WO2020248443A1

WO2020248443A1 - Two-dimensional phased antenna array-based rapid and accurate beam tracking method

Info

Publication number: WO2020248443A1
Application number: PCT/CN2019/110029
Authority: WO
Inventors: 刘羽; 周世东
Original assignee: 清华大学
Priority date: 2019-06-14
Filing date: 2019-10-09
Publication date: 2020-12-17
Also published as: CN110429964B; CN110429964A

Abstract

The present invention provides a two-dimensional phased antenna array-based rapid and accurate beam tracking method. The method is suitable for millimeter wave mobile communication. By dividing a time-varying channel into three situations, i.e., a quasi-static scene, a dynamic type I, and a dynamic type II, the method uses the optimized detection direction and tracks channel parameters under each situation according to a random Newton's method. The advantage of the method is that it can achieve the minimum pilot overhead of the joint tracking of the two-dimensional beam direction and the channel coefficient. For the quasi-static scene and the dynamic type I, the performance of the method can approach the Cramer-Rao lower bound of the tracking error. For the dynamic type II, the method can achieve higher tracking accuracy and faster tracking speed.

Description

A fast and accurate beam tracking method based on two-dimensional phased antenna array

Technical field

The invention belongs to the field of communication technology, relates to millimeter wave mobile communication, and particularly relates to a fast and accurate beam tracking method based on a two-dimensional phased antenna array.

Background technique

The huge bandwidth advantage makes millimeter wave communication have broad application prospects. Due to the short wavelength, millimeter wave communication systems are often equipped with large-scale antennas to achieve high array gain, which makes the millimeter wave beam very narrow, and a slight misalignment may cause huge energy loss. In a static scene or a quasi-static scene where the channel changes slowly, beam alignment can be achieved through pilot training, which will occupy huge pilot resources. However, in a scene where the channel changes rapidly, the alignment cannot be achieved directly by means of pilot training. Therefore, it is necessary to achieve accurate beam tracking with as little pilot overhead as possible in mobile scenarios.

At the same time, with the increase in the number of antennas and the increase in carrier frequency, the hardware overhead and energy loss of the all-digital beamforming architecture used in millimeter wave communications are huge. An economically feasible solution is to use an analog beamforming architecture. When the analog beamforming architecture is adopted, the receiver can only obtain observations in a specific direction at a time, and the selected detection beamforming vector will have a great impact on system performance. Therefore, how to select the detection beamforming vector to achieve accurate beam tracking is an important issue for millimeter wave communications in fast-moving scenarios.

There have been a series of research algorithms for millimeter wave beam tracking in mobile scenes. By using the sparsity of the millimeter wave channel, an algorithm based on compressed sensing is proposed. However, this algorithm is mainly designed for static or quasi-static scenes, and its performance is poor in fast-moving scenes. IEEE 802.11ad proposes a simple dynamic tracking algorithm, which switches between the current beam and adjacent beams based on the received signal strength. Someone applied the idea of Kalman filtering to the beam tracking problem, using the beam direction and channel coefficients as the system state, and applying the extended Kalman filtering method for tracking. However, these methods do not optimize the detection beamforming vector. In the analog beamforming architecture, the tracking performance is closely related to the selected detection beamforming vector. Failure to optimize the detection beamforming vector makes the tracking accuracy of these algorithms insufficient. Based on a one-dimensional antenna array, an algorithm for joint tracking of channel coefficients and beam direction is proposed, and the detection beamforming vector is optimized. However, this algorithm can only support one-dimensional beam direction tracking. In many scenarios such as UAV communication, the beam direction may arrive from different horizontal and vertical directions. Therefore, it is necessary to track the two-dimensional beam direction simultaneously.

Summary of the invention

In order to overcome the above-mentioned shortcomings of the prior art, the purpose of the present invention is to provide a fast and accurate beam tracking method based on a two-dimensional phased antenna array. The pilot overhead used should be as small as possible and the detection beamforming vector should be optimized. , In order to achieve the highest possible tracking accuracy and the fastest possible tracking speed.

In order to achieve the above objective, the technical solution adopted by the present invention is:

A fast and accurate beam tracking method based on a two-dimensional phased antenna array, which is characterized in that it periodically works in detection and communication modes alternately. During the detection period of a detection communication cycle, the beam is directed to several detection directions in sequence, and the communication The phase beam points to the beam direction estimated at the time.

The beam tracking method includes the following steps:

Step A: According to the time-varying characteristics of the channel, the channel is divided into three types: 1) quasi-static scene; 2) dynamic type I; 3) dynamic type II. Determine which type of channel to be tracked belongs to;

Step B: In the detection period of each detection communication cycle, based on the channel type determined in step A, according to the three types of beam tracking methods, determine the direction of sequential detection, and obtain according to the detection results and historically recorded information Estimation of channel coefficients and beam direction;

Step C: Point the beam to the beam direction estimation obtained in Step B for communication.

In step A, the quasi-static scene refers to a situation where the channel coefficient and the beam direction change very slowly: β _k ≈β _k-1 , x _k ≈x _k-1 . Where β _k refers to the channel coefficient of the k-th probing communication cycle, and x _k refers to the direction parameter vector of the k-th probing communication cycle:

x _{k, 1} , x _{k, 2} are the values of the two dimensions of the two-dimensional direction parameter vector x _k , respectively, θ _k , φ _k refer to the elevation angle and azimuth angle of the arrival beam direction in the k-th detection communication cycle. M, N are the number of antennas in two directional dimensions of the two-dimensional area array antenna, d ₁ , d ₂ are the distances between adjacent antennas in the two directional dimensions of the area array antenna, and λ is the millimeter wave wavelength.

In step A, the dynamic type I refers to the situation where the channel coefficient changes rapidly and the beam direction changes very slowly: β _k ≠ β _k-1 , x _k ≈x _k-1 .

In step A, the dynamic type II refers to the situation where both the channel coefficient and the beam direction change rapidly: β _k ≠ β _k-1 , x _k ≠ x _k-1 .

In step B, the tracking steps for the quasi-static scene are as follows:

Step 11: Obtain the initial direction estimation in the real beam main lobe range and the initial estimation of the channel coefficients. The initial estimation method used may include, but is not limited to, beam scanning: in one probing communication cycle, M×N pilots are sent, and these pilots are used to perform M×N detections in different directions. According to Orthogonal Matching Pursuit (OMP) algorithm, obtain the initial estimation of channel coefficients and beam direction;

Step 12: In the detection period of each detection communication cycle, three pilots are sent in sequence to perform three detections in different directions. The detection direction is the last estimated direction plus three sets of fixed offsets Δ _S,1 , Δ _S,2 , Δ _S,3 , these three sets of offsets are located on a ring of the two-dimensional plane: 0.4≤ |Δ _S,i |≤0.6,i=1, 2, 3. The three sets of offsets can be optimized according to the target to be tracked. The optimization targets include but are not limited to the mean square error of the channel vector and the mean square error of the direction parameter vector.

Step 13, based on the detection results and historically recorded information, update the estimated value using a random iteration method. The random iteration method used may include, but is not limited to, the random Newton method.

In step B, for the dynamic type I, only the beam direction is tracked.

In step B, the tracking steps for dynamic type I are as follows:

Step 21: Obtain an initial direction estimate in the real beam main lobe range. The initial estimation method used may include but is not limited to beam scanning: in the first L probing communication cycles, each probing communication cycle sends M×N pilots, using these pilots, each probing communication cycle performs M×N Detection in different directions. The selection of L must ensure that the maximum signal energy of M×N different detection directions received in the L-th detection communication cycle is greater than a certain threshold. Then use the detection result of the L-th detection communication cycle to obtain an initial estimate of the beam direction.

Step 22: In the detection period of each detection communication cycle, three pilots are sent in sequence to perform three detections in different directions. The detection direction is the last estimated direction plus three sets of fixed offsets Δ _DI,1 , Δ _DI,2 , Δ _DI,3 , these three sets of offsets are located on a ring of the two-dimensional plane 0.5≤| Δ _DI,i |≤0.7, i=1, 2, 3. The three sets of offsets can be optimized according to the target to be tracked. The optimization target includes but is not limited to the mean square error of the direction parameter vector.

Step 23, based on the detection result and historically recorded information, update the estimated value using a random iteration method. The random iteration method used may include, but is not limited to, the random Newton method.

In step B, the tracking steps for dynamic type II are as follows:

Step 31: Obtain an initial direction estimate in the real beam main lobe range and an initial estimate of channel coefficients. The initial estimation method adopted may include, but is not limited to, beam scanning: in one probing communication cycle, M×N pilots are sent in sequence, and these pilots are used to perform M×N detections in different directions. According to the OMP algorithm, an initial estimate of the channel coefficients and beam direction is obtained.

Step 32: In the detection period of each detection communication cycle, three pilots are sent in sequence to perform three detections in different directions. The detection direction is the last estimated direction plus three sets of fixed offsets Δ _DII,1 , Δ _DII,2 , Δ _DII,3 , these three sets of offsets are located on a ring in the two-dimensional plane: 0.4≤ |Δ _DII,i |≤0.6,i=1, 2, 3. The three sets of offsets can be optimized according to the target to be tracked. The optimization targets include but are not limited to the mean square error of the channel vector and the mean square error of the direction parameter vector.

Step 33, based on the detection result and historically recorded information, update the estimated value using a random iteration method. The random iteration method used may include, but is not limited to, the random Newton method.

Compared with the prior art, the beneficial effects of the present invention are:

(1) The minimum pilot overhead for joint tracking of two-dimensional channel coefficients and beam directions is realized.

(2) In static scenes and dynamic type I, the method can reach the Cramer-Road lower bound (CRLB) of the two-dimensional beam tracking error asymptotically.

(3) In the dynamic type II, compared with the prior art, the tracking method can achieve higher tracking accuracy and faster tracking speed.

Description of the drawings

Fig. 1 is a schematic diagram of a two-dimensional phased antenna array structure according to an embodiment of the present invention.

Fig. 2 is a frame structure of a beam tracking method in a quasi-static scene according to an embodiment of the present invention.

Fig. 3 is a diagram showing an example of the optimal offset of the beam tracking detection direction in a quasi-static scene according to an embodiment of the present invention.

Fig. 4 is a performance simulation diagram of a beam tracking method in a quasi-static scene according to an embodiment of the present invention.

Fig. 5 is a frame structure of a dynamic I-beam tracking method according to an embodiment of the present invention.

Fig. 6 is a diagram showing an example of the optimal offset of the dynamic I-beam tracking detection direction according to an embodiment of the present invention.

Fig. 7 is a performance simulation diagram of a dynamic I-beam tracking method according to an embodiment of the present invention.

Fig. 8 is a performance simulation diagram of a dynamic type II beam tracking method according to an embodiment of the present invention.

Detailed ways

The embodiments of the present invention are described in detail below. Examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals represent the same or similar elements or elements with the same or similar functions. The embodiments described below with reference to the drawings are exemplary, and are only used to explain the present invention, but should not be construed as limiting the present invention.

This method studies the tracking problem at the receiving end, and the tracking at the transmitting end can be achieved through channel reciprocity.

Consider the two-dimensional phased array antenna receiver shown in Figure 1: The planar antenna array is composed of M×N elements distributed in a rectangular area, and the distance between adjacent elements in the x(y) axis is d ₁ ( d ₂ ). The receiver uses an analog beamforming architecture, and each element is connected to the same RF link through a separate phase shifter. In this embodiment, the channel remains unchanged during the probing communication period. Each probe in the probe communication cycle period, the transmitter transmits the q same known pilot s _p, for estimating the channel coefficients and the beam direction of arrival.

The millimeter wave scattering effect is weak, and there is only a direct path and a small amount of reflection path in the channel. Because the angle extension is small, the mutual interference between multipaths is quite weak. Therefore, each path in the multipath can be tracked independently. This embodiment focuses on the tracking method of one path, and other multipaths can be tracked independently by the same method. In the k-th detection communication cycle, the direction of the beam arrival beam to be studied is (θ _k , φ _k ), where θ _k is the elevation angle of the arrival beam direction, and φ _k is the azimuth angle of the arrival beam direction. The channel vector of this path can be expressed as

h _k = β _k a(x _k ), (1)

among them

Is the complex channel coefficient,

Is the direction parameter vector determined by (θ _k ,φ _k ), a(x _k ) is the steering vector

a(x _k )=[a ₁₁ (x _k )...a _1N (x _k )a ₂₁ (x _k )...a _MN (x _k )] ^T , (2)

among them

λ is the millimeter wave wavelength.

In the probing period of the k-th probing communication cycle, use w _k,i (i=1,...q) to represent the probing beamforming vector used to receive the i-th pilot symbol, considering that w _k,i also has the form of a steering vector ,

Among them, γ _k,i represents the detection direction vector corresponding to w _k,i , and the detection direction of the receiver can be adjusted by changing γ _k,i . After phase shifting and combining, the baseband output signal finally obtained by the receiver is

among them

It is a complex loop Gaussian random variable. definition

Is the channel parameter vector of the k-th sounding communication cycle,

To detect the beamforming matrix,

Is the noise vector, the baseband output signal obtained by the receiver in (4) can be rewritten as

In quasi-static scenarios, dynamic type I and dynamic type II scenarios, it is necessary to ensure that the detection results of each detection communication cycle are used to obtain a unique estimation of the channel parameter vector. The minimum pilot overhead required is 3 pilots per detection communication cycle. Frequency, so q=3.

Next, the beam tracking method under different channel change scenarios is given.

1) Quasi-static scene

In this scenario, both the channel coefficient and the beam direction change very slowly, which can be approximated as β _k = β, x _k = x, so

At this time, the conditional probability density function of the observation vector y _k is

use

Represents a tracking strategy for channel coefficients and beam directions, considering a specific set of causal tracking strategies Ξ _S : the probe beamforming matrix W _k and system estimation for the k-th probe communication cycle

All are determined by the previously used detection beamforming matrices W ₁ , W ₂ ,...W _k-1 and historical observations y ₁ , y ₂ ,...y _k . Therefore, in the k-th probing communication cycle, the tracking problem of channel coefficient and beam direction can be formulated as:

Where constraints (8)

(8) Ensure

Is the channel vector

The unbiased estimation of, the constraint (9) ensures that the detection direction of each detection communication cycle is the last estimated direction plus three sets of fixed offsets.

In a quasi-static scenario, this embodiment designs a two-stage tracking scheme: Joint Beam and Channel Tracking (JBCT), which is specifically as follows:

(i) Coarse beam scanning. As shown in Figure 2, in a probing communication cycle, the receiver sequentially receives M×N pilots, corresponding to the observation

The detection beamforming vector is

The initial estimates of channel coefficients and beam directions are as follows:

among them

The codebook size used is M ₀ ×N ₀ (M ₀ ≥M,N ₀ ≥N),

X ⁺ ＝(X ^H X) ^-1 X ^H.

(ii) Joint tracking of channel coefficients and beam directions. In the probing period of the k-th probing communication cycle, the receiver sequentially receives 3 pilots, and the 3 probing beamforming vectors used are as follows:

among them

It is the asymptotically optimal detection direction offset when the number of antennas tends to infinity in a quasi-static scenario, as shown in Figure 3 and Table 1:

Table 1 Asymptotically optimal detection offset in quasi-static scene

Channel parameter vector estimation

Update by random Newton method:

among them

Is the tracking step. Fisher Information Matrix

Defined as follows

Set the system parameters as: antenna number M=N=8, antenna spacing

Coarse beam scanning codebook size M ₀ ＝2M, N ₀ ＝2N, pilot symbol _sp =1, transmit signal to noise ratio

The beam direction obeys uniform distribution:

The channel coefficient obeys the Rice distribution, and the Rice factor is 15dB. The average result obtained after 1000 system simulations is shown in Figure 4. It can be seen that the proposed tracking method can asymptotically reach the minimum CRLB of the channel vector mean square error, and the tracking accuracy is much higher than the existing algorithm.

2) Dynamic I type

In this scenario, the channel coefficient changes rapidly, but the beam direction changes very slowly, which can be approximated as x _k =x. In this embodiment, β _k obeys the Rayleigh distribution, and

Where ∑ _y (x) is the covariance matrix of the observation vector y _k :

J ₃ is the unit matrix of order 3, and the determinant of ∑ _y (x) is

use

Represents a tracking strategy for the beam direction. The present invention considers a specific set of causal tracking strategies Ξ _DI : the detection beamforming matrix W _k and system estimation of the k-th detection communication cycle

All are determined by the previously used detection beamforming matrices W ₁ , W ₂ ,...W _k-1 and historical observations y ₁ , y ₂ ,...y _k . Therefore, in the k-th detection communication cycle, the tracking problem of the beam direction can be formulated as:

Where constraints (19) ensure

It is an unbiased estimation of the direction parameter vector x, the constraint (20) ensures that the detection direction of each detection communication cycle is the last estimated direction plus three sets of fixed offsets

For the dynamic type I, this embodiment designs a two-stage tracking scheme: Recursive Beam Tracking (RBT), which is specifically as follows:

(i) Coarse beam scanning. As shown in Figure 5, in the first L probing communication cycles, the receiver receives M×N pilots in each probing communication cycle, corresponding to the observation

The detection beamforming vector is

The total number of detection communication cycles L at this stage is determined by the following formula

among them

The codebook size used is M ₀ ×N ₀ (M ₀ ≥M,N ₀ ≥N),

X ⁺ ＝(X ^H X) ^-1 X ^H.

(ii) Beam direction tracking. In the probing period of the k-th probing communication cycle, the receiver receives 3 pilots, and the 3 probing beamforming vectors used are as follows:

among them

It is the asymptotically optimal detection direction deviation when the number of antennas tends to infinity in the dynamic I type, as shown in Figure 6 and Table 2:

Table 2 Dynamic I-type asymptotically optimal detection direction shift

Orientation parameter vector estimation

Update by random Newton method:

among them

Is the tracking step. Fisher Information Matrix

The elements in the p-th row and s-th column (p=1, 2; s=1, 2) are given by

among them

Set the system parameters as: antenna number M=N=8, antenna spacing

The beam direction obeys uniform distribution:

φ∈(-π,π]. The result obtained by averaging after 1000 system simulations is shown in Figure 7. It can be seen that the proposed tracking method can approach the minimum CRLB of the mean square error of the direction parameter vector, and the tracking The accuracy is much higher than existing algorithms.

3) Dynamic type II

In this scenario, the channel coefficients and beam directions both change rapidly. At this time, the conditional probability density function of the observation vector y _k is

use

Represents a tracking strategy for channel coefficients and beam directions. The present invention considers a specific set of causal tracking strategies Ξ _DII : the probe beamforming matrix W _k and system estimation for the k-th probe communication cycle

Where constraints (29) ensure

Is the channel vector

The unbiased estimation of, the constraint (30) ensures that the detection direction of each detection communication cycle is the last estimated direction plus three sets of fixed offsets.

For the dynamic type II, this embodiment designs a two-stage tracking solution: Joint Beam and Channel Tracking (JBCT), which is specifically as follows:

(i) Coarse beam scanning. As shown in Figure 2, the receiver receives M×N pilots in one sounding communication cycle, which corresponds to the observation

The detection beamforming vector is

among them

The codebook size used is M ₀ ×N ₀ (M ₀ ≥M,N ₀ ≥N),

X ⁺ ＝(X ^H X) ^-1 X ^H.

(ii) Joint tracking of channel coefficients and beam directions. In the probing phase of the k-th probing communication cycle, the receiver receives 3 pilots, and the 3 probing beamforming vectors used are as follows:

among them

Given by Figure 3 and Table 1. Channel parameter vector estimation

Update by random Newton method:

among them

b _DII,k = 0.7 is the tracking step length. Fisher Information Matrix

Defined as follows

Set the system parameters as: antenna number M=N=8, antenna spacing

The direction of the arriving beam obeys a random walk: θ _k =θ _k-1 +Δ _θ ,φ _k =φ _k-1 +Δ _φ ;Δ _θ ,

The initial beam direction obeys uniform distribution:

φ ₀ ∈(-π,π], the channel coefficient obeys Gauss-Markov distribution: β _k =ρβ _k-1 +μ _k ; ρ=0.995,

The initial channel coefficient β ₀ obeys the Rice distribution, and the Rice factor is 15dB. The averaged results obtained after 1000 system simulations are shown in Figure 8. It can be seen that the proposed tracking method can achieve higher tracking accuracy. If a tolerable tracking error is set, such as e _t =0.15, the proposed method can achieve faster tracking speed.

Claims

A fast and accurate beam tracking method based on a two-dimensional phased antenna array, which is suitable for millimeter wave mobile communications. It is characterized in that it periodically works in detection and communication modes alternately. During the detection period of a detection communication cycle, the beams are sequentially Pointing to several detection directions, the beam during the communication period points to the beam direction estimated at that time.
The fast and accurate beam tracking method based on a two-dimensional phased antenna array according to claim 1, characterized in that it comprises the following steps:

Step A: According to the time-varying characteristics of the channel, the channel is divided into three types: 1) quasi-static scene; 2) dynamic type I; 3) dynamic type II, first determine which type of channel to be tracked belongs to;

Step B: In the detection period of each detection communication cycle, based on the channel type determined in step A, according to the corresponding type of beam tracking method, determine the direction of sequential detection, and obtain the channel coefficients and the channel coefficients according to the detection results and historically recorded information. Estimated beam direction;

Step C: Point the beam to the beam direction estimation obtained in Step B for communication.
The fast and accurate beam tracking method based on a two-dimensional phased antenna array according to claim 2, characterized in that, in the step A, the quasi-static scene refers to the situation where the channel coefficient and the beam direction are both changing slowly: β k ≈ β k-1 ,x k ≈x k-1 , where β k refers to the channel coefficient of the k-th probing communication cycle, and x k refers to the direction parameter vector of the k-th probing communication cycle:
x k, 1 , x k, 2 are the values of the two dimensions of the two-dimensional direction parameter vector x k , θ k , φ k respectively refer to the elevation angle and azimuth angle of the arrival beam direction of the k-th detection communication cycle, M, N Are the number of antennas in the two directions of the two-dimensional array antenna, d 1 and d 2 are the distances between adjacent antennas in the two directions of the array antenna, and λ is the millimeter wave wavelength;

Dynamic I type refers to the situation where the channel coefficient changes rapidly and the beam direction changes very slowly: β k ≠ β k-1 , x k ≈x k-1 ;

Dynamic type II refers to the situation where the channel coefficient and the beam direction change rapidly: β k ≠ β k-1 , x k ≠ x k-1 .
The fast and accurate beam tracking method based on a two-dimensional phased antenna array according to claim 2, characterized in that, in the step B, the tracking steps for a quasi-static scene are as follows:

Step 11: Obtain an initial direction estimate in the real beam main lobe range and an initial estimate of channel coefficients;

Step 12. In the detection period of each detection communication cycle, send 3 pilots in sequence to perform 3 detections in different directions. The detection direction is the last estimated direction plus three sets of fixed offsets ΔS ,1 ,Δ S,2 ,Δ S,3 ;

Step 13, based on the detection results and historically recorded information, update the estimated value using the random iteration method;

For dynamic type I, only the beam direction is tracked, and the tracking steps are as follows:

Step 21: Obtain an initial direction estimate in the real beam main lobe range;

Step 22: During the detection period of each detection communication cycle, three pilots are sent in sequence to perform three detections in different directions. The detection direction is the last estimated direction plus three sets of fixed offsets ΔDI,1 ,Δ DI,2 ,Δ DI,3 ;

Step 23, based on the detection result and historically recorded information, update the estimated value using a random iteration method;

The tracking steps for dynamic type II are as follows:

Step 31: Obtain initial direction estimates in the real beam main lobe range and initial estimates of channel coefficients;

Step 32: In the detection period of each detection communication cycle, three pilots are sent in sequence to perform three detections in different directions. The detection direction is the last estimated direction plus three sets of fixed offsets Δ DII,1 ,Δ DII,2 ,Δ DII,3 ;

Step 33, based on the detection result and historically recorded information, update the estimated value using a random iteration method.
The fast and accurate beam tracking method based on the two-dimensional phased antenna array according to claim 4, characterized in that, in the step 11, the initial estimation method adopted is beam scanning: within 1 detection communication cycle, send M ×N pilots, use these pilots to perform M×N detections in different directions, and obtain initial estimates of channel coefficients and beam directions according to the Orthogonal Matching Pursuit (OMP) algorithm; in step 12 , The three sets of fixed detection offsets are located on a ring of the two-dimensional plane 0.4≤|Δ S,i |≤0.6,i=1, 2, 3; in the step 13, the random iteration method used is random Newton law.
The fast and accurate beam tracking method based on a two-dimensional phased antenna array according to claim 5, wherein the three sets of fixed detection offsets located on the ring are optimized according to the target to be tracked, and the optimization target is the channel vector The mean square error of and the mean square error of the direction parameter vector.
The fast and accurate beam tracking method based on the two-dimensional phased antenna array according to claim 4, characterized in that, in the step 21, the initial estimation method adopted is beam scanning: in the first L detection communication cycles, every M×N pilots are sent in a probing communication cycle. Using these pilots, M×N detections in different directions are performed in each probing communication cycle. The selection of L must ensure that the M×N received in the L-th probing communication cycle The maximum signal energy of two different detection directions is greater than a certain threshold, and then the detection result of the L-th detection communication cycle is used to obtain the initial estimation of the beam direction according to the OMP algorithm; in step 22, the three sets of fixed detection offsets are located On a ring of the two-dimensional plane, 0.5≤|ΔDI ,i |≤0.7, i=1, 2, 3, and in the step 23, the random iteration method used is the random Newton method.
The fast and accurate beam tracking method based on a two-dimensional phased antenna array according to claim 7, wherein the three sets of fixed detection offsets located on the ring are optimized according to the target to be tracked, and the optimized target is a directional parameter The mean square error of the vector.
The fast and accurate beam tracking method based on the two-dimensional phased antenna array according to claim 4, characterized in that, in said step 31, the initial estimation method adopted is beam scanning: within 1 detection communication cycle, send M ×N pilots, use these pilots to perform M×N detections in different directions. According to the OMP algorithm, obtain the initial estimation of the channel coefficient and the beam direction; in the step 32, the three sets of fixed detection offsets are located on a ring of the two-dimensional plane: 0.4≤|Δ DII,i |≤0.6,i= 1,2,3; In the step 33, the random iteration method used is the random Newton method.
The fast and accurate beam tracking method based on a two-dimensional phased antenna array according to claim 9, wherein the three sets of fixed detection offsets located on the ring are optimized according to the target to be tracked, and the optimization target is the channel vector The mean square error of and the mean square error of the direction parameter vector.