CN110505597A - A kind of data transmission method of wireless sensor network - Google Patents
A kind of data transmission method of wireless sensor network Download PDFInfo
- Publication number
- CN110505597A CN110505597A CN201910701024.2A CN201910701024A CN110505597A CN 110505597 A CN110505597 A CN 110505597A CN 201910701024 A CN201910701024 A CN 201910701024A CN 110505597 A CN110505597 A CN 110505597A
- Authority
- CN
- China
- Prior art keywords
- data
- node
- module
- sensor node
- sensed data
- 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.)
- Pending
Links
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W28/00—Network traffic management; Network resource management
- H04W28/02—Traffic management, e.g. flow control or congestion control
- H04W28/021—Traffic management, e.g. flow control or congestion control in wireless networks with changing topologies, e.g. ad-hoc networks
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W28/00—Network traffic management; Network resource management
- H04W28/02—Traffic management, e.g. flow control or congestion control
- H04W28/06—Optimizing the usage of the radio link, e.g. header compression, information sizing, discarding information
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W4/00—Services specially adapted for wireless communication networks; Facilities therefor
- H04W4/30—Services specially adapted for particular environments, situations or purposes
- H04W4/38—Services specially adapted for particular environments, situations or purposes for collecting sensor information
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04W—WIRELESS COMMUNICATION NETWORKS
- H04W84/00—Network topologies
- H04W84/18—Self-organising networks, e.g. ad-hoc networks or sensor networks
Landscapes
- Engineering & Computer Science (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Mobile Radio Communication Systems (AREA)
- Arrangements For Transmission Of Measured Signals (AREA)
Abstract
This application discloses a kind of data transmission methods of wireless sensor network, it include: the data prediction model that sensor node receives the training of mist calculate node, acquire presently sensed data, the predicted value of presently sensed data is determined according to sense of history measured data, if the predicted value of presently sensed data and the difference of actual value are less than the first threshold of setting, then the sensor node does not send presently sensed data to the mist calculate node, otherwise, the sensor node sends presently sensed data to the mist calculate node；The mist calculate node is since the acquisition moment of presently sensed data, if receiving the presently sensed data within the waiting time of setting, the presently sensed data that then sensor node is sent carry out subsequent processing, if not receiving presently sensed data within the waiting time, the predicted value of presently sensed data is then determined according to sense of history measured data, and carries out subsequent processing for the predicted value as presently sensed data.By the application, the mass data transfers in wireless sensor network can be effectively reduced, and reduce the data processing load of core net.
Description
Technical field
This application involves wireless sensor network technology, in particular to a kind of transmission side data of wireless sensor network
Method.
Background technique
Wireless sensor network is a kind of multi-hop wireless network of self-organizing, is made of many static and movable sensors,
Have many advantages, such as that detection accuracy height, zmodem, coverage area are big.But the finite energy of sensor node and be difficult to supplement,
According to statistics, up to 80% total power consumption derives from wireless data transmission in wireless sensor network.Simultaneously as data
The periodicity of acquisition and the characteristic of wireless sensor network interior joint redundant deployment, lead to that a large amount of data can be generated in network
It flows, also will receive the data of overlapping at aggregation node, lead to the waste of energy and reduce the service life of wireless sensor network.Especially
It is when the Number dynamics of node deployment are flexible (scalability of system), problem can be more serious.
Existing wireless sensor network data compression and acquisition technique type are a lot of, such as based on compressed sensing, adjustment
The technologies such as sensor node dormancy time and sensing data prediction, but the real availability of most of compress techniques is not strong enough, does not have
There is the complexity for considering actual deployment scene, it is difficult to adapt to the data processing scene of huge wireless sensor network and magnanimity.
Wireless sensor network has been widely used for intelligent city, and the related fieldss such as safety monitoring and environment measuring are
Following 5G epoch essential a part, it is necessary first to which solution is exactly efficient and actually available wireless sensor network
The method and system of Data compression transmission.
And in the current situation, by the powerful computing capability of cloud computing, current most of Internet of Things solutions all bases
It in classic client-server framework, is analyzed in cloud back-end realization data, the service such as visualization and data storage.But this side
There are certain limitations for method, if such as the data of sensor node are transmitted to the Cloud Server far from edge network, then carry out
The analysis detection of data will cause and be difficult to endure delay, and the sensing data of the redundancy of wireless sensor network can aggravate core
The burden of heart net, to be difficult to dispose wireless sensor network on a large scale.
Summary of the invention
The application provides a kind of data transmission method of wireless sensor network, can effectively reduce wireless sensor network
In mass data transfers, and reduce the data processing load of core net.
To achieve the above object, the application adopts the following technical scheme that
A kind of data transmission method of wireless sensor network, in the leader cluster node and cloud computing of wireless sensor network
The sensing data that sensor node is collected are sent to the mist and calculated by middle addition mist calculate node between the heart, the leader cluster node
Node, the mist calculate node are sent to the cloud computing center, the data transmission after being compressed the sensing data
Method includes:
Sensor node receives the data prediction model of mist calculate node training by leader cluster node, acquires presently sensed number
According to according to the data prediction model, the predicted value of presently sensed data being determined according to sense of history measured data, if presently sensed number
According to predicted value and actual value difference be less than setting first threshold, then the sensor node do not pass through the leader cluster node to
The mist calculate node sends presently sensed data, and otherwise, the sensor node passes through the leader cluster node to the mist meter
Operator node sends presently sensed data；
The mist calculate node is since the acquisition moment of presently sensed data, if receiving institute within the waiting time of setting
Presently sensed data are stated, then the presently sensed data that sensor node is sent carry out subsequent processing, if within the waiting time
Presently sensed data are not received, then according to trained data prediction model, presently sensed number is determined according to sense of history measured data
According to predicted value, and using the predicted value as presently sensed data carry out subsequent processing.
Preferably, this method further comprises: the mist calculate node calculate the prediction of nearest continuous N time data acquisition at
Power re-starts the update of the data prediction model, and lead to when second threshold of the success rate prediction lower than setting
Cross the parameter that leader cluster node updates the data prediction model to sensor node；
Wherein, if the mist calculate node does not receive presently sensed data within the waiting time, record is current
Secondary data acquisition is predicts successfully, if the mist calculate node receives the presently sensed data within the waiting time,
Then current data acquisition of record is unsuccessful to predict；M is preset natural number.
3, the method according to claim 1, wherein being independently arranged waiting for different sensor nodes
Time；
According in each sensor node to the data transmission period of mist calculate node, sensor node and mist calculate node into
Line number it is predicted that time, setting sensor node corresponding waiting time.
Preferably, determining that the mode of the waiting time includes:
For sensor node before receiving the data prediction model, all the sensors node sends n to mist calculate node
The transmission time of actual sensed data, the actual sensed data sent according to sensor node each in same cluster determines the cluster
The corresponding waiting time initial value of middle all the sensors node；
The mist calculate node calculates this heartbeat letter after the heartbeat message for receiving the transmission of any sensor node every time
The transmission time of breath, updating the waiting time is
The mist calculate node calculates this data biography when receiving the sensing data of any sensor node transmission every time
The defeated timeIf the mist calculate node not yet sends current data, the mist meter when receiving the sensing data
Operator node updates any sensor node corresponding waiting time and is It is no
Then, the waiting time of all the sensors node is in cluster where the mist calculate node updates any sensor node
Wherein, α is preset Adaptation factor, 0 < α < 1, tnewFor the time index of nearest received actual sensed data；h
For the index of cluster where the sensor node, K is all the sensors number of nodes where any sensor node in cluster
Amount, k are index of any sensor node where it in cluster, and i is the index of actual sensed data, tiFor actual sensed data
Time index,I-th of the actual sensed data in n actual sensed data sent for k-th of sensor node in cluster
Transmission time.
Preferably, the transmission time of the actual sensed data sent according to all the sensors node in same cluster determines the cluster
The mode of the corresponding waiting time initial value of middle all the sensors node includes:
Calculate the transmission time for the actual sensed data that all the sensors node is sentAverage value
Calculate the distance between transmission time and the average value of each actual sensed data
The smallest preceding λ * n*K corresponding transmission times of distance DIS [j] are selected from all distancesIt calculates average
Value, using obtained average value as the waiting time initial value；
Wherein, λ is preset scale factor, 0 < λ < 1.
Sensor node equipment in a kind of wireless sensor network, comprising: sensing data acquisition module, model parameter are same
Walk module, data prediction module, data comparison module and sensing data transmission blocks；
The sensing data acquisition module, for periodically acquisition sensing data；
The model parameter synchronization module, the data for receiving the training of mist calculate node equipment by leader cluster node are predicted
Model；
The data prediction module, for determining current sense according to the data prediction model according to sense of history measured data
The predicted value of measured data；
The data comparison module, for from sensing data acquisition module sum number it is predicted that obtaining presently sensed number in module
According to actual value and predicted value, if the difference of actual value and predicted value be less than setting first threshold, notify the sensing data
Sending module does not send data, otherwise by the presently sensed data transmission of the sensing data collecting module collected to sensing data
Sending module；
The sensing data transmission blocks, for sending the presently sensed data to mist calculate node equipment.
Mist calculate node equipment in a kind of wireless sensor network, comprising: prediction model training module, model parameter are same
Walk module, sensing data reception module, data latency module and data uploading module；
The prediction model training module, for being based on data according to each sensor node sense of history measured data sequence
Time autocorrelation training data prediction model；
The model parameter synchronization module, for by the parameter synchronization of the data prediction model to corresponding sensor section
Point device；
The sensing data reception module, the presently sensed number sent for receiving sensor node by leader cluster node
According to；
The data latency module, for since the acquisition time of sensor node equipment, when waiting waiting to be set
Between after, if receiving the presently sensed data not over the sensing data reception module, from the data predict mould
Type training module obtains the predicted value of presently sensed data, and is sent to data uploading module；Otherwise, the sensing data are connect
It receives the presently sensed data that module receives and is sent to uploading module；
Data uploading module, for by presently sensed data or its predicted value carry out next step data compression process and/or
It is uploaded to cloud computing center.
Preferably, the equipment further includes waiting for time update module；
The sensing data reception module, the heartbeat sent for receiving the sensor node equipment by leader cluster node
Information, and the communications time of the received data of every deuterzooid module or heartbeat message is calculated, for calculating and more new sensor
The node device corresponding waiting time；
The waiting time update module, the transmission communicated every time for being determined according to the sensing data reception module
Time, using the variable thresholding strategy of environment self-adaption come in this equipment when the waiting corresponding with each sensor node equipment
Between.
As seen from the above technical solution, in the application, wireless sensor network leader cluster node and cloud computing center it
Between middle addition mist calculate node.Sensor node receives the data prediction model of mist calculate node training by leader cluster node, adopts
Collect presently sensed data, according to data prediction model, the predicted value of presently sensed data is determined according to sense of history measured data, if working as
It is preceding sensing data predicted value and actual value difference be less than set first threshold, then sensor node do not pass through leader cluster node to
Mist calculate node sends presently sensed data, and otherwise, sensor node is worked as by the leader cluster node to the transmission of mist calculate node
Preceding sensing data.Mist calculate node determines presently sensed data according to trained data prediction model, according to sense of history measured data
Predicted value, and the presently sensed data that the sensor node received in the waiting time of setting is sent are sent to cloud computing
The predicted value of presently sensed data is sent to cloud computing if not receiving presently sensed data within the waiting time by center
Center.By above-mentioned processing, on the one hand can be subtracted in the basic handling of network edge progress data by introducing mist calculate node
The processing load of few core net, reduces the time delay of data processing；On the other hand the number of wireless transmission side is reduced using prediction model
According to flow, extend network life；As a result, the application can effectively reduce the biography of the mass data in wireless sensor network
It is defeated, and reduce the data processing load of core net.
Detailed description of the invention
Fig. 1 is the network architecture schematic diagram of the wireless sensor data collection system calculated in the application based on mist；
Fig. 2 is the basic procedure schematic diagram of data transmission method in the application；
Fig. 3 is the interaction flow schematic diagram of mist calculate node and sensor node in the application data transmission method；
Fig. 4 is the corresponding waiting time schematic diagram of sensor node；
Fig. 5 is the structural schematic diagram of mist calculate node equipment and sensor node equipment in the application；
Fig. 6 is the interaction schematic diagram of each module in sensor node equipment；
Fig. 7 is the interaction schematic diagram of each module in mist calculate node equipment.
Specific embodiment
In order to which the purpose, technological means and advantage of the application is more clearly understood, the application is done below in conjunction with attached drawing
It is further described.
In the application, under the inspiration of mist computing technique, mist calculate node is introduced in wireless sensor network, utilizes mist
The features such as calculate node geographical distribution is wide, close to network edge and location aware carries out a part of data processing in network edge,
Reduce the processing load of core net.Specifically, the application provides a kind of wireless sensor network structure calculated based on mist, is passing
It unites middle addition mist calculate node between the leader cluster node and cloud computing center of wireless sensor network, leader cluster node is by sensor section
The sensing data that point is collected are sent to mist calculate node, and mist calculate node will be sensed after data are compressed and is sent in cloud computing
The heart.
It is further preferred that considering the complexity of actual deployment environment, the present invention also proposes a kind of environment self-adaption
Variable thresholding strategy, for the real-time selected of threshold parameter specific in model, so that model dynamic adapts to complicated sensor
Network environment.
Fig. 1 is the network architecture diagram of the large-scale wireless sensor data collection system calculated in the application based on mist.Such as
Shown in Fig. 1, the network architecture can divide three layers, be data sensing layer, mist computation layer and cloud central core respectively, wherein including sensor section
Point, leader cluster node, mist calculate node and cloud computing center.
It is multiple clusters by the sensor tissue in region in data sensing layer, sensor node is responsible for sampling ambient enviroment
Sensing data, be sent to leader cluster node.Leader cluster node and numerous sensor node direct communications, each sensor node from cluster
Data are collected, simple process data simultaneously will sense data transmission to mist calculate node.In order to keep sensor node energy consumption equal
Weighing apparatus is extended whole network life cycle, is chosen leader cluster node by the way of poll in cluster.
In mist computation layer, each mist calculate node is by the sensing data in leader cluster node collecting zone, then when carrying out
Cloud computing center is sent to after null Context compression.Wherein, mist calculate node utilizes the data prediction model pre-established, according to
The sensing data at the sense of history measured data prediction current time of sensor node, if not receiving biography within the waiting time of setting
The sensing data that sensor node is sent, then it is assumed that predict successfully, replace actual sensed data to be handled using prediction data；Such as
Fruit receives the sensing data that sensor node is sent within the waiting time of setting, then it is assumed that prediction of failure, utilization are received
Actual sensed data carry out subsequent processing.
In cloud central core, cloud computing center, which is collected, comes from the compressed time series data of mist calculate node, and to compressed
Data carry out storage and data visualize, and using stronger computing capability, more complicated algorithm comes intelligent excavating and analysis
Potential rule in data.
Fig. 2 is the basic procedure schematic diagram of data transmission method in the application.Wherein, with mist calculate node and a sensing
It is illustrated for interaction between device node, the processing of other sensors node is identical.Based on above-mentioned network knot shown in FIG. 1
Structure, as shown in Fig. 2, the data transmission method of the application includes:
Step 201, mist calculate node training data prediction model.
For sensor node when carrying out continuous sampling, the data of acquisition constitute a time series sequentially in time.It is logical
The autocorrelation analysis for crossing time series can obtain, and physical environment variation often has continuity, therefore same node is for a period of time
The data of interior acquisition temporal correlation with higher.So can use limited historical data to predict following a period of time
Measured value.For each sensor node, the corresponding data prediction model of the sensor node is individually trained in mist calculating.
Specifically, the sense of history measured data that mist calculate node transmits in advance according to certain sensor establishes data prediction model.
There are mainly two types of common prediction models: one is linear models, such as regression analysis, sliding average, exponential smoothing；It is another
It is nonlinear model, such as neural network, gray system equiprobability model.In view of sensor node computing capability and resource have
Limit, the application predict walking for sense of reality measured data using common autoregressive prediction AR (n) model in sensor network research
To.
The example of a Self-regression Forecast Model and its foundation is given below.
In sensing data sequence, node senses the set of data, i.e. TS=with some sample frequency continuous acquisition n
{(t1,x1),(t2,x2),……(ti,xi),……(tn,xn)}.Wherein, tiIndicate that data acquire moment, xiIndicate the sense of acquisition
Measured data.Assuming that x1,x2……xt-1With xtEstimated valueJust like offline sexual intercourse:
Wherein β1,β2……βt-nIt is parameter to be calculated when establishing prediction model.It is often in distress in actual measurement environment
With the error observed, therefore white noise ε is added in above-mentioned formulat, it may be assumed that
Wherein εt~NID (0, σ2), βiFor auto-regressive parameter, σ2For εtVariance.
The parameter of above-mentioned AR (n) model is calculated using least-squares estimation.By { xtSequence brings above formula into, use matrix form
It indicates are as follows: y=x β+ε, in which:
Therefore, the least-squares estimation of parameter beta is
Step 202, sensor node receives the data prediction model of mist calculate node training by leader cluster node.
The data prediction model parameter that training obtains is issued corresponding sensor section by leader cluster node by mist calculate node
Point.
In the processing of the application, the data transmission between mist calculate node and sensor node is all complete by leader cluster node
At, for description it is easy for the sake of, be just directly known as mist in the description that interact between mist calculate node and sensor below and calculate to save
Point is sent to sensor node or sensor node is sent to mist calculate node, is not repeated that description is complete by leader cluster node
At.
Step 203, sensor node acquires presently sensed data, according to data prediction model, according to sense of history measured data
It determines the predicted value of presently sensed data, and calculates the predicted value of sensing data and the difference of actual value.
The data prediction model parameter that sensor node is sent according to mist calculate node determines data prediction model, then,
The predicted value of current acquisition moment sensing data is determined according to historical dataMeanwhile in the current sense of currently acquisition moment acquisition
The actual value x of measured datat.Predicted value and actual value are compared in following step.
Specifically, sensor node carries out the prediction of sensing data using formula (2), and calculates predicted value and actual value
Error of fitting is
Step 204, judge whether the difference between the predicted value and actual value of current time sensing data is less than setting
Otherwise first threshold δ, executes step 206 if so, thening follow the steps 205.
Work as difftWhen < δ, sensor node does not upload sensing data x to mist calculate nodet；Work as difftWhen >=δ, sensing
Device node uploads sensing data x to mist calculate nodet。
Step 205, sensor node does not upload sensing data x to mist calculate nodet, and execute step 207.
Step 206, sensor node uploads sensing data x to mist calculate nodet, and execute step 207.
Step 207, mist calculate node judge sensor node acquire the moment start after the setting waiting time in whether
The sensing data that sensor node is sent are received, if so, 208 are thened follow the steps, it is no to then follow the steps 209.
Step 208, the sensing data received are carried out subsequent processing by this prediction of failure, mist calculate node.
If mist calculate node receives sensing data in the stipulated time, show sensor node determine predicted value with
The difference of actual value has been more than first threshold δ, then prediction accuracy is just unsatisfactory for system requirements, then determines this prediction of failure.
Step 209, this is predicted successfully, and mist calculate node trigger data prediction model generates predicted value, and by the prediction
Value carries out subsequent processing as when previous sensing data.
If mist calculate node does not receive sensing data in the stipulated time, then it is assumed that the predicted value that sensor node determines
And the difference of actual value is less than first threshold δ, then prediction accuracy just meets system requirements, then determines this time to predict successfully.
Step 210, mist calculate node calculates the success rate prediction of nearest continuous N time data acquisition, and judges the success rate
Whether lower than the second threshold set, if so, thening follow the steps 211, otherwise return step 203, carry out data next time and acquire.
Step 211, the update of data prediction model is carried out, and is predicted by leader cluster node to sensor node more new data
The parameter of model, return step 201.
As second threshold σ of the success rate prediction lower than setting, then show that current data prediction model model at this time cannot
The trend for continuing accurate prediction sensing data, needs to be updated data prediction model, return step 201, from sensor
Nearest n actual measured value is synchronized, re-establishes data prediction model, and send updated data prediction to sensor node
Model parameter.
As second threshold σ of the success rate prediction lower than setting, then show that current data prediction model model at this time can be with
The trend for continuing accurate prediction sensing data, does not need to update data prediction model, return step 203 continues next number
According to acquisition.
Wherein, when calculating success rate prediction, it is preferable that diff can be recorded after each determination is predicted successfullyt=0.So
Afterwards, it calculates success rate prediction and calculates P (difft=0).
So far, the basic procedure of data transmission method terminates in the application.Fig. 3 is mist calculate node in above-mentioned transmission method
With the interaction flow schematic diagram of sensor node.
In above-mentioned Fig. 2 and process shown in Fig. 3, the waiting time ω set in mist calculate node determines two predictions
Model data synchronous opportunity is to guarantee that the processing steps such as follow-up data fusion, distribution are correct and the key factor of quick execution.
Suitable waiting time ω needs to guarantee when mist calculate node is in the actual measured value x for receiving t moment sensortWhen, mist meter
Operator node is not yet by predicted valueIt is transmitted to next data processing step (such as be sent to cloud computing center etc.), otherwise
It is synchronous with data to need to carry out identical calculating again, waste calculating and the communication resource.
In true sensor network, different sensors node is different with the distance of mist calculate node, and sensor
Node is deployed in mostly under network bandwidth constrained environment, therefore different sensors node sends a message to mist meter in different moments
The transmission delay of operator node is not substantially identical.So in the application preferably, different sensor nodes can be corresponded to, it is independent
Corresponding waiting time ω is set.Meanwhile the space based on sensor node in the same cluster Yu mist calculate node communication distance
The dynamic change characterization of similitude and network communication bandwidth, the application it is further proposed that a kind of environment self-adaption it is variable etc.
To time parameter method.In prediction model initial phase, the spatial correlation Exception Filter value based on sensor node position, according to
The propagation delay time sequence of each sensor node determines the initial value of waiting time ω in the same cluster, then examines in subsequent heartbeat
The waiting time ω for surveying and dynamically updating each sensor node in data synchronization process according to time delay in network.Such as Fig. 4 institute
Show.
Specifically, in actual deployed environment, the selection of waiting time ω is dependent on k-th node in t moment cluster h
The propagation delay time of data packetWith the factors such as calculating duration of the prediction model in sensor node and mist calculate node.For
The data prediction model of k-th of sensor node, corresponding waiting time may be expressed as: in cluster h
Wherein,And CfogRespectively indicate the calculating time of prediction model in sensor node and mist calculate node, this with
The computing capability of node is related to model complexity.In synchronization prediction model of the invention, it is contemplated that the energy of sensor node
Consume limited, we have selected simple and effective linear prediction model, therefore sensor node and mist calculate node prediction model
It is negligible to calculate the time.Therefore, under the premise of without loss of generality, can approximatively think
In more detail, the specific choice mode of the waiting time of any sensor node can carry out as follows:
1) initial threshold is chosen
Prediction model initialize the synchrodata stage, each sensor node need to send n sense of reality measured data to
Basis at mist calculate node, as prediction model training.Based on the above process, k-th node in our available cluster h
Message transmission time delay sequenceBy all K sensors in combination cluster h
The propagation delay time sequence of node can obtain: First
The sequence of calculationThe average value of middle n*K propagation delay timeThen calculation delay sequenceIn each time delay
DataDifference between average valueAnd λ * n*K is minimum before selecting from DIS
Difference, by the corresponding delay data of corresponding differenceThe average value of sample is as in cluster when the waiting of all the sensors node
Between ω initial value.Wherein, λ is preset scale factor, 0 < λ < 1.
2) dynamically adapting network environment
If 2.1) receive the heartbeat message that sensor node is sent, the communication that mist calculate node calculates this heartbeat message is prolonged
The slow timeUpdate the waiting time be
Wherein α is Adaptation factor, represents the speed for adapting to environmental change.
If 2.2) receive the actual measured value for the sensing data that sensor node k is sent, mist calculate node calculates this reality
The communication delay of border measured valueAnd it is handled respectively by the following two kinds situation:
2.2.1) if sync fail, i.e., predicted value is transmitted to next data step by mist calculate node, then is represented
Network more congestion at this time.Using the spatial coherence of sensor node each in cluster, this situation can be fed back into cluster other
The waiting time of all the sensors node in sensor node, that is, synchronized update cluster are as follows:
2.2.2) if synchronizing success, i.e., mist calculate node does not generate predicted value also, then is updated using formula (5) and wait this single
The waiting time of sensor node.
Above-mentioned is the variable latency strategy of environment self-adaption in the application.
In addition, present invention also provides a kind of mists that can be used for implementing data transmission method in above-mentioned the application to calculate section
Point device and sensor node equipment, structure are as shown in Figure 5.
Wherein, mist calculate node equipment includes sensing data reception module, data latency module, the synchronous mould of model parameter
Block, prediction model training module, waiting time update module, cumulative errors memory module and data uploading module.Sensor section
Point device includes model parameter synchronization module, data prediction module, sensing data transmission blocks, data comparison module and sensing number
According to acquisition module.
1) mist calculate node equipment
A. data reception module is sensed: for receiving sense of reality measured data and heartbeat message from bottom sensor, together
When calculate the transmission time that communicates every time, for calculating the waiting time corresponding with more new sensor node；
B. waiting time update module: the transmission time communicated every time determined according to sensing data reception module utilizes
The variable thresholding strategy of environment self-adaption calculates in mist calculate node equipment the duration that synchronous prediction model needs to wait；
C. data latency module: after waiting waiting time to be set, if being received not over sensing data reception module
To true sensing data this moment, then it is assumed that prediction data this moment can be met the requirements, and will be obtained from prediction model training module
Prediction data is sent to data uploading module；If having received sense of reality measured data, true value is sent to uploading module.Together
When the error predicted every time is sent to cumulative errors memory module；
D. prediction model training module: according to each sensor node sense of history measured data sequence, the time based on data
Autocorrelation trains prediction model；
E. model parameter synchronization module: by the parameter synchronization of prediction model to corresponding sensor node equipment；
F. cumulative errors memory module: the prediction error that storage is calculated from data latency module, if accumulative prediction mistake
Difference is more than that certain threshold value or the successful probability of prediction are less than certain threshold value, then it is assumed that cannot continue accurately to predict sensing data
Trend, to trigger prediction model training module re -training model；
G. data uploading module: data are subjected to next step data compression process and are uploaded to cloud computing center.
2) sensor node equipment
A. data acquisition module is sensed: periodically acquisition sensing data；
B. newest prediction model parameters model parameter synchronization module: are obtained from mist calculate node equipment；
C. it data prediction module: according to the prediction model of mist calculate node equipment training, obtains and senses the pre- of data this moment
Measured value；
D. data comparison module: from sensing data acquisition module sum number it is predicted that obtaining true value this moment and prediction in module
Value does not send data, sensing value is otherwise sent to sensing data transmission blocks if prediction error is less than threshold value；
E. data transmission blocks are sensed: sending true sensing data to mist calculate node equipment.
3) module interaction
The application is using synchronous prediction model come from reducing bandwidth consumption in network, therefore mist meter from data source header as far as possible
Operator node equipment is disposed on corresponding sensor node equipment identical according to sense of history measured data training prediction model
Prediction model, to achieve the effect that synchronous prediction.
For sensor node equipment, the sensing data of sensing data acquisition module periodically acquisition ambient enviroment,
Data prediction module obtains prediction data according to model, and then data comparison module calculates the mistake of sensing data and prediction data
Difference, if error is less than in endurable threshold value, no need to send sensing data, otherwise, then need to send by sensing data
Module sends this moment really sensing data to mist calculate node equipment.Sensor node equipment passes through model parameter synchronization module
Newest prediction model parameters synchronous with mist calculate node equipment guarantee that synchronous forecasting mechanism operates normally.Sensor node is set
The interaction of each module is as shown in Figure 6 in standby.
In mist calculate node equipment, sensing data reception module is for receiving the sense of reality from sensor node equipment
Measured data, and the time delay of this time communication is obtained, when waiting time update module updates data latency according to real-time network state
Between.Data latency module is after waiting for a period of time, if obtaining true sensing this moment not over sensing data reception module
Data, then it is assumed that prediction data this moment can be met the requirements, and prediction data can be sent to uploading module and carry out next step data
Compression processing and it is uploaded to cloud computing center.The interaction of each module is as shown in Figure 7 in mist calculate node equipment.
Above-mentioned is the specific implementation of the application.By the processing of the application, on the one hand by introducing mist calculate node energy
Enough basic handlings that data are carried out in network edge, reduce the processing load of core net, reduce the time delay of data processing；Another party
The data traffic of wireless transmission side is reduced in face using prediction model, extends network life；As a result, the application can effectively subtract
Mass data transfers in few wireless sensor network, and reduce the data processing load of core net.
The foregoing is merely illustrative of the preferred embodiments of the present invention, is not intended to limit the invention, all in essence of the invention
Within mind and principle, any modification, equivalent substitution, improvement and etc. done be should be included within the scope of the present invention.
Claims (8)
1. a kind of data transmission method of wireless sensor network, which is characterized in that in the leader cluster node of wireless sensor network
The sensing data that sensor node is collected are sent to by middle addition mist calculate node, the leader cluster node between cloud computing center
The mist calculate node, the mist calculate node are sent to the cloud computing center, institute after being compressed the sensing data
Stating data transmission method includes:
Sensor node receives the data prediction model of mist calculate node training by leader cluster node, acquires presently sensed data,
According to the data prediction model, the predicted value of presently sensed data is determined according to sense of history measured data, if presently sensed data
Predicted value and actual value difference be less than setting first threshold, then the sensor node does not pass through the leader cluster node to institute
It states mist calculate node and sends presently sensed data, otherwise, the sensor node is calculated by the leader cluster node to the mist
Node sends presently sensed data；
The mist calculate node is since the acquisition moment of presently sensed data, if receiving described work as within the waiting time of setting
Preceding sensing data, then the presently sensed data that sensor node is sent carry out subsequent processing, if not connecing within the waiting time
Presently sensed data are received, then according to trained data prediction model, presently sensed data are determined according to sense of history measured data
Predicted value, and subsequent processing is carried out using the predicted value as presently sensed data.
2. the method according to claim 1, wherein this method further comprises: the mist calculate node calculates
The success rate prediction of nearest continuous N time data acquisition is re-started when second threshold of the success rate prediction lower than setting
The update of the data prediction model, and the parameter of the data prediction model is updated by leader cluster node to sensor node；
Wherein, if the mist calculate node does not receive presently sensed data within the waiting time, current number is recorded
According to acquisition to predict successfully, if the mist calculate node receives the presently sensed data within the waiting time, remember
Current data acquisition of record is unsuccessful to predict；M is preset natural number.
3. the method according to claim 1, wherein when being independently arranged waiting for different sensor nodes
Between；
It is counted according in each sensor node to the data transmission period of mist calculate node, sensor node and mist calculate node
It is predicted that time, setting sensor node corresponding waiting time.
4. according to the method described in claim 3, it is characterized in that, determining that the mode of the waiting time includes:
For sensor node before receiving the data prediction model, all the sensors node sends n reality to mist calculate node
Data are sensed, the transmission time of the actual sensed data sent according to sensor node each in same cluster determines institute in the cluster
There is the corresponding waiting time initial value of sensor node；
The mist calculate node calculates this heartbeat message after the heartbeat message for receiving the transmission of any sensor node every time
Transmission time, updating the waiting time is
The mist calculate node calculates the transmission of this data when receiving the sensing data of any sensor node transmission every time
TimeIf the mist calculate node not yet sends current data when receiving the sensing data, the mist calculates section
Point updates any sensor node corresponding waiting time and is Otherwise, institute
The waiting time for stating all the sensors node in cluster where mist calculate node updates any sensor node is
Wherein, α is preset Adaptation factor, 0 < α < 1, tnewFor the time index of nearest received actual sensed data；H is institute
The index of cluster where stating sensor node, K are all the sensors number of nodes where any sensor node in cluster, k
For index of any sensor node where it in cluster, i is the index of actual sensed data, tiFor actual sensed data when
Between index,The biography of i-th of the actual sensed data in n actual sensed data sent for k-th of sensor node in cluster
The defeated time.
5. according to the method described in claim 4, it is characterized in that, the reality sent according to all the sensors node in same cluster
The transmission time of sensing data determines that the mode of the corresponding waiting time initial value of all the sensors node in the cluster includes:
Calculate the transmission time for the actual sensed data that all the sensors node is sentAverage value AVG
Calculate the distance between transmission time and the average value of each actual sensed data
The smallest preceding λ * n*K corresponding transmission times of distance DIS [j] are selected from all distancesAverage value is calculated, it will
Obtained average value is as the waiting time initial value；
Wherein, λ is preset scale factor, 0 < λ < 1.
6. the sensor node equipment in a kind of wireless sensor network, which is characterized in that the equipment includes: sensing data acquisition
Module, model parameter synchronization module, data prediction module, data comparison module and sensing data transmission blocks；
The sensing data acquisition module, for periodically acquisition sensing data；
The model parameter synchronization module, the data for receiving the training of mist calculate node equipment by leader cluster node predict mould
Type；
The data prediction module, for determining presently sensed number according to the data prediction model according to sense of history measured data
According to predicted value；
The data comparison module, for from sensing data acquisition module sum number it is predicted that obtaining presently sensed data in module
Actual value and predicted value notify the sensing data to send if the difference of actual value and predicted value is less than the first threshold of setting
Module does not send data, otherwise sends the presently sensed data transmission of the sensing data collecting module collected to sensing data
Module；
The sensing data transmission blocks, for sending the presently sensed data to mist calculate node equipment.
7. the mist calculate node equipment in a kind of wireless sensor network, which is characterized in that the equipment includes: prediction model training
Module, model parameter synchronization module, sensing data reception module, data latency module and data uploading module；
The prediction model training module, for according to each sensor node sense of history measured data sequence, based on data when
Between autocorrelation training data prediction model；
The model parameter synchronization module, for setting the parameter synchronization of the data prediction model to corresponding sensor node
It is standby；
The sensing data reception module, the presently sensed data sent for receiving sensor node by leader cluster node；
The data latency module, after since the acquisition time of sensor node equipment, waiting waiting time to be set,
If receiving the presently sensed data not over the sensing data reception module, from data prediction model training
Module obtains the predicted value of presently sensed data, and is sent to data uploading module；Otherwise, by the sensing data reception module
The presently sensed data received are sent to uploading module；
Data uploading module, for presently sensed data or its predicted value to be carried out next step data compression process and/or upload
To cloud computing center.
8. equipment according to claim 7, which is characterized in that the equipment further includes waiting for time update module；
The sensing data reception module is believed for receiving the sensor node equipment by the heartbeat that leader cluster node is sent
Breath, and the communications time of the received data of every deuterzooid module or heartbeat message is calculated, for calculating and more new sensor section
The point device corresponding waiting time；
The waiting time update module, when for according to the determining transmission communicated every time of the sensing data reception module
Between, using the variable thresholding strategy of environment self-adaption come in this equipment when the waiting corresponding with each sensor node equipment
Between.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910701024.2A CN110505597A (en) | 2019-07-31 | 2019-07-31 | A kind of data transmission method of wireless sensor network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910701024.2A CN110505597A (en) | 2019-07-31 | 2019-07-31 | A kind of data transmission method of wireless sensor network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110505597A true CN110505597A (en) | 2019-11-26 |
Family
ID=68586891
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910701024.2A Pending CN110505597A (en) | 2019-07-31 | 2019-07-31 | A kind of data transmission method of wireless sensor network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110505597A (en) |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111417178A (en) * | 2020-03-17 | 2020-07-14 | 重庆大学 | Industrial internet energy-saving system and method based on machine learning and edge calculation |
| CN111417127A (en) * | 2020-03-25 | 2020-07-14 | 杭州一鸣惊人网络科技有限公司 | Sensor network formed by sensor nodes with 5G communication capability |
| CN111447583A (en) * | 2020-03-03 | 2020-07-24 | 中国海洋石油集团有限公司 | Time division multiplexing transmission method for riser monitoring system |
| CN113498071A (en) * | 2020-03-20 | 2021-10-12 | 大众汽车股份公司 | Method, apparatus and program for predicting future quality of service of a wireless communication link |
| CN113965565A (en) * | 2021-09-14 | 2022-01-21 | 南京融才交通科技研究院有限公司 | Internet of things networking method and system based on information age |
| CN114609329A (en) * | 2022-01-28 | 2022-06-10 | 西安电子科技大学 | Gas monitoring system based on sensor networking under industrial environment |
| CN115361363A (en) * | 2022-08-24 | 2022-11-18 | 杭州磐诚科技有限公司 | Edge calculation method of intelligent gateway |
| WO2023287331A1 (en) * | 2021-07-15 | 2023-01-19 | Telefonaktiebolaget Lm Ericsson (Publ) | Achieving uplink data compression using estimated environment models |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160182639A1 (en) * | 2014-12-17 | 2016-06-23 | University-Industry Cooperation Group Of Kyung-Hee University | Internet of things network system using fog computing network |
| CN106060148A (en) * | 2016-06-24 | 2016-10-26 | 华东师范大学 | Vehicle information secure collection method applicable to fog computing in intelligent traffic light system |
| CN106341842A (en) * | 2016-08-23 | 2017-01-18 | 北京邮电大学 | Method and device for transmitting data of wireless sensor network |
| CN107392398A (en) * | 2017-08-28 | 2017-11-24 | 济南浪潮高新科技投资发展有限公司 | A kind of agricultural management method, mist calculating platform and system |
| CN109379225A (en) * | 2018-10-22 | 2019-02-22 | 北京邮电大学 | A kind of data request processing method and apparatus between edge network |
-
2019
- 2019-07-31 CN CN201910701024.2A patent/CN110505597A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20160182639A1 (en) * | 2014-12-17 | 2016-06-23 | University-Industry Cooperation Group Of Kyung-Hee University | Internet of things network system using fog computing network |
| CN106060148A (en) * | 2016-06-24 | 2016-10-26 | 华东师范大学 | Vehicle information secure collection method applicable to fog computing in intelligent traffic light system |
| CN106341842A (en) * | 2016-08-23 | 2017-01-18 | 北京邮电大学 | Method and device for transmitting data of wireless sensor network |
| CN107392398A (en) * | 2017-08-28 | 2017-11-24 | 济南浪潮高新科技投资发展有限公司 | A kind of agricultural management method, mist calculating platform and system |
| CN109379225A (en) * | 2018-10-22 | 2019-02-22 | 北京邮电大学 | A kind of data request processing method and apparatus between edge network |
Non-Patent Citations (1)
| Title |
|---|
| HAO DENG 等: "Fog Computing Architecture-Based Data Reduction Scheme for WSN", 《2019 1ST INTERNATIONAL CONFERENCE ON INDUSTRIAL ARTIFICIAL INTELLIGENCE (IAI)》 * |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111447583A (en) * | 2020-03-03 | 2020-07-24 | 中国海洋石油集团有限公司 | Time division multiplexing transmission method for riser monitoring system |
| CN111447583B (en) * | 2020-03-03 | 2022-08-02 | 中国海洋石油集团有限公司 | Time division multiplexing transmission method for riser monitoring system |
| CN111417178A (en) * | 2020-03-17 | 2020-07-14 | 重庆大学 | Industrial internet energy-saving system and method based on machine learning and edge calculation |
| CN111417178B (en) * | 2020-03-17 | 2021-12-31 | 重庆大学 | Industrial internet energy-saving system and method based on machine learning and edge calculation |
| CN113498071A (en) * | 2020-03-20 | 2021-10-12 | 大众汽车股份公司 | Method, apparatus and program for predicting future quality of service of a wireless communication link |
| CN111417127A (en) * | 2020-03-25 | 2020-07-14 | 杭州一鸣惊人网络科技有限公司 | Sensor network formed by sensor nodes with 5G communication capability |
| WO2023287331A1 (en) * | 2021-07-15 | 2023-01-19 | Telefonaktiebolaget Lm Ericsson (Publ) | Achieving uplink data compression using estimated environment models |
| CN113965565A (en) * | 2021-09-14 | 2022-01-21 | 南京融才交通科技研究院有限公司 | Internet of things networking method and system based on information age |
| CN114609329A (en) * | 2022-01-28 | 2022-06-10 | 西安电子科技大学 | Gas monitoring system based on sensor networking under industrial environment |
| CN115361363A (en) * | 2022-08-24 | 2022-11-18 | 杭州磐诚科技有限公司 | Edge calculation method of intelligent gateway |
| CN115361363B (en) * | 2022-08-24 | 2023-08-11 | 杭州磐诚科技有限公司 | Edge computing method of intelligent gateway |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110505597A (en) | A kind of data transmission method of wireless sensor network | |
| Tayeh et al. | A distributed real-time data prediction and adaptive sensing approach for wireless sensor networks | |
| CN110390246A (en) | A kind of video analysis method in side cloud environment | |
| CN108399745A (en) | A kind of city road network trend prediction method at times based on unmanned plane | |
| CN101267446A (en) | Time domain data amalgamation method for wireless sensor network | |
| CN111722540A (en) | Energy Internet digital twin simulation system and method | |
| TWI393376B (en) | Method of data transmission | |
| CN105120217A (en) | Intelligent camera motion detection alarm system and method based on big data analysis and user feedback | |
| CN106686533B (en) | Wireless body area network data transmission method based on reinforcement learning algorithm | |
| CN112684301B (en) | Method and device for detecting power grid faults | |
| CN103079289A (en) | Wireless multi-person motion data collector and collection method | |
| CN104023356B (en) | A kind of wireless sensor network data transmission method towards facilities environment control | |
| CN104507096B (en) | The data transmission method and system of model-driven in wireless sense network | |
| CN109525956A (en) | The energy-efficient method of data capture of sub-clustering in wireless sense network based on data-driven | |
| CN108379818A (en) | A kind of technology analysis system and method for swimming exercise | |
| CN108334831A (en) | A kind of monitoring image processing method, monitoring terminal and system | |
| Wang et al. | Edge intelligence for mission cognitive wireless emergency networks | |
| CN109816978A (en) | Consider the region group system for traffic guiding and method of driver dynamic respondent behavior | |
| CN102118884B (en) | Data transmission method for opportunistic mobile sensor network based on closeness centrality | |
| CN108401235A (en) | A kind of agriculture site environment parameter intelligent acquisition processing system based on big data | |
| CN113904948B (en) | 5G network bandwidth prediction system and method based on cross-layer multidimensional parameters | |
| CN103338472B (en) | A kind of wireless network links quality estimation method | |
| Medeiros et al. | Swarm-based and energy-aware unmanned aerial vehicle system for video delivery of mobile objects | |
| CN114723480A (en) | Passenger flow prediction method and cargo scheduling system for rural tourism | |
| CN116436980A (en) | Real-time video task end network edge cooperative scheduling method and device |
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 | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20191126 |