CN115290130B - Distributed information estimation method based on multivariate probability quantization - Google Patents
Distributed information estimation method based on multivariate probability quantization Download PDFInfo
- Publication number
- CN115290130B CN115290130B CN202211229494.1A CN202211229494A CN115290130B CN 115290130 B CN115290130 B CN 115290130B CN 202211229494 A CN202211229494 A CN 202211229494A CN 115290130 B CN115290130 B CN 115290130B
- Authority
- CN
- China
- Prior art keywords
- probability
- quantization
- multivariate
- data
- function
- 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.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01D—MEASURING NOT SPECIALLY ADAPTED FOR A SPECIFIC VARIABLE; ARRANGEMENTS FOR MEASURING TWO OR MORE VARIABLES NOT COVERED IN A SINGLE OTHER SUBCLASS; TARIFF METERING APPARATUS; MEASURING OR TESTING NOT OTHERWISE PROVIDED FOR
- G01D21/00—Measuring or testing not otherwise provided for
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16Y—INFORMATION AND COMMUNICATION TECHNOLOGY SPECIALLY ADAPTED FOR THE INTERNET OF THINGS [IoT]
- G16Y20/00—Information sensed or collected by the things
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16Y—INFORMATION AND COMMUNICATION TECHNOLOGY SPECIALLY ADAPTED FOR THE INTERNET OF THINGS [IoT]
- G16Y20/00—Information sensed or collected by the things
- G16Y20/10—Information sensed or collected by the things relating to the environment, e.g. temperature; relating to location
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16Y—INFORMATION AND COMMUNICATION TECHNOLOGY SPECIALLY ADAPTED FOR THE INTERNET OF THINGS [IoT]
- G16Y40/00—IoT characterised by the purpose of the information processing
- G16Y40/10—Detection; Monitoring
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16Y—INFORMATION AND COMMUNICATION TECHNOLOGY SPECIALLY ADAPTED FOR THE INTERNET OF THINGS [IoT]
- G16Y40/00—IoT characterised by the purpose of the information processing
- G16Y40/20—Analytics; Diagnosis
-
- G—PHYSICS
- G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
- G16Y—INFORMATION AND COMMUNICATION TECHNOLOGY SPECIALLY ADAPTED FOR THE INTERNET OF THINGS [IoT]
- G16Y40/00—IoT characterised by the purpose of the information processing
- G16Y40/60—Positioning; Navigation
-
- 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
-
- 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
- Engineering & Computer Science (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Biomedical Technology (AREA)
- Environmental & Geological Engineering (AREA)
- Toxicology (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Computer Networks & Wireless Communication (AREA)
- Signal Processing (AREA)
- Arrangements For Transmission Of Measured Signals (AREA)
Abstract
The invention discloses a distributed information estimation method based on multivariate probability quantization, which comprises the following steps: s1, constructing a distributed information estimation scene: the system comprises a fusion center FC positioned in the center of a wireless communication network and a plurality of sensors distributed at the edge of the wireless communication network; s2, constructing a multivariate probability quantizer for quantizing local observation data by a sensor; s3, optimizing design parameters of multi-element quantization probability function(ii) a S4, designing a quantitative fusion estimator by a fusion center FC and optimizing to obtain an optimal estimation function(ii) a S5. Based onAndthe distributed information estimation is quantized with multivariate probability. The method can adapt to the situation that a quantization result has a plurality of elements and keep higher estimation performance.
Description
Technical Field
The invention relates to distributed information estimation, in particular to a distributed information estimation method based on multivariate probability quantization.
Background
Distributed information estimation based on quantized data has been an active area of research. In a typical distributed estimation framework, local sensors send local observation data of raw information to a fusion center. The fusion center receives data sent from different local sensors and estimates unknown original information by using an estimation algorithm. However, due to bandwidth/energy limitations, the local observed data on the sensors typically needs to be quantified before transmission to the fusion center. The use of the same quantizer for all sensors is a widely adopted solution because it simplifies the design issues.
However, many conventional solutions mainly consider the problem of quantizer optimization in an environment where ideally no observation noise exists. And further research into quantizer designs that take into account observed noise conditions is lacking. Furthermore, many performance analyses and theories regarding the optimal quantizer only consider the case of binary quantization, i.e., the length of the quantized data on the sensor is limited to 1 bit. For the case of quantizing the observation data into multi-bit data on the sensor, i.e. there is a plurality of possibilities for quantization results, the corresponding quantizer design scheme is also lack of research.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provide a distributed information estimation method based on multivariate probability quantization, which can adapt to the situation that a quantization result has multivariate and keep higher estimation performance.
The purpose of the invention is realized by the following technical scheme: a distributed information estimation method based on multivariate probability quantization comprises the following steps:
s1, constructing a distributed information estimation scene: the system comprises a fusion center FC positioned in the center of a wireless communication network and a plurality of sensors distributed at the edge of the wireless communication network;
each sensor is responsible for the raw information required by the fusion centerObserving to obtain own local observation data, performing multivariate probability quantization operation on the local observation data, converting continuous observation data into binary discrete data which can be used for digital communication, and sending the binary discrete data to a fusion center FC; the FC fusion center estimates the original information according to the quantized data sent by all the sensors;
S2, constructing a multivariate probability quantizer for quantizing local observation data by a sensor;
S4, designing a quantitative fusion estimator by a fusion center FC and optimizing to obtain an optimal estimation function;
The beneficial effects of the invention are: the multivariate quantization probability method still keeps the capability of approximately linearly decreasing with the total bit number under the condition that the total bit number of the network is changed, and the performance of the multivariate quantization probability method is efficiently estimated in the distributed wireless sensor network.
Drawings
FIG. 1 is a flow chart of a method of the present invention;
FIG. 2 is a schematic diagram of a distributed information estimation scenario;
FIG. 3 is a block diagram of a multivariate probability quantizer;
FIG. 4 is a diagram of a quantization function structure;
FIG. 5 is a block diagram of a quantized fusion estimator;
fig. 6 is a diagram illustrating MSE estimated by the network for the original information under the condition that the total quantization bit number of the entire network changes.
Detailed Description
The technical solutions of the present invention are further described in detail below with reference to the accompanying drawings, but the scope of the present invention is not limited to the following.
Aiming at the problem of information estimation based on a bandwidth/energy limited distributed wireless sensor in a future wireless communication network, the invention designs a distributed information estimation scheme based on multivariate probability quantization: the method comprises the steps of designing a multivariate probability quantizer positioned on a sensor and a corresponding multivariate quantization probability function optimization algorithm; and designing a quantitative fusion estimator positioned on the fusion center and a corresponding estimation function optimization algorithm. Consider a generalized scenario in which a distributed wireless sensor network estimates an unknown raw information, the network comprising a Fusion Center (FC) located at a hub node of the network and a plurality of sensors distributed at different locations at the edge of the network. The original information may be any kind of data required by the network, and is determined by the specific requirements of the network, such as common positioning information or weather information. Each sensor observes the original information and obtains its own local observation data, and usually, in an actual environment, due to the influence of environmental noise on the observation, an error exists between the local observation data and the original information. For a sensor with limited bandwidth/energy, the local observation data needs to be quantized first, and the continuous observation data is converted into binary discrete data which can be used for modern digital communication, so that the own observation data can be smoothly sent to the FC. FC can only estimate the raw information using the quantized data sent from all sensors. The measurement index of the estimation performance of the original information generally uses Mean Squared Error (MSE) of the original information and the estimation value thereof, and smaller MSE means more accurate estimation and better estimation performance;
as shown in fig. 1, a distributed information estimation method based on multivariate probability quantization includes the following steps:
s1, constructing a distributed information estimation scene: as shown in fig. 2, the system comprises a fusion center FC located at the center of the wireless communication network and a plurality of sensors distributed at the edge of the wireless communication network;
each sensor is responsible for the raw information required by the fusion centerTo carry outObserving to obtain own local observation data, performing multivariate probability quantization operation on the local observation data, converting continuous observation data into binary discrete data which can be used for digital communication, and sending the binary discrete data to a fusion center FC; the FC fusion center estimates original information according to the quantitative data sent by all the sensors;
s2, constructing a multivariate probability quantizer for quantizing local observation data by a sensor:
multiple sensors distributed at the edge of the network, together with the original information required by the fusion centerAnd observing to respectively obtain own local observation. When all sensor observations are considered here, the observation noise affected by the environment is independently and identically distributed. Therefore, to reduce the difficulty in sensor design and deployment, we also consider using the exact same multivariate probability quantizer structure on all sensors, including any adjustable design parameters on the quantizer.
Since independent and identically distributed observation noise on all sensors is considered and the same multivariate quantizer is used. We take an example of any sensor (ignoring sensor numbers) herein to describe the process of observation and data quantization on the sensor, and the structure, function and design scheme of the multivariate probability quantizer. As shown in FIG. 3, the sensor observes the raw informationObtaining local observationsBy usingTo represent the observed valueRelative to what is observedTo describe the randomness between them. The sensor obtains the observed valueThen inputting the data into a multi-element probability quantizer and outputting a final quantization resultQuantifying the resultsIs a one containsBinary data of bits. Inside the quantizer, the observed value of the inputIs first fed into a multivariate quantization probability control functionIs mapped to oneProbability vector of dimensionProbability vectorAll of the elements in (1) areThe interval takes a value while satisfying the sum of 1, i.e.
Wherein, the first and the second end of the pipe are connected with each other,is a function of design parametersA variable function of control having
WhereinComprises aAll of the parameters that can be adjusted in (c),is the number of design parameters. As can be seen from equation (2), by changing the design parametersBy changing the parameter function accordinglyThe functional expressions and structures of (1). FromOutput probability vectorThen fed into the quantization functionIn (a) to (b)Detailed below and in fig. 4), a decimal one-dimensional discrete value is outputThen we go through decimalQuantization result converted into binary. For quantization functionIts outputA is common toDifferent results, it is desirable to makeTaking the probability of each result entirely fromProbability vector of dimensionControl, i.e. effecting
WhereinRepresenting observations at a given input multivariate probability quantizerUnder the premise of (1), output is quantizedTake a value ofThe probability of (c).
To implement the above-described function of probability quantization of local observation data by the multivariate probability quantizer, we apply quantization function in the multivariate probability quantizerThe structure shown in fig. 4 is designed.
Wherein the function is quantizedIs inputted byProbability vector of dimensionThe output being a one-dimensional discrete valueIt is composed ofThe serial sublayers with the same structure are composed, and the specific structure functions are as follows:
Mth sublayer, M =1,2, …, M: input of mth sublayer (if M =1, its input isAnd) Is output from the previous sublayer (m-1 th sublayer)Vector of dimensionsAnd quantized value(ii) a First, in the m-th sublayer, to be inputtedDivided into two sub-vectors of equal length, each comprisingAll elements of the first half and all elements of the second half, i.e. twoSubvectors of dimensionAnd(ii) a Then, the mth sublayer utilizesAndoutputting the quantized valueIn which
Is [0,1]Random noise, function, evenly distributed over intervalsAn input non-negative number will output a1, whereas an input negative number will output a 0. Definition ofThe mth sublayer outputVector of dimensionsWherein
And (3) outputting: quantization functionIs output quantized valueIs that it isQuantized values output by the sub-layersI.e. by。
With the structure as in FIG. 4, the function is quantizedImplements a quantized value for its outputTaking the probability of each possible outcome entirely from the probability vectorTo control, the function in equation (3) is implemented.
Probability control function is quantized by adjusting the multivariate in the multivariate probability quantizer on the sensor as shown in equations (2) and (3)Design parameters ofCan be varied accordinglyAnd further changing the probability distribution of the quantized data on the sensor relative to the local observed data and the original information. This means that we can optimize the multivariate quantization probability function for the original information subject to different random distributions and the observation noise with different random characteristics under different observation environmentsDesign parameters ofTo obtain the optimal quantized data probability distribution adapted to the current environment. By using Bayesian estimation theory, we consider minimizing the channel by an algorithmAnd determining an estimated MSE lower bound which can be achieved by using quantized data in the fusion center after the sensor quantizes the local observation so as to find out the optimal design parameter suitable for the current observation environmentAnd corresponding use-optimized design parameters on the sensorIs optimized to a multivariate quantization probability function。
We assume a commonThe individual sensors are distributed throughout the network, a multivariate probability quantizer structure as shown in FIG. 3 is used on all sensors, and we use the exact same multivariate quantization probability function in all multivariate probability quantizersAnd corresponding design parametersSo as to reduce the design cost and difficulty of the whole network.Each sensor is used for respectively comparing the original informationObserving to obtain own local observation data with observation noise. We consider the presence of independent identically distributed observation noise on each sensor and define a probability density functionTo describe the distribution of observed noise: for the firstA sensor, its local observation dataIs quantized into a plurality of probability quantizers through a local multivariate probability quantizerBinary data of bitsAnd sent to the FC at the hub. Therefore, it is not only easy to useA sensor jointly generatesAnQuantized data of bitsAnd sent to the FC, which needs to estimate the original information using all the received quantized data.
Quantitative data on all sensors at a given time by Bayesian probability theoryAre independently and simultaneously distributed. Therefore, based on the probability distribution of the quantized data, first, the quantized data is calculated when the FC receives the quantized dataFor the original informationThe lower bound of the estimated MSE that the estimation can achieve, i.e. the lower bound
Wherein,Indicating that FC receives all the quantized data sent from the sensor,representing FC utilization quantized dataWhat can be achieved isRespectively, to the left of the inequality in the formulaRepresenting original informationAnd itEstimated valueThe MSE between the two is the MSE,represents the operation of calculating mathematical expectation, the right side of the inequality in the formula (6) represents the lower bound of the calculated MSE,is the number of combinations in the mathematical definition,
is formed by multi-component quantization of probability functionsDesign parameters ofAnd original informationA series of intermediate calculation terms for the decision. As can be seen from equation (6), in the original informationProbability distribution and observed noise distribution ofFC with both determinants utilizes quantized data pairsMSE estimated, i.e.Whose lower bound is fully defined by the multivariate quantization probability functionDesign parameters ofAnd (6) determining. Thus, the right side of the inequality in equation (6) is minimized algorithmically byThe determined FC uses the lower bound of the MSE estimated by the quantized data on the original information to find the optimal design parameter adapted to the current observation environmentAnd corresponding optimal multivariate quantization probability function。
Based on the above analysis, we consider an iterative algorithm that approximately solves the optimization problem for the design parameters in each iteration of the algorithm based on a series of samples of raw information collected from the actual observation environment and sensor local observation data, and gradually approaches the optimal design parameters during the iteration process。
Initialization: total number of sensors isSample set, Is the total number of samples contained in the sample set,is a sample of the original information that was,indicates the serial number of the sample, andrepresenting sensor versus raw information samplesTotal observationsObtained secondarily(ii) an observation sample; setting initial design parametersSetting the tolerance threshold of iteration toSetting an initial iteration count to。
The method comprises the following steps: in the first placeIn the second iteration, the pairs are obtained according to the formula (7)Definition of (2) toCalculating fromMultivariate quantization probability function design parameters obtained in sub-iterationA series of intermediate calculation items of the decisionWherein
Step two: using the equation (8)And sample setThe lower MSE bound on the right side of the inequality in equation (6) is approximately calculated as
Then using the interior point method and the gradient descent method, the following minimization problem is solved
If not, the explanation needs to continue iteration, and needs to beStep A2 is carried out to carry out the next iteration, and the iteration count is updated(ii) a If the convergence condition is established, outputtingAs an optimal design parameter.
The optimal design parameters can be obtained through the algorithmAnd corresponding to the optimal multivariate quantization probability function。
S4, designing a quantitative fusion estimator by a fusion center FC and optimizing to obtain an optimal estimation function;
As mentioned in the foregoing description,a sensor jointly generatesAnOf bitsAnd sent to the FC, and the FC needs to utilize all received quantized data to correct the original informationThe information is estimated. Therefore, we first present a quantitative fusion estimator design on FC.
As shown in fig. 5, the FC receives a slaveTransmitted from a sensorQuantized dataAnd inputting it into the quantization fusion estimator, and outputting the original informationIs estimated value of. In the quantitative fusion estimator, of the inputAnBinary quantized data of bitsIs first converted intoIs arranged atMean decimal discrete data;
Decimal dataThen the data is sent to an Onehot function for carrying out one-hot coding operation to obtain the corresponding dataOne-hot coded vector. To a first orderDecimal dataBy way of example only, it is possible to use,is its corresponding one-hot coded vector,is composed ofbinary data of bit length toIn total ofThe bits bit are numbered in sequence asBit, decimal dataThe value range of (1) is just the serial number of all bits, and the one-hot coding means that only one bit is codedTo (1)The bit will be set to 1 and all the rest of the bit bits will be set to 0, i.e. the bit is set to 1,
Then, the process of the present invention is carried out,one-hot coded vectorIs fed into an averager to obtain their mean vectors;
After thatIs then fed into the estimation functionIn the method, an estimated value of the original information is output,Is also a design parameterA function of control, i.e.
When the same multivariate probability quantizer as shown in FIG. 3 is used on all sensors, and with the same optimal design parametersOf the multivariate quantization probability functionThe quantized data sent by all sensors to the FC is compared to the original informationAre conditionally independent and identically distributed;
at the moment, based on Bayesian estimation theory and probability model, the estimation value of FC to the original information is calculatedWith the original informationMSE between
following the pair in equation (7)By quantifying the optimal design parameters of the probability functionAnd original informationA determined series of intermediate calculation terms;
the optimal design parameters of the probability function in the multivariate quantization are obtained from the formula (13)The MSE estimated on FC for the original information is determined entirely by the estimation functionVariable design parameters ofControlling;
s404, a series of samples based on the original information collected from the actual observation environment and the local observation data of the sensor, and an optimal multivariate quantization probability function on the sensorSolving the optimum estimation function design parameters that minimize the estimated MSE on FC。
Initialization: total number of input sensorsOptimal multivariate quantization probability function on sensorAnd corresponding optimum design parametersSample setFollowing the pair in equation (14)For any ofWill be composed ofAnd original information samplesIntermediate item of decisionIs approximately calculated as
Setting initial design parametersSetting the tolerance threshold of iteration toSetting an initial iteration count to。
The method comprises the following steps: in the first placeIn the second iteration, defineFor the estimated MSE at FC, using equation (13),and the design parameters obtained in the first iterationWill beIs approximately calculated as
Using the interior point method and the gradient descent method, the following minimization problem is solved
Step two: calculating and viewing convergence criteriaWhether the result is true or not; if not, continue iteration, willStep B2 is carried out to carry out the next iteration and the iteration count is updated(ii) a If the convergence condition is established, outputtingAs an optimal design parameter.
The optimal design parameters can be obtained through the algorithmAnd corresponding optimal estimation function。
Based on the two algorithms, the optimal multivariate quantization probability functions on all the sensors are obtained respectivelyAnd the best estimation function on FC. Next, we will briefly describeThe whole process of estimating the original information by the whole network. The functional structures of the multivariate probability quantizer on the sensor and the quantized fusion estimator on the FC are described in detail above, and are not described herein again.
First of all, the first step is to,individual sensor pair raw informationThe observation was performed separately. To a first orderAn example of a sensor, itObtaining own local observation data after observationAnd will beInto a multivariate probability quantizer as shown in FIG. 3 (note that at this time, the multivariate quantization probability function in FIG. 3Has been optimizedReplacement), and finally outputbinary quantization data of bitIs sent to the FC. All ofAll sensors shareGenerate aQuantized data. FC receiving signals from all sensorsQuantizes the data and feeds them into the quantized fusion estimator shown in fig. 5 (note also that at this point, the estimation function in fig. 5Has been optimizedReplacement), and finally outputs an estimate of the original information。
In the embodiment of the present application, considering the estimation performance of the proposed distributed information estimation method based on multivariate probability quantization on original information in a practical environment, as described above, we use the Mean Square Error (MSE) of the original information and its estimation value as an evaluation criterion, and a smaller MSE indicates better estimation performance. Specifically, we have tested the MSE performance of the network on the original information estimate when the total quantized bit number of the entire network changes and compared it with the current optimal binary quantization SQMLF method and the lower bound of the theoretical minimum MSE that can be reached on the original information estimate under binary quantization (one-bit quantization). As can be seen from fig. 6, although the SQMLF method has almost completely approximated the lower MSE bound for the original information at any time with the constraint of binary quantization, the estimated MSE for the original information by the distributed wireless sensor network using the multivariate probability quantization method is much smaller than both, which means that the multivariate probability quantization method we propose is superior to any binary quantization method without considering the constraint on the number of bits of quantized data on the sensor. Furthermore, it can be observed that the multivariate quantization probability approach still maintains the ability to decrease approximately linearly with the total number of quantized bits, as the total number of bits of the network changes. The method verifies the efficient estimation performance of the multivariate probability quantification method in the distributed wireless sensor network and the adaptability and the expansibility of the total quantification bit number dynamic change of the network in the actual environment.
While the foregoing description shows and describes a preferred embodiment of the invention, it is to be understood, as noted above, that the invention is not limited to the form disclosed herein, but is not intended to be exhaustive or to exclude other embodiments and may be used in various other combinations, modifications, and environments and may be modified within the scope of the inventive concept described herein by the above teachings or the skill or knowledge of the relevant art. And that modifications and variations may be effected by those skilled in the art without departing from the spirit and scope of the invention as defined by the appended claims.
Claims (9)
1. A distributed information estimation method based on multivariate probability quantization is characterized in that: the method comprises the following steps:
s1, constructing a distributed information estimation scene: the system comprises a fusion center FC positioned in the center of a wireless communication network and a plurality of sensors distributed at the edge of the wireless communication network;
each sensor is responsible for the raw information required by the fusion centerObserving to obtain local observation data of the user, performing multivariate probability quantization operation on the local observation data, converting continuous observation data into binary discrete data which can be used for digital communication, and sending the binary discrete data to a fusion center FC; the FC fusion center estimates original information according to the quantitative data sent by all the sensors;
s2, constructing a multivariate probability quantizer for quantizing local observation data by a sensor;
the step S2 includes:
s201, observing original information by a sensorObtaining local observationsBy usingTo express the observed valueRelative to what is observedTo describe randomness therebetween;
s202, obtaining an observed value by a sensorThen inputting the data into a multi-element probability quantizer and outputting a final quantization resultQuantifying the resultsIs a one containsBinary data of bits:
inside the quantizer, the observed value of the inputIs first fed into a multivariate quantization probability functionIs mapped to oneProbability vector of dimensionProbability vectorAll of the elements in (1) areThe interval takes a value while satisfying the sum of 1, i.e.
Wherein the content of the first and second substances,is a function of design parametersA variable function of control having
WhereinComprises aAll of the parameters that can be adjusted in (c),is the number of design parameters; by varying design parametersBy changing the parameter function accordinglyThe functional expressions and structures of (a);
fromOutput probability vectorThen fed into the quantization functionIn the method, a decimal one-dimensional discrete value is outputThen we go through decimalQuantization result converted into binary(ii) a For quantization functionIts outputA share ofDifferent results, it is desirable to makeTaking each kind of knotThe probability of the fruit is totally composed ofProbability vector of dimensionControl, i.e. effecting
WhereinRepresenting observations at a given input multivariate probability quantizerUnder the premise of (1), output is quantizedTake a value ofThe probability of (d);
S4, designing a quantitative fusion estimator by a fusion center FC and optimizing to obtain an optimal estimation function;
2. The distributed information estimation method based on multivariate probability quantization as claimed in claim 1, characterized in that: when each sensor observes the original information, the observation noises influenced by the environment are independently and identically distributed, and all the sensors use the same multivariate probability quantizer structure.
3. The distributed information estimation method based on multivariate probability quantization as claimed in claim 1, characterized in that: the quantization functionIs inputted byProbability vector of dimensionThe output being a one-dimensional discrete valueIt is composed ofThe serial sublayers with the same structure are composed, and the specific structure functions are as follows:
The input of the mth sublayer is the output of the previous sublayer, M =1,2, …, MVector of dimensionsAnd quantized value(ii) a First, in the m-th sublayer, to be inputtedDivided into two sub-vectors of equal length, each containingAll elements of the first half and all elements of the second half, i.e. twoSubvectors of dimensionAnd(ii) a Then, the m sub-layer utilizesAndoutputting the quantized valueIn which
Is [0,1]Random noise, function, evenly distributed over intervalsInputting a non-negative number and outputting 1, otherwise, inputting a negative number and outputting 0; definition ofThe mth sublayer outputVector of dimensionsWherein
4. A multivariate based probability mass as defined in claim 1The distributed information estimation method is characterized by comprising the following steps: in the step S3, bayesian estimation theory is used, and the method is consideredAfter the determined sensor quantifies the local observation, an estimated MSE lower bound which can be reached by using the quantified data is found in the fusion center to find the optimal design parameter suitable for the current observation environmentAnd corresponding use-optimized design parameters on the sensorIs optimized for the multivariate quantization probability function。
5. The distributed information estimation method based on multivariate probability quantization as claimed in claim 4, wherein: the step S3 includes:
are all provided withThe independent sensors are distributed in the whole network, each sensor adopts the step S2 to construct a multi-element probability quantizer structure, and all the multi-element probability quantizers use the same multi-element quantization probability functionAnd corresponding design parameters;
Each sensor respectively corresponding to the original informationObserving to obtain own local observation data with observation noise;
Considering the presence of independent and identically distributed observation noise on each sensor, and defining a probability density functionTo describe the distribution of observed noise: for the firstA sensor, its local observation dataIs quantized into a plurality of elements by a local multivariate probability quantizerBinary data of bitsAnd is sent to the FC of the network center, soA sensor jointly generatesAnQuantized data of bitsAnd sending the information to the FC, wherein the FC needs to estimate the original information by using all received quantized data;
quantitative data on all sensors at a given time by Bayesian probability theoryAlso in the case of (2), since they are independently and identically distributed, based on the probability distribution of the quantized data, it is first calculated when the FC receives the quantized dataFor the original informationThe lower bound of the estimated MSE that the estimation can achieve, i.e. the lower bound
Wherein,Indicating that FC receives all the quantized data sent from the sensor,representing FC utilization quantized dataWhat can be achieved isArbitrary estimation ofEvaluating the value, correspondingly to the left of the inequality in the formulaRepresenting original informationAnd its estimated valueThe MSE between the two is the MSE,shows the operation of calculating mathematical expectation, the right side of the inequality in the formula (6) shows the lower bound of the MSE,is the number of combinations in the mathematical definition,
is based on a multivariate quantization probability functionDesign parameters ofAnd original informationThe determined series of intermediate calculation terms, as can be seen from equation (6), are in the original informationProbability distribution and observed noise distribution ofAre all determinedFC utilizes quantized data pairsMSE estimated, i.e.Whose lower bound is fully defined by the multivariate quantization probability functionDesign parameters ofDetermining;
minimizing the right side of the inequality in equation (6) by an algorithmThe determined FC uses the lower bound of the MSE estimated by the quantized data on the original information to find the optimal design parameter adapted to the current observation environmentAnd corresponding optimal multivariate quantization probability function。
6. The distributed information estimation method based on multivariate probability quantization as claimed in claim 5, wherein: in the step S3, optimal design parameters suitable for the current observation environment are obtainedAnd corresponding optimal multivariate quantization probability functionThe process comprises the following steps:
a1, setting the total number of sensors asSample set, Is the total number of samples contained in the sample set,is a sample of the original information that was,indicates the serial number of the sample, andrepresenting sensor versus raw information samplesTotal observationsObtained by(ii) an observation sample; setting initial design parametersSetting the tolerance threshold of iteration toSetting an initial iteration count to;
A2 is atFirst, theIn the second iteration, the pairs are shown in formula (7)Definition of (2) toIs calculated byMultivariate quantization probability function design parameters obtained in sub-iterationA determined series of intermediate calculation termsWherein
A3, using the formula (8)And sample setThe lower MSE bound on the right side of the inequality in equation (6) is approximately calculated as
Then using the interior point method and the gradient descent method, the following minimization problem is solved
if not, the explanation needs to continue iteration, and needs to beStep A2 is carried out to carry out the next iteration, and the iteration count is updated(ii) a If the convergence condition is established, outputtingAs an optimal design parameter, and determining a corresponding optimal multivariate quantization probability function。
7. The distributed information estimation method based on multivariate probability quantization as claimed in claim 5, wherein: the step S4 includes:
s401.FC reception fromTransmitted from a sensorQuantized dataAnd inputting it into the quantization fusion estimator, and outputting the original informationIs estimated value of;
In the quantitative fusion estimator, of the inputAnBinary quantized data of bitsIs first converted intoIs arranged atMean decimal discrete data;
Decimal dataThen the data is sent to an Onehot function for carrying out one-hot coding operation to obtain the corresponding dataOne-hot coded vector;
S402, with the firstDecimal dataBy way of example only, it is possible to use,is its corresponding one-hot coded vector,is composed ofbinary data of bit length, toIn total ofThe bits bit are numbered in sequence asBit, decimal dataThe value range of (1) is just the serial number of all bits, and the one-hot coding means that only one bit is codedToThe bit will be set to 1 and all the rest of the bit bits will be set to 0, i.e. the bit is set to 1,
Then, the process of the present invention is carried out,one-hot coded vectorIs sent to an averager to obtain their mean vectors;
After thatIs then fed into the estimation functionIn the method, an estimated value of the original information is output,Is also a design parameterA function of control, i.e.
s403. When the same multivariate probability quantizer is used on all the sensors, and the multivariate probability quantizer with the same optimal design parameters is usedOf the multivariate quantization probability functionThe quantized data sent by all sensors to the FC is compared to the original informationAre conditionally independent and identically distributed;
at the moment, based on Bayesian estimation theory and probability model, the estimation value of FC to the original information is calculatedWith the original informationMSE between
following the pair in equation (7)By quantifying the optimal design parameters of the probability functionAnd original informationA determined series of intermediate calculation terms;
the optimal design parameters in the multivariate quantization probability function obtained from the formula (13)The MSE estimated on FC for the original information is determined entirely by the estimation functionVariable design parameters ofControlling;
s404, a series of samples based on the original information collected from the actual observation environment and the local observation data of the sensor, and an optimal multivariate quantization probability function on the sensorSolving the optimum estimation function design parameters that minimize the estimated MSE on FC。
8. The distributed information estimation method based on multivariate probability quantization as claimed in claim 7, wherein: the step S404 includes:
b1, total number of input sensorsOptimal multivariate quantization probability function on sensorAnd their corresponding optimum design parametersSample setFollowing the pair in equation (14)Definition of (2) to arbitraryWill be composed ofAnd original information samplesIntermediate item of decisionIs approximately calculated as
Setting initial settingsMetering parametersSetting the tolerance threshold of iteration toSetting an initial iteration count to;
B2 in the firstIn the second iteration, defineFor the estimated MSE at FC, using equation (13),and the design parameters obtained in the first iterationWill beIs approximately calculated as
Using the interior point method and the gradient descent method, the following minimization problem is solved
B3, calculating and checking convergence conditionsWhether the result is true or not; if not, continue iteration, willStep B2 is carried out for the next iteration, and the iteration count is updated(ii) a If the convergence condition is established, outputtingAs optimal design parameters and determining corresponding optimal estimation functions。
9. The distributed information estimation method based on multivariate probability quantization as claimed in claim 1, characterized in that: the step S5 includes:
for the firstA sensor, it is toObtaining own local observation data after observationAnd will beFed into a multivariate probability quantizer, multivariate quantization probability functionsGet the bestAnd finally outputbinary quantization data of bitIs sent to the FC; all ofA sensor jointly generatesQuantized data;
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211229494.1A CN115290130B (en) | 2022-10-08 | 2022-10-08 | Distributed information estimation method based on multivariate probability quantization |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202211229494.1A CN115290130B (en) | 2022-10-08 | 2022-10-08 | Distributed information estimation method based on multivariate probability quantization |
Publications (2)
Publication Number | Publication Date |
---|---|
CN115290130A CN115290130A (en) | 2022-11-04 |
CN115290130B true CN115290130B (en) | 2022-12-16 |
Family
ID=83819224
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202211229494.1A Active CN115290130B (en) | 2022-10-08 | 2022-10-08 | Distributed information estimation method based on multivariate probability quantization |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115290130B (en) |
Family Cites Families (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US8886526B2 (en) * | 2012-05-04 | 2014-11-11 | Sony Computer Entertainment Inc. | Source separation using independent component analysis with mixed multi-variate probability density function |
CN107451101B (en) * | 2017-07-21 | 2020-06-09 | 江南大学 | Method for predicting concentration of butane at bottom of debutanizer by hierarchical integrated Gaussian process regression soft measurement modeling |
CN110035405B (en) * | 2019-04-18 | 2020-04-28 | 电子科技大学 | Multi-sensor distributed efficient fusion method based on random set theory |
CN114235257A (en) * | 2021-12-24 | 2022-03-25 | 长春工业大学 | Bolt axial force online accurate monitoring method and device based on data fusion |
CN114900817B (en) * | 2022-07-12 | 2022-09-30 | 香港中文大学(深圳) | Distributed wireless sensor network information estimation method based on binary quantization |
-
2022
- 2022-10-08 CN CN202211229494.1A patent/CN115290130B/en active Active
Also Published As
Publication number | Publication date |
---|---|
CN115290130A (en) | 2022-11-04 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Luo | An isotropic universal decentralized estimation scheme for a bandwidth constrained ad hoc sensor network | |
CN113537514A (en) | High-energy-efficiency federal learning framework based on digital twins | |
CN113381952A (en) | Multi-antenna system channel estimation method based on deep learning | |
CN113179109A (en) | Honeycomb-removing large-scale MIMO uplink spectrum efficiency optimization method | |
Kar et al. | Distributed detection in noisy sensor networks | |
US11888703B1 (en) | Machine learning algorithms for quality of service assurance in network traffic | |
Zhang et al. | Deep learning based user association in heterogeneous wireless networks | |
CN115290130B (en) | Distributed information estimation method based on multivariate probability quantization | |
Shamaiah et al. | Greedy sensor selection under channel uncertainty | |
CN116321255A (en) | Compression and user scheduling method for high-timeliness model in wireless federal learning | |
CN114900817B (en) | Distributed wireless sensor network information estimation method based on binary quantization | |
Ozfatura et al. | Fast federated edge learning with overlapped communication and computation and channel-aware fair client scheduling | |
Malka et al. | Decentralized low-latency collaborative inference via ensembles on the edge | |
Vucic et al. | Fixed point iteration for max-min SIR balancing with general interference functions | |
CN114884957B (en) | Method and device for unloading computing tasks in air-space-ground network and electronic equipment | |
CN116471612A (en) | Method for jointly optimizing quantization level and client selection in federal learning network | |
Wang et al. | Distributed online quantile regression over networks with quantized communication | |
Wang et al. | Distortion outage minimization in Nakagami fading using limited feedback | |
CN114745389A (en) | Computing offloading method for mobile edge computing system | |
EP2193606A1 (en) | Method and apparatus for multiple description coding | |
Reisizadeh et al. | Adaptive bit allocation for communication-efficient distributed optimization | |
CN114337881A (en) | Wireless spectrum intelligent sensing method based on multi-unmanned aerial vehicle distribution and LMS | |
Azimi-Abarghouyi et al. | Quantized Hierarchical Federated Learning: A Robust Approach to Statistical Heterogeneity | |
CN108183757B (en) | Segmented approximate convex decomposition method for perception duration and resource allocation joint optimization | |
WO2020087293A1 (en) | Communication receiver and method for processing signal |
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 |