CN116756699B - Dynamic arrival angle estimation method, device and storage medium - Google Patents
Dynamic arrival angle estimation method, device and storage medium Download PDFInfo
- Publication number
- CN116756699B CN116756699B CN202311049609.3A CN202311049609A CN116756699B CN 116756699 B CN116756699 B CN 116756699B CN 202311049609 A CN202311049609 A CN 202311049609A CN 116756699 B CN116756699 B CN 116756699B
- Authority
- CN
- China
- Prior art keywords
- angle
- arrival
- estimation
- neural network
- antenna
- 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
- 238000000034 method Methods 0.000 title claims abstract description 57
- 238000003860 storage Methods 0.000 title claims abstract description 22
- 238000013528 artificial neural network Methods 0.000 claims abstract description 61
- 238000005259 measurement Methods 0.000 claims abstract description 48
- 230000008859 change Effects 0.000 claims abstract description 20
- 230000006870 function Effects 0.000 claims description 49
- 239000013598 vector Substances 0.000 claims description 40
- 238000009826 distribution Methods 0.000 claims description 19
- 238000012549 training Methods 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 7
- 238000006073 displacement reaction Methods 0.000 claims description 4
- 238000004891 communication Methods 0.000 abstract description 11
- 238000010586 diagram Methods 0.000 description 13
- 238000012545 processing Methods 0.000 description 9
- 230000005540 biological transmission Effects 0.000 description 7
- 230000004927 fusion Effects 0.000 description 5
- 230000008569 process Effects 0.000 description 4
- 238000003491 array Methods 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 239000000835 fiber Substances 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 230000033001 locomotion Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000001902 propagating effect Effects 0.000 description 2
- RYGMFSIKBFXOCR-UHFFFAOYSA-N Copper Chemical compound [Cu] RYGMFSIKBFXOCR-UHFFFAOYSA-N 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 229910052802 copper Inorganic materials 0.000 description 1
- 239000010949 copper Substances 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000010295 mobile communication Methods 0.000 description 1
- 238000003062 neural network model Methods 0.000 description 1
- 230000010287 polarization Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/047—Probabilistic or stochastic networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/09—Supervised learning
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computing Systems (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Probability & Statistics with Applications (AREA)
- Radar Systems Or Details Thereof (AREA)
Abstract
The embodiment of the application discloses a dynamic arrival angle estimation method, a device and a storage medium, wherein the dynamic arrival angle estimation method comprises the following steps: collecting a time sequence of the change of the arrival angle before the current moment, and using the time sequence as the input of a deep neural network, and predicting the prior estimation of the arrival angle at the current moment through the deep neural network; and obtaining an estimation of the arrival angle of the current moment according to the measurement, thereby obtaining more accurate posterior estimation of the arrival angle of the current moment so as to carry out phase adjustment of the antenna. The method solves the problem that the regularity of the change of the arrival angle caused by the fixed track of the train in the prior art is not applied to the estimation of the arrival angle, so that the railway communication system does not reach the performance which can be achieved originally.
Description
Technical Field
The present application relates to the field of wireless communications technologies, and in particular, to a dynamic arrival angle estimation method, a device, and a storage medium.
Background
Railway communication systems have high reliability requirements. To achieve this, directional antennas may be used to increase the strength of the received signal and reduce interference from other directions, as they may increase the signal strength in the main lobe direction and decrease the signal strength in other directions.
However, directional antennas require accurate estimation of the direction of signal arrival (angle of arrival), otherwise the signal strength may be lower than omni-directional antennas. The adjustable directional antenna may be implemented by electrical adjustment (e.g., phased antenna) or mechanical adjustment (e.g., parabolic antenna). The use of directional antennas requires accurate estimation of the signal angle of arrival, adjusting the antenna main lobe to the direction of the angle of arrival. For trains traveling at high speeds, angle of arrival estimation is not an easy matter. Since the movement speed of the train is very fast, a dynamic angle of arrival estimation is required. Dynamic angle of arrival estimation requires consideration of factors such as speed, direction and position of the train, and uncertainty and complexity of these factors make dynamic angle of arrival estimation very difficult.
Although techniques for angle of arrival tracking exist, they are not suitable for railroad communications. For example, the continuity of train operation allows the angle of arrival to be continuously varied. The train travels along a fixed track and thus the angle of arrival changes have a stronger regularity than other vehicles and hand held devices. This rule depends on factors such as the track line type, the position of the base station, and the speed of the train, and is thus difficult to express with a simple analytical expression.
Because of these difficulties, the regularity of angle of arrival changes imposed by the fixed track of the train is not used in the angle of arrival estimation, thereby causing the railroad communication system to not achieve what it would otherwise have achieved.
Disclosure of Invention
An object of an embodiment of the present application is to provide a dynamic arrival angle estimation method, apparatus and storage medium, which are used for solving the problem that in the prior art, regularity of arrival angle change caused by a fixed track of a train is not applied to arrival angle estimation, so that a railway communication system does not reach performance which can be achieved originally.
In order to achieve the above object, an embodiment of the present application provides a dynamic arrival angle estimation method, including: collecting a time sequence of the change of the arrival angle before the current moment, taking the time sequence as the input of a built deep neural network, and predicting the prior estimation of the arrival angle at the current moment through the deep neural network;
and combining the measurement result of the arrival angle at the current moment obtained through measurement with the prior estimation to obtain a posterior probability density function, thereby obtaining more accurate posterior estimation of the arrival angle at the current moment, and carrying out phase adjustment of the antenna.
Further, before the input serving as the built deep neural network predicts the prior estimation of the arrival angle at the current moment through the deep neural network, the method further comprises:
using a priori, posterior and estimated likelihood functions of normal distribution expression of angle of arrival, the angle of arrival is expected to beStandard deviation of->The probability density function of (2) is expressed as:
，
wherein delta is the directionAnd->The included angle between the two is cos delta as the product of the unit direction vectors of the two,
down to spherical coordinate systemThe unit vectors in Cartesian coordinate system areThe components of the distance of any antenna to the reference antenna in the signal proceeding direction are therefore:
，
where γ is the angle between d and e, this is true because the vector inner product is equal to the product of their modulo and the cosine of the angle between them, and e modulo is 1, and therefore,
。
further, the predicting, as an input to the built deep neural network, an a priori estimate of an arrival angle at a current time by the deep neural network includes:
the input of the deep neural network isOutputting the prior estimate of the angle of arrival at the current time, i.e., the angle of arrival at time t:
，
wherein the method comprises the steps ofAnd->Desired pitch angle and direction angle at time t, respectively,/->And the depth neural network is an RNN neural network architecture of LSTM or GRU, and is the standard deviation of the angle of arrival at the moment t.
Further, the method further comprises the following steps:
training the deep neural network by supervised learning, and training the deep neural network by using a time sequence of arrival angles obtained by actual measurement, wherein a training objective function is a likelihood function for maximizing the actual measurement angle for all time t:
，
wherein the method comprises the steps ofIs a neural network N β By inputting->Expected value of pitch angle at time t.
Further, the method further comprises the following steps:
optimizing a parameter set beta of the deep neural network by a gradient ascent method;
and adding a penalty term of system complexity in training the deep neural network, so as to avoid overfitting.
Further, the combining the obtained measurement result of the arrival angle at the current time with the prior estimation to obtain a posterior probability density function, thereby obtaining a more accurate posterior estimation of the arrival angle at the current time, including:
and obtaining the measured arrival angle probability distribution by measuring the signal intensity of the antenna main lobe in different directions. Describing arrival angle probability distribution by normal distribution according to the theorem of the large number, thereby obtaining the measurement result of the arrival angle at the current moment;
and expanding the sphere near the prior estimation and the measurement result into a plane to obtain the posterior probability density function, wherein the posterior probability density function is the product of the prior probability density function and the measurement probability density function.
Further, the determining a merging vector according to the obtained prior estimate of the arrival angle of the current moment includes:
defining one antenna in the reflecting surface as a reference antenna, and the displacement of any other antenna from the reference antenna is d= (deltax, deltay, deltaz);
down to spherical coordinate systemThe unit vectors in Cartesian coordinate system areThe components of the distance of any antenna to the reference antenna in the signal proceeding direction are therefore:
，
where γ is the angle between d and e, this holds because the vector inner product is equal to the product of their modulo and the cosine of the angle between them, and the modulo of e is 1;
the phase compensation of the antenna is then obtained according to l:
，
wherein f is the signal frequency, c is the speed of light, the compensating phase is the opposite number of the sum of the phase differences caused by the two, and the vector composed of the unit complex numbers of the compensating phases of all antennas is the combined vector.
In order to achieve the above object, the present application further provides a dynamic angle of arrival estimation apparatus, including: a memory; and
a processor coupled to the memory, the processor configured to perform the steps of the method as described above.
To achieve the above object, the present application also provides a computer storage medium having stored thereon a computer program which, when executed by a machine, implements the steps of the method as described above.
The embodiment of the application has the following advantages:
the embodiment of the application provides a dynamic arrival angle estimation method, which comprises the following steps: collecting a time sequence of the change of the arrival angle before the current moment, taking the time sequence as the input of a built deep neural network, and predicting the prior estimation of the arrival angle at the current moment through the deep neural network; determining a combining vector according to the obtained prior estimation of the arrival angle at the current moment so as to adjust the phase of the antenna; or combining the obtained measurement result of the arrival angle at the current moment with the prior estimation to obtain a posterior probability density function, thereby obtaining more accurate posterior estimation of the arrival angle at the current moment, and carrying out phase adjustment of the antenna.
By means of the method, the arrival angle can be estimated more accurately from the measurement result interfered by noise by data fusion of the arrival angle measurement result before the current moment, so that higher antenna gain is achieved, interference is reduced, the method can be used for a base station antenna and an antenna of a vehicle-mounted or handheld device, and therefore the problem that the regularity of arrival angle change caused by a fixed track of a train in the prior art is not applied to arrival angle estimation, and therefore the railway communication system does not reach the performance which can be achieved originally is solved.
Drawings
In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below. It will be apparent to those skilled in the art from this disclosure that the drawings described below are merely exemplary and that other embodiments may be derived from the drawings provided without undue effort.
FIG. 1 is a flow chart of a dynamic angle of arrival estimation method according to an embodiment of the present application;
fig. 2 is a schematic diagram of describing directions in a three-dimensional space by using a pitch angle and an azimuth angle according to an embodiment of the present application;
fig. 3 is a block diagram of a dynamic angle-of-arrival estimation device according to an embodiment of the present application.
Detailed Description
Other advantages and advantages of the present application will become apparent to those skilled in the art from the following detailed description, which, by way of illustration, is to be read in connection with certain specific embodiments, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the application without making any inventive effort, are intended to be within the scope of the application.
In addition, the technical features of the different embodiments of the present application described below may be combined with each other as long as they do not collide with each other.
In order to ensure safe operation of the railway, the railway needs to build a private network of the railway, the frequency band is independent of the frequency bands of other operators and industries, but due to the fact that more and more networks exist and are built, signals such as other networks, a broad-spectrum repeater, microwaves and the like often interfere railway signals, the interference problem is serious, and railway departments need to check the interference regularly.
Besides the interference from other networks, the railway is often required to shuttle in environments such as mountains, tunnels and the like, so that the transmission of wireless signals is very unstable, and besides the signals transmitted in a straight line, the railway signals have the influence of multipath interference such as diffraction, diffraction and the like and Doppler effect from other directions.
The omni-directional antenna adopted by the vehicle-mounted equipment is only a common civil mobile communication antenna except for being installed on the roof for changing the appearance, and is not specifically adjusted for special environments of railways, particularly high-speed railways. The railway line is fixed, the train runs in a certain section, the transmission path of the signal received by the vehicle-mounted equipment is basically fixed, the omnidirectional antenna has the advantages that the signal transmitted from all directions can be received, but the polarization direction of the omnidirectional antenna is fixed, the captured signal cannot be guaranteed to be the strongest signal, and the signal is not the strongest signal under normal conditions. Railroad communication devices are also optimized for interference from different product aspects, such as adding customized narrowband bandpass filters, physical layer algorithms, but all of these tasks are based on the signal that has been received and not from the front-end, i.e., antenna end, of the signal reception.
The application processes the received signal from the forefront end so as to ensure that the vehicle-mounted equipment can always receive the strongest signal.
The method comprises the following steps: the communication standard adopted by the current global railway is mostly GSM-R, and mostly works in the 900MHz frequency band, and an omni-directional antenna is adopted, and the basic parameters are shown in table 1.
TABLE 1 basic parameters of the railway communication System at present
The angle of arrival estimation is the basis for using high gain antennas. However, due to the high speed motion of the train, the angle of arrival estimation must be performed in real time. It is common practice to estimate and adjust the antenna main lobe direction once every fixed time (e.g., 0.1 seconds). However, since thermal noise and interference are unavoidable in the estimation process, the complexity of the estimation algorithm must also be controlled, and the accuracy of the angle-of-arrival estimation is difficult to be ensured.
Although the accuracy of the real-time angle of arrival estimation is difficult to guarantee, the train travels along a fixed route, and the change of the angle of arrival with time is very regular. Therefore, if the rule can be effectively utilized, the past estimation result and the current estimation result are subjected to data fusion, and the accuracy of the arrival angle estimation can be greatly improved.
An embodiment of the present application provides a dynamic angle of arrival estimation method, referring to fig. 1, fig. 1 is a flowchart of a dynamic angle of arrival estimation method provided in an embodiment of the present application, it should be understood that the method may further include additional blocks not shown and/or may omit the blocks shown, and the scope of the present application is not limited in this respect.
The embodiment of the application uses the pitch angle theta and the direction angle in the spherical coordinate systemDirections in three-dimensional space are described. Wherein θ=0° points directly above, θ=90° points in the horizontal direction, and θ=180° points directly below. />Is directed to the north-south direction of the human body,direct to the right west>Direct to the right south,/->Pointing to the forward direction. A method of describing directions in a three-dimensional space using pitch and azimuth angles is shown in fig. 2. Spherical coordinate system lower direction->The unit vector in Cartesian coordinate system is +.>。
The application uses a Kalman filter-like method to fuse the arrival angles at successive moments. The angle of arrival at the next time depends on the angle of arrival at the previous time and the rate of change of the angle of arrival between the two times. Unlike the kalman filter which performs data fusion in a linear cartesian coordinate system, the spherical coordinate system is not a linear coordinate system. The same three-dimensional angle change at different pitch angles corresponds to very different direction angle changes. For example, when the pitch angle θ=90° (the pitch angle points in the horizontal direction), a smaller three-dimensional angle change corresponds to a smaller direction angle change. However, when θ=0° (pitch angle is directed upward), a small three-dimensional angular change may correspond to a very large change in direction angle. Therefore, the present application cannot describe the relationship between the angle of arrival and the rate of change of the angle of arrival in a closed form.
Aiming at the problem, the application provides a method for replacing the change rate of the arrival angle by using the history record of the arrival angle, and obtaining the conditional probability of the current arrival angle in the history record of the specific arrival angle by using a deep neural network, thereby completing the data fusion under a nonlinear spherical coordinate system. The basis for this is that if the angle of arrival at a time before the current time t is known, i.e. the time series of changes in angle of arrival before the current time is acquiredThen the angle of arrival at the present moment +>Or may be approximately defined. By training a deep neural network, a priori estimates of the angle of arrival at the current time can be predicted from the history.
Thus, at step 101, a time series of angle of arrival changes before the current moment is acquired as input to a built deep neural network by which a priori estimates of the angle of arrival at the current moment are predicted.
In some embodiments, the actual angle of arrival is not known and cannot be used directly as a deep neural network input. Thus using a normal distribution to express a priori, a posteriori, and estimated likelihood functions of the angle of arrival, the arrivalThe angle is expected to beStandard deviation of->The probability density function of (2) is expressed as:
（1）
wherein, the liquid crystal display device comprises a liquid crystal display device,is the direction->And->Included angle between them, the product of unit direction vectors of them is，
Down to spherical coordinate systemThe unit vectors in Cartesian coordinate system areThe components of the distance of any antenna to the reference antenna in the signal proceeding direction are therefore:
，
where γ is the angle between d and e, this is true because the vector inner product is equal to the product of their modulo and the cosine of the angle between them, and e modulo is 1, and therefore,
（2）
in some embodimentsUsing deep neural network N β The change in angle of arrival at successive moments is described, where β is the parameter set of the deep neural network. The input of the deep neural network isOutputting the prior estimate of the angle of arrival at the current time, i.e., the angle of arrival at time t:
（3）
wherein the method comprises the steps ofAnd->Desired pitch angle and direction angle at time t, respectively,/->The standard deviation of the angle of arrival at time t. Since the input is a time series, the deep neural network uses RNN neural network architecture of LSTM or GRU, etc.
In some embodiments, the deep neural network is trained by supervised learning, with a time series of arrival angles obtained by actual measurement, the trained objective function being a likelihood function that maximizes the actual measurement angle for all times t (equation (1)):
（4）
wherein the method comprises the steps ofIs a neural network N β By input ofExpected value of pitch angle at time t.
In some embodiments, deep neural networks are optimized by gradient ascentThe complex parameter set beta can obtain the prior probability density of the direction angle change. In order to improve the stability of the prediction prior probability density estimation to the extreme condition, the application can define the standard deviation sigma of the output t Lower limit sigma of min Meanwhile, penalty items of system complexity are added into the training deep neural network, so that overfitting is avoided, and the reliability of the model for coping with the extreme case of small probability is improved.
At step 102, the measurement result of the arrival angle at the current time obtained by measurement is combined with the prior estimation to obtain a posterior probability density function, so as to obtain a more accurate posterior estimation of the arrival angle at the current time, so as to perform phase adjustment of the antenna.
In some embodiments, a combining vector may also be determined for phase adjustment of the antenna based on the obtained a priori estimate of the angle of arrival at the current time.
Specifically, a priori estimates of the angle of arrival at time t are obtainedThere are two applications. The first is to determine the combining vector (i.e. which direction the phased array should "look" from the obtained a priori estimate of the angle of arrival at the current time, since the a priori estimate should not deviate too much from the actual one (here embodying the introduction of sigma) min Is provided). The purpose of this is to reduce the number of merging vectors that need to be considered, reducing the complexity of the angle of arrival estimation. The calculation method of the merging vector is briefly described as follows:
defining one antenna in the reflecting surface as a reference antenna, and the displacement of any other antenna from the reference antenna is d= (deltax, deltay, deltaz);
as described above, the spherical coordinate system is downwardThe unit vectors in Cartesian coordinate system areThe component of the distance of any antenna to the reference antenna in the signal advancing direction is therefore：
，
Where γ is the angle between d and e, this holds because the vector inner product is equal to the product of their modulo and the cosine of the angle between them, and the modulo of e is 1;
the phase compensation of the antenna is then obtained according to l:
，
wherein f is the signal frequency, c is the speed of light, the compensating phase is the opposite number of the sum of the phase differences caused by the two, and the vector composed of the unit complex numbers of the compensating phases of all antennas is the combined vector.
The second application is to combine the measurement result with the prior estimate to obtain a posterior probability density function, thereby obtaining a more accurate posterior estimate. Because of measurement accuracy limitations, estimates based entirely on measurements are not accurate. And combining the acquired measurement result of the arrival angle at the current moment with the prior estimation to obtain an estimation result which is more accurate than the estimation based on measurement completely. The specific method comprises the following steps: and obtaining the measured arrival angle probability distribution by measuring the signal intensity of the antenna main lobe in different directions. And describing the arrival angle probability distribution by using normal distribution according to the big number theorem, thereby obtaining a measurement result. Assume that the measured estimate of the angle of arrival isThe standard deviation of the measurement error is +.>. In case the prior estimate and the measurement result do not differ much (in practical applications this assumption should be made), the sphere around the prior estimate and the measurement result can be unfolded into a plane, resulting in a posterior probability density function, which is a prior probability density function (the angle of arrival predicted from the previous estimate is taken differentlyThe probability magnitude at the value) and the measured probability density function (the probability magnitude of the angle of arrival at different values from the measurement result). The posterior probability density function is also a normal distribution according to its characteristics, and its expected value is:
（5）
（6）
the variance is:
（7）
the relation between the posterior probability and the prior probability, the measurement results can be clearly seen by formulas (5) - (7): standard deviation of，/>Accuracy of estimation or measurement is described: the higher the accuracy, the smaller the standard deviation. The expected value of the posterior estimate is a weighted average of the a priori estimated expected value and the measured expected value. The smaller the standard deviation, the more reliable the self-estimation, and the smaller the weight of the counterpart, and vice versa. Standard deviation of posterior estimation>Less than->And->It is shown that the accuracy of the posterior estimation is higher than both the prior estimation and the measurement. This demonstrates the effectiveness of the proposed method.
The method of the present application can be summarized as follows, in combination with the description of the above embodiments:
rail traffic information arrival angle estimation based on Bayesian inference and neural network prior knowledge:
for the estimated instants t=1, 2, …, T is calculated as follows:
according to formula (3), the result of estimating the angle of arrival before time tCalculating prior probability distribution of current arrival angle。
Estimating from a priori anglesDetermining the combined vector of the phased antenna, and measuring the arrival angle to obtain the measurement result +.>。
Calculating the posterior probability distribution of the arrival angle by formulas (5) - (7). And determining the merging vector of the phased antenna for data transmission.
The specific application scenarios of the scheme provided by the embodiment of the application comprise:
1. phased antennas are installed on a base station, train body or hand-held device for railroad communication, and the phase adjustment of each antenna can be adjusted by software.
2. And building a neural network model and a data acquisition platform of an LSTM or GRU architecture.
3. Acquisition of time series of angle of arrival changesWhere T is the total number of travel sample times at this time.
4. After collecting enough data, the neural network is trained using equation (4) as an objective function.
5. After training is completed, in the application stage, the formula is used(3) Calculating a priori estimate of the angle of arrival, estimating the angle of arrival around the expected value of the prior probability, or obtaining a measurement resultAnd (3) calculating to obtain a more accurate arrival angle posterior estimation result by using the steps (5), (6) and (7).
By the method, dynamic arrival angle estimation by using Bayesian inference is proposed. Because the train travel track is fixed, the change in the angle of arrival at the front and rear moments is highly correlated. By carrying out data fusion on the measurement result of the angle of arrival before the current moment, the application can more accurately estimate the angle of arrival from the measurement result interfered by noise, thereby realizing higher antenna gain and reducing interference. The application can be used for base station antennas and antennas of vehicle-mounted or handheld devices.
Fig. 3 is a block diagram of a dynamic angle-of-arrival estimation device according to an embodiment of the present application. The device comprises:
a memory 201; and a processor 202 connected to the memory 201, the processor 202 configured to: collecting a time sequence of the change of the arrival angle before the current moment, taking the time sequence as the input of a built deep neural network, and predicting the prior estimation of the arrival angle at the current moment through the deep neural network;
determining a combining vector according to the obtained prior estimation of the arrival angle at the current moment so as to adjust the phase of the antenna; or alternatively, the process may be performed,
and combining the acquired measurement result of the arrival angle at the current moment with the prior estimation to obtain a posterior probability density function, thereby obtaining more accurate posterior estimation of the arrival angle at the current moment, and carrying out phase adjustment of the antenna.
In some embodiments, the processor 202 is further configured to: the method for estimating the arrival angle of the current moment by using the depth neural network as the input of the built depth neural network comprises the following steps:
using a priori, posterior and estimated likelihood functions of normal distribution expression of angle of arrival, the angle of arrival is expected to beStandard deviation of->The probability density function of (2) is expressed as:
，
wherein delta is the directionAnd->The included angle between the two is cos delta as the product of the unit direction vectors of the two,
down to spherical coordinate systemThe unit vectors in Cartesian coordinate system areThe components of the distance of any antenna to the reference antenna in the signal proceeding direction are therefore:
，
where γ is the angle between d and e, this is true because the vector inner product is equal to the product of their modulo and the cosine of the angle between them, and e modulo is 1, and therefore,
。
in some embodiments, the processor 202 is further configured to: the input serving as the built deep neural network predicts the prior estimation of the arrival angle at the current moment through the deep neural network, and comprises the following steps:
the input of the deep neural network isOutputting the prior estimate of the angle of arrival at the current time, i.e., the angle of arrival at time t:
，
wherein the method comprises the steps ofAnd->Desired pitch angle and direction angle at time t, respectively,/->And the depth neural network is an RNN neural network architecture of LSTM or GRU, and is the standard deviation of the angle of arrival at the moment t.
In some embodiments, the processor 202 is further configured to: further comprises:
training the deep neural network by supervised learning, and training the deep neural network by using a time sequence of arrival angles obtained by actual measurement, wherein a training objective function is a likelihood function for maximizing the actual measurement angle for all time t:
，
wherein the method comprises the steps ofIs a neural network N β By input ofExpected value of pitch angle at time t.
In some embodiments, the processor 202 is further configured to: further comprises:
optimizing a parameter set beta of the deep neural network by a gradient ascent method;
and adding a penalty term of system complexity in training the deep neural network, so as to avoid overfitting.
In some embodiments, the processor 202 is further configured to: combining the obtained measurement result of the arrival angle at the current moment with the prior estimation to obtain a posterior probability density function, thereby obtaining more accurate posterior estimation of the arrival angle at the current moment, and comprising the following steps:
and obtaining the measured arrival angle probability distribution by measuring the signal intensity of the antenna main lobe in different directions. Describing arrival angle probability distribution by normal distribution according to the theorem of the large number, thereby obtaining the measurement result of the arrival angle at the current moment;
and expanding the sphere near the prior estimation and the measurement result into a plane to obtain the posterior probability density function, wherein the posterior probability density function is the product of the prior probability density function and the measurement probability density function.
In some embodiments, the processor 202 is further configured to: the determining a merging vector according to the obtained prior estimation of the arrival angle of the current moment comprises the following steps:
defining one antenna in the reflecting surface as a reference antenna, and the displacement of any other antenna from the reference antenna is d= (deltax, deltay, deltaz);
down to spherical coordinate systemThe unit vectors in Cartesian coordinate system areThe components of the distance of any antenna to the reference antenna in the signal proceeding direction are therefore:
，
where γ is the angle between d and e, this holds because the vector inner product is equal to the product of their modulo and the cosine of the angle between them, and the modulo of e is 1;
the phase compensation of the antenna is then obtained according to l:
，
wherein f is the signal frequency, c is the speed of light, the compensating phase is the opposite number of the sum of the phase differences caused by the two, and the vector composed of the unit complex numbers of the compensating phases of all antennas is the combined vector.
Reference is made to the foregoing method embodiments for specific implementation methods, and details are not repeated here.
The present application may be a method, apparatus, system, and/or computer program product. The computer program product may include a computer readable storage medium having computer readable program instructions embodied thereon for performing various aspects of the present application.
The computer readable storage medium may be a tangible device that can hold and store instructions for use by an instruction execution device. The computer readable storage medium may be, for example, but not limited to, an electronic storage device, a magnetic storage device, an optical storage device, an electromagnetic storage device, a semiconductor storage device, or any suitable combination of the foregoing. More specific examples (a non-exhaustive list) of the computer-readable storage medium would include the following: portable computer disks, hard disks, random Access Memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), static Random Access Memory (SRAM), portable compact disk read-only memory (CD-ROM), digital Versatile Disks (DVD), memory sticks, floppy disks, mechanical coding devices, punch cards or in-groove structures such as punch cards or grooves having instructions stored thereon, and any suitable combination of the foregoing. Computer-readable storage media, as used herein, are not to be construed as transitory signals per se, such as radio waves or other freely propagating electromagnetic waves, electromagnetic waves propagating through waveguides or other transmission media (e.g., optical pulses through fiber optic cables), or electrical signals transmitted through wires.
The computer readable program instructions described herein may be downloaded from a computer readable storage medium to a respective computing/processing device or to an external computer or external storage device over a network, such as the internet, a local area network, a wide area network, and/or a wireless network. The network may include copper transmission cables, fiber optic transmissions, wireless transmissions, routers, firewalls, switches, gateway computers and/or edge servers. The network interface card or network interface in each computing/processing device receives computer readable program instructions from the network and forwards the computer readable program instructions for storage in a computer readable storage medium in the respective computing/processing device.
Computer program instructions for carrying out operations of the present application may be assembly instructions, instruction Set Architecture (ISA) instructions, machine-related instructions, microcode, firmware instructions, state setting data, or source or object code written in any combination of one or more programming languages, including an object oriented programming language such as Smalltalk, c++ or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The computer readable program instructions may be executed entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computer (for example, through the Internet using an Internet service provider). In some embodiments, aspects of the present application are implemented by personalizing electronic circuitry, such as programmable logic circuitry, field Programmable Gate Arrays (FPGAs), or Programmable Logic Arrays (PLAs), with state information for computer readable program instructions, which can execute the computer readable program instructions.
Various aspects of the present application are described herein with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the application. It will be understood that each block of the flowchart illustrations and/or block diagrams, and combinations of blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer-readable program instructions.
These computer readable program instructions may be provided to a processing unit of a general purpose computer, special purpose computer, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processing unit of the computer or other programmable data processing apparatus, create means for implementing the functions/acts specified in the flowchart and/or block diagram block or blocks. These computer readable program instructions may also be stored in a computer readable storage medium that can direct a computer, programmable data processing apparatus, and/or other devices to function in a particular manner, such that the computer readable medium having the instructions stored therein includes an article of manufacture including instructions which implement the function/act specified in the flowchart and/or block diagram block or blocks.
The computer readable program instructions may also be loaded onto a computer, other programmable data processing apparatus, or other devices to cause a series of operational steps to be performed on the computer, other programmable apparatus or other devices to produce a computer implemented process such that the instructions which execute on the computer, other programmable apparatus or other devices implement the functions/acts specified in the flowchart and/or block diagram block or blocks.
The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present application. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of instructions, which comprises one or more executable instructions for implementing the specified logical function(s). In some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
Note that all features disclosed in this specification (including any accompanying claims, abstract and drawings) may be replaced by alternative features serving the same, equivalent or similar purpose, unless expressly stated otherwise. Thus, unless expressly stated otherwise, each feature disclosed is one example only of a generic set of equivalent or similar features. Where used, further, preferably, still further and preferably, the brief description of the other embodiment is provided on the basis of the foregoing embodiment, and further, preferably, further or more preferably, the combination of the contents of the rear band with the foregoing embodiment is provided as a complete construct of the other embodiment. A further embodiment is composed of several further, preferably, still further or preferably arrangements of the strips after the same embodiment, which may be combined arbitrarily.
While the application has been described in detail in the foregoing general description and specific examples, it will be apparent to those skilled in the art that modifications and improvements can be made thereto. Accordingly, such modifications or improvements may be made without departing from the spirit of the application and are intended to be within the scope of the application as claimed.
Claims (7)
1. A method for dynamic angle of arrival estimation, comprising:
collecting a time sequence of the change of the arrival angle before the current moment, taking the time sequence as the input of a built deep neural network, and predicting the prior estimation of the arrival angle at the current moment through the deep neural network;
combining the measurement result of the arrival angle at the current moment obtained through measurement with the prior estimation to obtain a posterior probability density function, thereby obtaining more accurate posterior estimation of the arrival angle at the current moment so as to carry out phase adjustment of the antenna;
the step of combining the obtained measurement result of the arrival angle at the current time with the prior estimation to obtain a posterior probability density function, thereby obtaining more accurate posterior estimation of the arrival angle at the current time, including:
obtaining measured arrival angle probability distribution by measuring signal intensity of the antenna main lobe in different directions, and describing the arrival angle probability distribution by normal distribution according to a big number theorem, thereby obtaining the measurement result of the arrival angle at the current moment;
expanding the sphere near the prior estimation and the measurement result into a plane to obtain the posterior probability density function, wherein the posterior probability density function is the product of the prior probability density function and the measurement probability density function;
the determining a merging vector according to the obtained prior estimation of the arrival angle of the current moment comprises the following steps:
defining one antenna in the reflecting surface as a reference antenna, and the displacement of any other antenna from the reference antenna is d= (deltax, deltay, deltaz);
the direction (theta) below the spherical coordinate system,) The unit vector in the cartesian coordinate system is e= (sin θcos +.>, sin θ sinCos θ), the component of the distance of any antenna to the reference antenna in the signal proceeding direction is:
，
where γ is the angle between d and e, this holds because the vector inner product is equal to the product of their modulo and the cosine of the angle between them, and the modulo of e is 1;
the phase compensation of the antenna is then obtained according to l:
，
wherein f is the signal frequency, c is the speed of light, the compensating phase is the opposite number of the sum of the phase differences caused by the two, and the vector composed of the unit complex numbers of the compensating phases of all antennas is the combined vector.
2. The method for estimating a dynamic angle of arrival according to claim 1, wherein the method further comprises, as an input to a built deep neural network, before predicting a priori estimates of the angle of arrival at the current time by the deep neural network:
using a priori, a posteriori, and estimated likelihood functions that express angles of arrival in normal distribution, the angles of arrival are expected to be (θ e ,) Standard deviation is sigma e The probability density function of (2) is expressed as:
，
where δ is the direction (θ,) Sum (theta) e ,/>) The included angle between the two is cos delta as the product of the unit direction vectors of the two,
the direction (theta) below the spherical coordinate system,) The unit vector in the cartesian coordinate system is e= (sin θcos +.>, sin θ sinCos θ), the component of the distance of any antenna to the reference antenna in the signal proceeding direction is:
，
where γ is the angle between d and e, this is true because the vector inner product is equal to the product of their modulo and the cosine of the angle between them, and e modulo is 1, and therefore,
。
3. the method for estimating a dynamic angle of arrival according to claim 2, wherein the predicting the prior estimate of the angle of arrival at the current time by the depth neural network as an input to the built depth neural network comprises:
the input of the deep neural network isOutputting the prior estimate of the angle of arrival at the current time, i.e., the angle of arrival at time t:
，
wherein θ is e t Andexpected, σ, for pitch angle and direction angle at time t, respectively e t And the depth neural network is an RNN neural network architecture of LSTM or GRU, and is the standard deviation of the angle of arrival at the moment t.
4. The dynamic angle of arrival estimation method according to claim 3, further comprising:
training the deep neural network by supervised learning, and training the deep neural network by using a time sequence of arrival angles obtained by actual measurement, wherein a training objective function is a likelihood function for maximizing the actual measurement angle for all time t:
，
wherein the method comprises the steps ofIs a neural network N β By inputting->Expected value of pitch angle at time t.
5. The method of dynamic angle of arrival estimation according to claim 4, further comprising:
optimizing a parameter set beta of the deep neural network by a gradient ascent method;
and adding a penalty term of system complexity in training the deep neural network, so as to avoid overfitting.
6. A dynamic angle of arrival estimation apparatus, comprising:
a memory; and
a processor connected to the memory, the processor being configured to perform the steps of the method of any one of claims 1 to 5.
7. A computer storage medium having stored thereon a computer program, which when executed by a machine performs the steps of the method according to any of claims 1 to 5.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311049609.3A CN116756699B (en) | 2023-08-21 | 2023-08-21 | Dynamic arrival angle estimation method, device and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311049609.3A CN116756699B (en) | 2023-08-21 | 2023-08-21 | Dynamic arrival angle estimation method, device and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116756699A CN116756699A (en) | 2023-09-15 |
| CN116756699B true CN116756699B (en) | 2023-11-03 |
Family
ID=87950081
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311049609.3A Active CN116756699B (en) | 2023-08-21 | 2023-08-21 | Dynamic arrival angle estimation method, device and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116756699B (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113347125A (en) * | 2021-06-09 | 2021-09-03 | 浙江大学 | Bayesian neural network channel estimation method and device for MIMO-OFDM communication system |
| CN113472415A (en) * | 2021-09-02 | 2021-10-01 | 网络通信与安全紫金山实验室 | Signal arrival angle estimation method and device, electronic equipment and storage medium |
| CN114004336A (en) * | 2021-10-29 | 2022-02-01 | 上海科技大学 | Three-dimensional ray reconstruction method based on enhanced variational self-encoder |
| CN114545342A (en) * | 2022-02-16 | 2022-05-27 | 西安电子科技大学 | Radar pulse signal parameter measuring method using multi-channel reconnaissance receiver |
-
2023
- 2023-08-21 CN CN202311049609.3A patent/CN116756699B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113347125A (en) * | 2021-06-09 | 2021-09-03 | 浙江大学 | Bayesian neural network channel estimation method and device for MIMO-OFDM communication system |
| CN113472415A (en) * | 2021-09-02 | 2021-10-01 | 网络通信与安全紫金山实验室 | Signal arrival angle estimation method and device, electronic equipment and storage medium |
| CN114004336A (en) * | 2021-10-29 | 2022-02-01 | 上海科技大学 | Three-dimensional ray reconstruction method based on enhanced variational self-encoder |
| CN114545342A (en) * | 2022-02-16 | 2022-05-27 | 西安电子科技大学 | Radar pulse signal parameter measuring method using multi-channel reconnaissance receiver |
Non-Patent Citations (2)
| Title |
|---|
| Bayesian Compressive Sensing Approaches for Direction of Arrival Estimation With Mutual Coupling Effects;Matthew Hawes 等;《IEEE TRANSACTIONS ON ANTENNAS AND PROPAGATION》;全文 * |
| 基于空间相关性的混合RSS-AOA 定位;张俊 等;《通信技术》;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116756699A (en) | 2023-09-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6643883B2 (en) | GPS drift processing method and system | |
| CN105093198A (en) | Flight path fusion method for networking detection of distributed external radiation source radars | |
| Meng et al. | Vehicular connectivity on complex trajectories: Roadway-geometry aware isac beam-tracking | |
| Ulmschneider et al. | Multiple hypothesis data association for multipath-assisted positioning | |
| CN114537477B (en) | Train positioning tracking method based on TDOA | |
| Dehkordi et al. | Adaptive beam tracking based on recurrent neural networks for mmWave channels | |
| CN116756699B (en) | Dynamic arrival angle estimation method, device and storage medium | |
| Yin et al. | A novel distributed sensor fusion algorithm for RSSI-based location estimation using the unscented kalman filter | |
| Rana et al. | Position and velocity estimations of mobile device incorporate GNSS | |
| Roth et al. | On-road trajectory generation from GPS data: A particle filtering/smoothing application | |
| US20220091252A1 (en) | Motion state determining method and apparatus | |
| CN114035154B (en) | Single-station radio frequency signal positioning method assisted by motion parameters | |
| CN106885567A (en) | A kind of inertial navigation Cooperative Localization Method and location equipment | |
| CN113329416B (en) | Beam alignment method and system of unmanned aerial vehicle-mounted antenna array | |
| Krishanth et al. | Prediction and retrodiction algorithms for path-constrained targets | |
| US20100289699A1 (en) | Method and Apparatus for Determining a Position of a Mobile Station in a Wireless Communication System | |
| Bhardwaj et al. | Federated Learning-Based Joint Radar-Communication mmWave Beamtracking with Imperfect CSI for V2X Communications | |
| Niculescu et al. | An energy-efficient localization system for imprecisely positioned sensor nodes with flying UAVs | |
| JP2005308428A (en) | Position estimation system | |
| Correa et al. | Indoor pedestrian tracking system exploiting multiple receivers on the body | |
| CN113093171A (en) | Airborne radar path and radiation resource joint optimization method based on target tracking | |
| CN111757242A (en) | Sea area communication beam control method and device based on AIS information calculation | |
| Bian et al. | Reconfigurable Intelligent Surface Assisted Mobile Target Tracking With Agile Phase Control | |
| Zhao et al. | Integrated sensing and communication‐promoted beam tracking and coverage for complex vehicular networks | |
| US20230068500A1 (en) | Method, device and system of prediction of direction of arrival of a 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 |