetooth communication prediction technique according to claim 4, it is characterised in that: the server-side
Including first service end and second service end；In the training process: the first service end samples intelligent wearable device
RSSI value, and generate training sample；The second service end is trained neural network model according to the training sample,
And trained neural network model is passed into the first service end.
6. intelligence wearable device Bluetooth communication prediction technique according to any one of claims 1 to 5, it is characterised in that:
Further include verifying adjustment process: server-side monitors the RSSI value of the intelligent wearable device, and when judging the peak value predicted
The deviation between true peak is carved, when the deviation is unsatisfactory for preset decision condition, further to the neural network
Model is trained.
7. intelligence wearable device Bluetooth communication prediction technique according to claim 6, it is characterised in that: described preset
Decision condition includes:
P1: when the deviation is less than predetermined deviation threshold value, determine that deviation is met the requirements；
P2: when recurring the number for being unsatisfactory for P1 greater than preset threshold value, judgement is unsatisfactory for decision condition.
8. a kind of intelligence wearable device Bluetooth communication forecasting system, it is characterised in that: set including server-side with intelligently wearable
It is standby；The intelligence wearable device is communicated to connect by blueteeth network and the server-side；
The server-side is used to sample the RSSI value of intelligent wearable device, generates training sample, carries out to neural network model
Training, obtains trained neural network model；And the RSSI value of intelligent wearable device is sampled, analysis sample is generated, is passed through
The neural network model analyzes the analysis sample, predicts the peak value moment of next RSSI value；
The intelligence wearable device carries out Bluetooth communication according to the peak value moment and the server-side.
9. intelligence wearable device Bluetooth communication forecasting system according to claim 8, it is characterised in that: the server-side
Including first service end and second service end；
The first service end samples the RSSI value of intelligent wearable device, and generates training sample；Second service end root
Neural network model is trained according to the training sample, and trained neural network model is passed into first clothes
Business end.
10. intelligence wearable device Bluetooth communication forecasting system according to claim 9, it is characterised in that: the service
End is also used to: the RSSI value of the monitoring intelligent wearable device, and is judged between the peak value moment predicted and true peak
Deviation further the neural network model is trained when the deviation is unsatisfactory for preset decision condition.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910148244.7A CN109905868B (en) | 2019-02-27 | 2019-02-27 | Bluetooth communication prediction method and system for intelligent wearable device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910148244.7A CN109905868B (en) | 2019-02-27 | 2019-02-27 | Bluetooth communication prediction method and system for intelligent wearable device |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109905868A true CN109905868A (en) | 2019-06-18 |
| CN109905868B CN109905868B (en) | 2020-07-24 |
Family
ID=66945795
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910148244.7A Active CN109905868B (en) | 2019-02-27 | 2019-02-27 | Bluetooth communication prediction method and system for intelligent wearable device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109905868B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114696925A (en) * | 2020-12-31 | 2022-07-01 | 华为技术有限公司 | Channel quality assessment method and related device |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104657792A (en) * | 2015-03-06 | 2015-05-27 | 中电海康集团有限公司 | Early warning method based on rule engine and intelligent prediction |
| EP3060935A1 (en) * | 2013-10-25 | 2016-08-31 | Cubic Corporation | Positioning system with wireless beacons that vary transmission power levels |
| CN106527664A (en) * | 2016-11-07 | 2017-03-22 | 东软集团股份有限公司 | Energy-saving control method and device for self-service terminal |
| CN107863099A (en) * | 2017-10-10 | 2018-03-30 | 成都启英泰伦科技有限公司 | A kind of new dual microphone speech detection and Enhancement Method |
| CN108199892A (en) * | 2018-01-12 | 2018-06-22 | 电子科技大学 | LoRa terminal maximum transmission rate dynamic prediction methods based on neural network |
| CN108964815A (en) * | 2018-07-30 | 2018-12-07 | 太原理工大学 | A kind of channel selection and bit rate adaptive approach based on BP neural network |
| WO2019012029A1 (en) * | 2017-07-12 | 2019-01-17 | Siemens Aktiengesellschaft | Method and system for deviation detection in sensor datasets |
| CN109308782A (en) * | 2018-08-29 | 2019-02-05 | 中国科学院信息工程研究所 | A kind of behavioral value method and system of target object |
-
2019
- 2019-02-27 CN CN201910148244.7A patent/CN109905868B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3060935A1 (en) * | 2013-10-25 | 2016-08-31 | Cubic Corporation | Positioning system with wireless beacons that vary transmission power levels |
| CN104657792A (en) * | 2015-03-06 | 2015-05-27 | 中电海康集团有限公司 | Early warning method based on rule engine and intelligent prediction |
| CN106527664A (en) * | 2016-11-07 | 2017-03-22 | 东软集团股份有限公司 | Energy-saving control method and device for self-service terminal |
| WO2019012029A1 (en) * | 2017-07-12 | 2019-01-17 | Siemens Aktiengesellschaft | Method and system for deviation detection in sensor datasets |
| CN107863099A (en) * | 2017-10-10 | 2018-03-30 | 成都启英泰伦科技有限公司 | A kind of new dual microphone speech detection and Enhancement Method |
| CN108199892A (en) * | 2018-01-12 | 2018-06-22 | 电子科技大学 | LoRa terminal maximum transmission rate dynamic prediction methods based on neural network |
| CN108964815A (en) * | 2018-07-30 | 2018-12-07 | 太原理工大学 | A kind of channel selection and bit rate adaptive approach based on BP neural network |
| CN109308782A (en) * | 2018-08-29 | 2019-02-05 | 中国科学院信息工程研究所 | A kind of behavioral value method and system of target object |
Cited By (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114696925A (en) * | 2020-12-31 | 2022-07-01 | 华为技术有限公司 | Channel quality assessment method and related device |
| WO2022143724A1 (en) * | 2020-12-31 | 2022-07-07 | 华为技术有限公司 | Channel quality evaluation method and related apparatus |
| CN114696925B (en) * | 2020-12-31 | 2023-12-15 | 华为技术有限公司 | Channel quality assessment method and related device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109905868B (en) | 2020-07-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101459445B (en) | Cooperative spectrum sensing method in cognitive radio system | |
| Keally et al. | Watchdog: Confident event detection in heterogeneous sensor networks | |
| CN111314473A (en) | Environmental monitoring system based on artificial intelligence | |
| CN106990217B (en) | Sewage monitoring system based on wireless sensor network | |
| Wittenburg et al. | Cooperative event detection in wireless sensor networks | |
| CN108322891A (en) | Region congestion recognition methods based on user mobile phone signaling data | |
| CN104200667A (en) | Traffic jam grade detection method based on cellphone signal data | |
| CN102869079A (en) | Method for adjusting regular sleep period of terminal node in self-adaptive manner | |
| CN103152745A (en) | Method of locating mobile node with strong adaptivity | |
| CN109525956A (en) | The energy-efficient method of data capture of sub-clustering in wireless sense network based on data-driven | |
| CN107426680A (en) | Towards the wireless sensor network data collection system of building monitoring | |
| CN107592621A (en) | Localization method and device | |
| CN104811991A (en) | Wireless link quality predicting method based on dynamic time warping algorithm | |
| CN105848187A (en) | Wireless network link quality evaluation method based on signal-to-noise ration threshold adaptation | |
| Xue et al. | Prediction-based protocol for mobile target tracking in wireless sensor networks | |
| CN110167114A (en) | A kind of underwater acoustic communication waking up nodes signal detecting method based on frame synchronizing signal | |
| CN105611626A (en) | Target tracking algorithm based on regional clustering of perceptron in wireless sensor network | |
| CN109905868A (en) | A kind of intelligence wearable device Bluetooth communication prediction technique and system | |
| CN106443164A (en) | Transient state event and steady state event monitoring method during electric energy classification metering process | |
| CN101237357B (en) | Online failure detection method for industrial wireless sensor network | |
| CN108494511A (en) | A kind of dynamic arrival frequency spectrum sensing method based on absolute value accumulation | |
| CN105721083A (en) | Frequency spectrum detection method based on autocorrelation energy | |
| CN114757237B (en) | Speed-independent gait recognition method based on WiFi signals | |
| CN101212236B (en) | TD-SCDMA terminal and interfered cell information detection and channel estimation method | |
| CN106093909A (en) | It is applicable to the multi-hop CFAR information fusion decision method of cluster structured RSN |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |