A millimeter wave wireless communication beam tracking method based on special-shaped beams

Abstract

This invention discloses a beam tracking method for millimeter-wave wireless communications based on shaped beams. By using a shaped beam, beam scanning is performed while the millimeter-wave wireless communication transceiver is aligned for communication. Based on the scanning results, the current angle is estimated, and future angles are predicted. The beam coverage range is then determined, and the method is continuously iterated, thereby achieving low-overhead, high-precision beam tracking. The beam tracking method disclosed in this invention achieves a higher average achievable rate than existing methods and, under high signal-to-noise ratio conditions, can reach 96% of the ideal average achievable rate.

Description

Technical Field

[0001] The present invention relates to the technical field of low-cost and high-precision beam tracking, and in particular to a millimeter-wave wireless communication beam tracking method based on special-shaped beams. Background Art

[0002] With the advent of next-generation wireless communications and the rapid development of the Internet of Things (IoT), mobile communication networks are increasingly connected to a vast number of wireless devices, and various data services are growing at an explosive rate. Millimeter-wave communication, due to its abundant spectrum resources and extremely high transmission rates, is considered a key technology with the greatest potential and application value for future mobile communications. Furthermore, communication systems equipped with large-scale antenna arrays, supplemented by beamforming technology, can significantly increase transmission range and capacity. However, the presence of numerous antennas results in narrow beam coverage, making it difficult to maintain high-gain communication in mobile user systems. Therefore, a beam tracking method is needed to maintain acceptable communication quality.

[0003] The existing beam tracking schemes can be mainly divided into two categories: codebook scanning and filtering. The codebook scanning type mainly performs beam scanning in a specific time slot to obtain the current angle information, thereby meeting the beam tracking requirements. However, this scheme requires a specific time slot and specific energy for beam tracking, which is relatively expensive and requires at least one beam energy for tracking scanning. Existing filter-based channel Angle of Departure (AoD) estimation schemes (such as those based on Kalman filter [1] and deep learning filter) all use the beam vector directly pointing to the estimated direction as the beam for transmitting data at the next moment. However, when the AoD change within a frame is greater than the beam width of the main lobe, beam misalignment may occur; at the same time, the influence of airflow fluctuations on AoD makes the AoD measurement error relatively large, which increases the interruption probability and deteriorates the communication quality. In this regard, the literature [2] proposes a method of beam widening through prediction to reduce the interruption probability and improve communication performance. However, the new information collected by the filtering method is only the current return value. Too little new information will also lead to inaccurate estimation, resulting in a decrease in overall communication performance. To this end, we hope to propose a scanning tracking scheme that can obtain the most accurate angle estimation with the least overhead by combining the characteristics of the two current beam scanning methods. (Reference [1]: V.Va, H.VikaloandR.W.Heath, "Beam tracking for mobile millimeter wave communication systems," 2016IEEE Global Conference on Signal and Information Processing (GlobalSIP), Washington DC, USA, Dec. 2016, pp.743-747. Reference [2]: H.-L.Song and Y.-C.Ko, "Beam Alignment for High-Speed ​​UAV via Angle Prediction and Adaptive Beam Coverage," in IEEE Transactions on Vehicular Technology, vol. 70, no. 10, pp. 10185-10192, Oct. 2021.). Summary of the Invention

[0004] In light of this, the present invention aims to provide a beam tracking method for millimeter-wave wireless communications based on shaped beams. This method designs a beam that distributes energy within the beam's coverage area. Specifically, a small portion of the energy is concentrated within a small scanning area within the coverage area, while maintaining a uniform distribution of the majority of the energy across the entire beam's coverage area. Furthermore, beam scanning using this type of beam can obtain accurate AoD information. Simultaneously, the AoD of the next frame is predicted to determine the beam coverage for that time slot, thereby ensuring uninterrupted communication.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions:

[0006] A beam tracking method for millimeter wave wireless communication based on a special-shaped beam, wherein the frame structure of the beam tracking method proposed in the method includes two units, namely a beam tracking unit and a data transmission unit. t Frame beam training or traditional beam tracking to obtain L for prediction t When entering the special-shaped beam tracking, the beam scanning unit in each frame includes the following steps:

[0007] Step S1: Generate multiple special-shaped beams for the current frame based on the beam coverage determined in the previous frame;

[0008] Step S2: The base station transmits a special-shaped beam to perform beam scanning on the user, and the user feeds back the beam scanning result;

[0009] Step S3: The base station estimates the AoD of the user in the current frame;

[0010] Step S4: The base station t The estimated value of AoD of the frame and the current frame, and the mean and variance of the probability distribution of the user's AoD in the next frame are predicted;

[0011] Step S5: Determine the beam coverage of the current frame based on the mean and variance obtained in step S4 and the estimated value of the AoD of the user in the current frame obtained in step S3.

[0012] Furthermore, in step S1, a shaped beam is defined Where N is the number of antennas in the base station, represents the complex domain. The modulus of the beam gain of the shaped beam in the angular domain ω∈[-1,1] satisfies

[0013]

[0014] where θ c and B are the beam center angle and beam width corresponding to the beam coverage range determined in the previous frame, respectively. is the center angle of each shaped beam scanning range, L w is the number of shaped beams, by transmitting L w Different shaped beams are used to achieve a complete scan of the beam coverage area. w The set of center angles of the shaped beam scanning range is defined as β1 represents the beam gain modulus within the scanning range of the shaped beam, and β2 represents the beam gain modulus within the beam coverage but outside the scanning range, satisfying β1>β2 and It can be seen that the special-shaped beam forms a beam bulge with a beam gain modulus of β1 within its scanning range, which can achieve a smaller range while keeping the beam coverage unchanged. By changing the coverage of the beam bulge, multiple special-shaped beams are generated.

[0015] Furthermore, the beam gain of the shaped beam w in the ω direction is defined as

[0016]

[0017] Among them, [w] n represents the nth element of vector w; (·) H represents the conjugate transpose operation. a(N,φ) represents the array steering vector with N elements and AoD φ. a(N,φ) is specifically expressed as:

[0018]

[0019] Where λ is the carrier wavelength, d is the distance between adjacent antenna elements, (·) T Indicates the transpose operation. Usually d = λ / 2.

[0020] Furthermore, define is a matrix containing Q steering vectors, where Then the shaped beam w can be obtained by solving the following optimization problem:

[0021]

[0022] in, is the desired beam gain, and The optimization problem can be solved based on the existing method to obtain the current frame and the tth frame f The Lx shaped beams w(1), w(2), ..., w(L w ).

[0023] Furthermore, in step S2, in a millimeter wave mobile communication system consisting of a multi-antenna base station and a single-antenna user, the transmission model of beam scanning is specifically expressed as follows:

[0024]

[0025] Where y(t) is the user received signal at time t, P is the signal transmission power of the base station, The beamforming vector at the base station at time t satisfies s(t) represents the base station transmitted symbol at time t, and n(t) is a symbol with mean 0 and variance σ 2 The additive white noise of complex Gaussian distribution.

[0026] Furthermore, in the transmission model, the channel model is:

[0027]

[0028] Where L represents the total number of multipath channels between the base station and the user, α l and φ l (t) represent the complex gain and AoD of the lth path at the tth moment, respectively.

[0029] Furthermore, the step S3 uses the Bayesian formula to obtain the update rule of the probability distribution of AoD according to the prior information of the noise distribution, the channel model and the shaped beam, and iteratively updates the probability distribution in chronological order to obtain the current frame and the t-th frame. f Maximum a posteriori estimation of frame AoD

[0030] Specifically, let the signals received by Lw shaped beams in time sequence be y(1), y(2), ..., y(L w ), the corresponding shaped beams are w(1),w(2),…,w(L w ). The posterior probability distribution can be obtained as a Q-dimensional vector π(k):

[0031]

[0032] Furthermore, according to the Bayesian formula, channel model and noise distribution, the posterior probability update method is as follows:

[0033]

[0034]

[0035] Assume that t f -1 frame data transmission beam coverage is [θ rr ,θ rl ], the total number of discrete angles contained in the interval is L rTherefore, it can be assumed that mobile users are uniformly distributed within the beam coverage area, that is, π(0)=[0,…,0,a T ,0,…,0] T ,in,

[0036] Through the above two formulas, recursively update the posterior probability to obtain π(L w ). Thus, the maximum a posteriori estimate of the angle can be obtained

[0037] Furthermore, in step S4, a fitting prediction algorithm is used based on the previous L t Estimated value of the frame's AoD And the estimated value of AoD of the current frame The mean and standard deviation of the AoD of the predicted next frame are denoted as μ and ξ respectively.

[0038] Specifically, the Gaussian random process algorithm is used to predict the AoD of the next frame. The AoD model is assumed to be:

[0039] φ(t f +1)=g(φ(t f ),φ(t f -1),φ(t f -2))+η(t f )

[0040] The input of the model is from the tth f Frame to t f - 3 estimated AoDs collected from 3 frames, indicating that the motion model takes into account angular velocity and acceleration.

[0041] The entire time process is regarded as a Gaussian random process, which is determined by the mean function and the covariance matrix function, and a data point is a sampling of the Gaussian process, that is,

[0042]

[0043] Where z(x) and x are the output and input of the model respectively, μ(x) is the mean function, and k(x, x') is the kernel function. Here we use the square exponential kernel function There are two hyperparameters involved Let the likelihood function be in are the output and input of the training parameters, L tr =L t -1 is the training input length, and γ is a hyperparameter. Hyperparameters can be obtained.

[0044] Using the Bayesian formula, we can get the predicted input data x * And the posterior distribution of the training data is where g * (x * ) is the predicted output, and μ is obtained by derivation and calculation * ,cov(g * (x * ))for

[0045]

[0046]

[0047] Use the MATLAB built-in function fitrgp to implement the training of the Gaussian process regression model and obtain the mean μ and standard deviation ξ of the distribution of the next frame AoD.

[0048] Furthermore, in step S5, the beam center angle and beam width corresponding to the beam coverage range of the current frame are defined as θ c (t f ) and B(t f ), which is determined by the following expression:

[0049]

[0050]

[0051] The above expression is mainly based on the principle that the beam should cover the user motion range within a frame as much as possible and the beam should be as narrow as possible. The beam coverage range is set to include the range of the next frame AoD and the current frame AoD. In addition, considering the noise and estimation error, plus Width tolerance.

[0052] Compared with the prior art, the present invention has the following beneficial effects:

[0053] 1. Compared with traditional filtering methods, it has higher accuracy of AoD estimation, especially at high speeds;

[0054] 2. Compared with the traditional beam scanning tracking method, it reduces the energy consumption for tracking and scanning and improves energy utilization efficiency;

[0055] 3. In terms of path averaging, the communication performance is almost close to that when the path is known. BRIEF DESCRIPTION OF THE DRAWINGS

[0056] Figure 1 and Figure 2 It is a time structure diagram of the beam tracking solution in Example 1 of the present invention and an operation flow chart of the beam tracking unit.

[0057] Figure 3 This is a comparison diagram of the expected and actual special-shaped beams used in the present invention and commonly used communication beams.

[0058] Figure 4(a) is a schematic diagram of the simulation scene, and Figure 4(b) is a diagram of the user's initial position x u (0)=-20,y u Real-time AoD change curve when (0) = 10, movement direction θ = -1.47 (rad), and movement speed v = 30 m / s.

[0059] Figure 5 is a comparison of the real-time beam center angle and beam coverage of the present invention, the base station's known real-time AoD solution, the base station's known AoD and ideal beam width solution, and the traditional neighborhood beam scanning solution, when the base station is equipped with 64 array elements, the array element spacing is half a wavelength, the number of RF links is 6, the base station obtains an accurate AoD, and the total number of transmission paths between the user and the base station is 1.

[0060] Figure 6 compares the real-time signal-to-noise ratios of Example 1 of the present invention, a known real-time AoD solution for the base station, a known AoD solution with an ideal beam width, and a traditional neighborhood beam scanning solution, when the base station is equipped with 64 array elements, the array element spacing is half a wavelength, the number of RF links is 6, the base station obtains an accurate AoD, and the total number of transmission paths between the user and the base station is 1.

[0061] FIG7 is a comparison diagram of the real-time beam center angle and coverage range provided by Example 1 of the present invention and the solutions in references [1] and [2] when the base station is equipped with 64 array elements, the array element spacing is half a wavelength, the number of RF links is 6, and the total number of transmission paths between the user and the base station is 1.

[0062] FIG8 is a comparison diagram of the real-time beam center angle and coverage range provided by Example 1 of the present invention and the solutions in references [1] and [2] when the base station is equipped with 64 array elements, the array element spacing is half a wavelength, the number of RF links is 6, and the total number of transmission paths between the user and the base station is equal to 1.

[0063] Figure 9 The comparison diagram of the estimated AoD error between Example 1 of the present invention and the solution in reference [2] is shown when the base station is equipped with 64 array elements, the array element spacing is half a wavelength, the number of RF links is 6, and the total number of transmission paths between the user and the base station is equal to 1.

[0064] Figure 10 When the base station is equipped with 64 array elements, the array element spacing is half a wavelength, the number of radio frequency links is 6, and the total number of transmission paths between the user and the base station is equal to 1, the present invention and all the above schemes have the same effect on the path direction. Comparison of average achievable rates under different signal-to-noise ratios obtained by 50 uniform samplings. DETAILED DESCRIPTION

[0065] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0066] Example 1

[0067] See also Figures 1-10 This embodiment provides a low-overhead, high-precision beam tracking method for a millimeter-wave mobile wireless communication system. The method specifically includes the following steps:

[0068] like Figure 1 As shown in the time structure diagram, the whole process consists of multiple time frames, and the duration of a frame is T f , where a frame contains multiple time slots, and the length of a time slot is T s First, perform L t Frame beam training or traditional beam tracking to obtain L for prediction t Initial value. When entering the special beam tracking, Figure 2 The operation process of its beam tracking unit is demonstrated. The specific operation process includes the following steps:

[0069] Step S1: Design the beam coverage range determined in the previous frame as follows: Figure 2 The shaped beam is shown. First, define the shaped beam The beam gain in the direction ω∈[-1,1] is

[0070]

[0071] Among them, [w] n Represents the nth element of vector v.

[0072] Furthermore, the gain of the above-mentioned shaped beam based on energy optimization allocation needs to meet the following constraints:

[0073]

[0074] where θ c and B are the beam center angle and beam width corresponding to the beam coverage range determined in the previous frame, respectively. is the center angle of each shaped beam scanning range, L w is the number of shaped beams, by transmitting L wDifferent shaped beams are used to achieve a complete scan of the beam coverage. In this case, L w =4. L w The set of center angles of the shaped beam scanning range is defined as β1 represents the beam gain modulus within the scanning range of the shaped beam, and β2 represents the beam gain modulus within the beam coverage but outside the scanning range, satisfying β1>β2 and This case setting That is, for a single beam, the scanning energy allocation ratio is about 10%, and the energy for maintaining communication is about 90%. When the shaped beam is compared with the data transmission beam Figure 3 As shown in the figure, the shaped beam forms a beam bulge with a beam gain modulus of β1 within its scanning range, which can achieve a smaller range while keeping the beam coverage unchanged. By changing the coverage of the beam bulge, multiple special-shaped beams are generated.

[0075] make is a matrix containing Q steering vectors, where Then the shaped beam w can be obtained by solving the following optimization problem:

[0076]

[0077] in, is the desired beam gain, and The optimization problem can be solved based on the existing method to obtain the current frame and the tth frame f Frame L w Shaped beams w(1), w(2),…, w( L w).

[0078] Step S2: For a communication system consisting of a multi-antenna base station and a single-antenna user, the channel model of this system can be expressed as follows according to the SV model:

[0079]

[0080] Where L represents the total number of multipath channels between the base station and the user, α l and φ l (t) represent the complex gain and AoD of the lth path at time t, respectively. Assume that the number of transmitting array elements N = 64, a(N,φ) represents the array steering vector with N elements and AoD φ, and its specific expression is:

[0081]

[0082] Where λ is the carrier wavelength, d is the antenna element spacing, (·) T represents the transpose operation, usually d = λ / 2. In general, the base station can obtain the AoD through other means, so it only needs to focus on tracking φ(t).

[0083] To simplify the analysis, assume that the total number of transmission paths between the user and the base station is equal to 1 and the path loss α = 1, then the channel model can be simplified to

[0084]

[0085] The downlink signal transmission model of the base station is specifically expressed as follows:

[0086]

[0087] Where y(t) is the received signal at time t, P is the signal transmission power at the base station, The beamforming vector at the base station at time t satisfies N is the number of transmitting antennas, represents the complex field, (·) H represents the conjugate transpose operation; s(t) represents the symbol transmitted at time t. n(t) is white noise with a mean of 0 and a variance of σ 2 The complex Gaussian distribution of .

[0088] Use the shaped beams w(1), w(2), ..., w(L w ) to perform beam scanning. Through the above transmission model, the user end will receive the signals y(1), y(2),…, y(L w ), and feed it back to the base station through the uplink channel.

[0089] Step S3: The base station uses the Bayesian formula to obtain the update rule of the probability distribution of AoD, and iteratively updates the probability distribution in chronological order, starting from the feedback values ​​y(1), y(2), ..., y(L w ) to get the current frame and the tth frame f Maximum a posteriori estimation of frame AoD First, discretize the AoD, and the posterior probability distribution can be obtained as a Q-dimensional vector π(k):

[0090]

[0091] According to the Bayesian formula, channel model and noise distribution, the posterior probability update method is as follows:

[0092]

[0093]

[0094] Assume that t f -1 frame data transmission beam coverage is [θ rr ,θ rl ], the total number of discrete angles contained in the interval is L r Therefore, it can be assumed that mobile users are uniformly distributed within the beam coverage area, that is, π(0)=[0,…,0,a T ,0,…,0] T ,in,

[0095] Through the above two formulas, recursively update the posterior probability to obtain π(L w ). Thus, the maximum a posteriori estimate of AoD can be obtained

[0096] Step S4: According to the previous L t Estimated value of the frame's AoD The Gaussian process regression algorithm is used to fit and predict the mean and standard deviation of the AoD of the next frame scanning transmission unit, which are expressed as μ and ξ respectively. First, assume that the AoD model is:

[0097] φ(t f +1)=g(φ(t f ),φ(t f -1),φ(t f -2))+η(t f )

[0098] The input of the model is from the tth f Frame to t f - 3 estimated AoDs collected from 3 frames, indicating that the motion model takes into account angular velocity and acceleration.

[0099] The entire time process is regarded as a Gaussian random process, which is determined by the mean function and the covariance matrix function, and a data point is a sampling of the Gaussian process, that is,

[0100]

[0101] Where z(x) and x are the output and input of the model respectively, μ(x) is the mean function, and k(x, x') is the kernel function. Here we use the square exponential kernel function There are two hyperparameters Let the likelihood function be where z tr , are the output and input of the training parameters, L tr =L t -1 is the training input length, and γ is a hyperparameter. Hyperparameters can be obtained.

[0102] Using the Bayesian formula, we can get the predicted input data x * And the posterior distribution of the training data is where g * (x * ) is the predicted output, and μ is obtained by derivation and calculation * ,cov(g * (x * ))for

[0103]

[0104]

[0105] Use MATLAB's own function fitrgp to implement the training of the Gaussian process regression model, and set the length of the training data set to L tr = 10. Directly obtain the mean μ and standard deviation ξ of the AoD distribution of the next frame.

[0106] Step S5: The beam center angle and beam width corresponding to the current frame beam coverage range are θ c (t f ) and B(t f ), which is determined by the following expression:

[0107]

[0108]

[0109] The present invention is further described below in conjunction with simulation conditions and results:

[0110] Consider a two-dimensional planar system as shown in Figure 4(a). With the base station as the origin, the number of RF chains is N. RF =6. The array is placed along the y-axis, with the center point at the origin. Initially, the user is at (x u (0),y u (0)), the coordinate unit is m, and the user moves in a straight line along the direction θ with a speed v. θ is a random variable that obeys the uniform distribution of [-π,π). Then we can get AoD at each moment as the user position (x u ,y u ) vector and the x-axis, the AoD time-varying function is obtained as

[0111]

[0112]

[0113] Let the initial user position xu (0)=-20,y u (0) = 10. To better demonstrate the performance of the solution, the simulation is performed with the angular velocity as large as possible. Therefore, the user motion direction θ is set to -1.47 (rad) and the motion velocity v is set to 30 m / s. The AoD dispersion Q is set to 512. The transmission signal-to-noise ratio is SNR0 = 15 dB. The total duration T is set to 1 s and the frame length is set to T. f =10ms, the data transmission block length is T s =1ms, the amount of Gaussian regression data is L tr =10, beam training frame length L t = 11. The obtained AoD real-time change diagram is shown in Figure 4(b).

[0114] Figures 5(a) and 5(b) compare the real-time beam center angle and beam coverage of the present invention with the ideal upper limit, robust ideal scheme, and neighborhood beam scanning scheme under low and high angular velocities. Note that the present invention is able to achieve effective tracking throughout the entire process and achieve high-precision estimation at the estimated time slot. In addition, regardless of the low angular velocity stage or the high angular velocity stage, the beam coverage is close to the ideal coverage, confirming the effectiveness of the prediction scheme. Relative to the domain search, it has a more accurate estimate, which is due to the low resolution of the neighborhood search codebook, but increasing its resolution will result in higher energy overhead.

[0115] Figures 6(a) and 6(b) compare the real-time signal-to-noise ratio of the present invention with that of the ideal upper limit, the robust ideal solution, and the neighborhood beam scanning solution under low and high angular velocities. The real-time signal-to-noise ratio of the present invention approaches that of the robust ideal solution during the entire tracking process, while the real-time signal-to-noise ratio of the neighborhood beam scanning solution is frequently interrupted due to movement, triggering neighborhood beam scanning, especially in areas with high angular velocities, which reduces communication performance. Specifically, it can be obtained that the time average achievable sum rates of the present invention, the ideal upper limit, the robust ideal solution, and the neighborhood beam scanning solution are 4.5973, 5.0388, 4.7167, and 3.7776, respectively, in bit / s / Hz. The present invention approaches the robust ideal solution under high-speed motion, that is, the gap with the ideal upper limit is mainly caused by beam broadening, and is significantly better than the neighborhood beam scanning solution.

[0116] Figure 7(a) and Figure 7(b) compare the real-time beam center angle and beam coverage of the present invention, the solution in reference [1], and the solution in reference [2] under low and high angular velocities. Reference [1] is a Kalman filter method, and reference [2] is a Kalman filter method with beam broadening. In the early stage of low angular velocity, the present invention and the solution in reference [2] show similar AoD tracking performance and equally accurate beam coverage, but the solution in reference [1] shows relatively poor AoD estimation performance. This is because the solution in reference [1] needs to predict the AoD coverage before performing beam adjustment, while reference [2] performs beam adjustment for each frame. However, in the mid-term high angular velocity, the present solution is significantly better than reference [2]. The coverage of reference [2] is significantly too wide. From the analysis of Figure 7 below, it can be seen that this is due to inaccurate AoD estimation.

[0117] Figures 8(a) and 8(b) compare the real-time signal-to-noise ratios of the present invention, the solution in reference [1], and the solution in reference [2] under low and high angular velocities. As shown in Figure 7(a), under low angular velocities, due to the real-time adjustment of the beam, both reference [2] and the present invention do not trigger the interruption of the rescanning process, while the solution in reference [1] has a small amount of interrupted scanning in the later stage. As the angular velocity increases, both reference [1] and reference [2] have frequent interrupted scanning, while when the estimation was accurate before, there was no frequent triggering, indicating that this is caused by inaccurate AoD estimation. Specifically, it can be obtained that the time average achievable sum rates of the present invention, reference [1], and reference [2] are 4.5973, 4.0366, and 3.6166, respectively, in units of bit / s / Hz. The present invention is significantly superior to the solutions in reference [1] and reference [2] under high-speed motion, mainly because the introduction of the special-shaped beam scanning improves the accuracy of AoD estimation. The performance of reference [2] is worse than that of reference [1] because when the beam width is selected too large, the signal-to-noise ratio is lower than the trigger interrupt scanning threshold, causing it to be triggered frequently, resulting in performance degradation.

[0118] Figure 9 The AoD estimation errors of the present invention and the solution in reference [2] are compared. It is clear from the figure that the present invention maintains high-precision estimation during the entire tracking process, while the solution in reference [2] has a huge error when the angular velocity is large in the middle period. Specifically, the maximum error of the present invention during the tracking process is 0.0034, which is about a discrete degree. The maximum AoD estimation error in reference [2] is 0.1065.

[0119] Figure 10The average achievable sum rates of all the above schemes under different SNR conditions are compared. It is noted that when the SNR is higher than 15dB, that is, under high SNR conditions, the scheme proposed in this section approaches the achievable sum rate of the ideal case, reaching 96% of the ideal performance and almost close to the robust ideal scheme. At the same time, it is much higher than the method in reference [1], the method in reference [2] and the neighborhood search method. That is, under high SNR, the estimated AoD accuracy is high, which makes the predicted AoD more accurate, and the variance of the obtained prediction is reduced, making the set beam coverage close to the ideal beam coverage. When the SNR is lower than 15dB, that is, under low SNR conditions, the achievable sum rate of the proposed scheme is slightly lower than that of the higher SNR condition, and the maximum difference from the ideal condition is about 0.5bit / s / Hz. This is because the low SNR affects the estimation accuracy, but it still shows better performance than the neighborhood and reference [1] schemes.

[0120] Anything not described in detail in the present invention is well known to those skilled in the art.

[0121] The above describes in detail the preferred embodiments of the present invention. It should be understood that those skilled in the art can make numerous modifications and variations based on the concepts of the present invention without inventive effort. Therefore, any technical solutions that can be derived by those skilled in the art through logical analysis, reasoning, or limited experimentation based on the concepts of the present invention and the prior art should be within the scope of protection defined by the claims.

Claims

1. A millimeter wave wireless communication beam tracking method based on a special-shaped beam, characterized in that: In a millimeter wave mobile communication system consisting of a multi-antenna base station and a single-antenna user, the base station determines the beam coverage range determined by the previous frame and the previous L t The estimated value of the user channel departure angle of the frame is used to estimate the beam coverage of the current frame and the channel departure angle of the user in the current frame, so as to achieve low-cost and high-precision beam tracking for the user. t positive integer; It includes the following 5 steps: Step S1: Generate multiple special-shaped beams for the current frame based on the beam coverage determined in the previous frame; define the special-shaped beams Where N is the number of antennas in the base station, Representing the complex domain, the modulus of the beam gain of the shaped beam in the angular domain ω∈[-1,1] satisfies where θ c and B are the beam center angle and beam width corresponding to the beam coverage range determined in the previous frame, respectively. is the center angle of each shaped beam scanning range, L w is the number of shaped beams, by transmitting L w Different shaped beams can realize complete scanning of the beam coverage area. w The set of center angles of the shaped beam scanning range is defined as β1 represents the beam gain modulus within the scanning range of the shaped beam, and β2 represents the beam gain modulus within the beam coverage but outside the scanning range, satisfying β1>β2 and Step S2: The base station transmits a special-shaped beam to perform beam scanning on the user, and the user feeds back the beam scanning result; Step S3: The base station estimates the channel departure angle of the user in the current frame; Step S4: The base station t The estimated value of the user channel departure angle of the frame and the current frame, and the mean and variance of the probability distribution of the user's channel departure angle in the next frame are predicted; Step S5: Determine the beam coverage of the current frame based on the mean and variance obtained in step S4 and the estimated value of the channel departure angle of the user in the current frame obtained in step S3.

2. The millimeter wave wireless communication beam tracking method based on a special-shaped beam according to claim 1, characterized in that: The beam gain of the shaped beam w in the ω direction is defined as Among them, [w] n represents the nth element of vector w; (·) H represents the conjugate transpose operation, a(N,φ) represents the array steering vector with N array elements and a departure angle of φ. a(N,φ) is specifically expressed as: Where λ is the carrier wavelength, d is the distance between adjacent antenna elements, (·) T Represents the transpose operation; definition is a matrix containing Q steering vectors, where Then the shaped beam w is obtained by solving the following optimization problem: in, is the desired beam gain, and Ψ=[ψ1,…,ψ q ] T .

3. The millimeter wave wireless communication beam tracking method based on a special-shaped beam according to claim 2, characterized in that: The transmission model of the beam scanning in step S2 is specifically expressed as follows: Where y(t) is the user received signal at time t, P is the signal transmission power of the base station, The beamforming vector at the base station at time t satisfies s(t) represents the base station transmitted symbol at time t, and n(t) is a symbol with mean 0 and variance σ 2 The additive white noise of complex Gaussian distribution.

4. The millimeter wave wireless communication beam tracking method based on a special-shaped beam according to claim 3, characterized in that: In the transmission model described, the channel model is: Where L represents the total number of multipath channels between the base station and the user, α l and φ l (t) represent the complex gain of the lth path and the channel departure angle at the tth moment respectively.

5. The millimeter wave wireless communication beam tracking method based on shaped beam according to claim 4, characterized in that: The step S3 uses the Bayesian formula to obtain the update rule of the probability distribution of the channel departure angle according to the prior information of the noise distribution, the channel model and the shaped beam, and iteratively updates the probability distribution in chronological order to obtain the current frame and the t-th frame. f Maximum a posteriori estimation of the frame channel departure angle 6. The millimeter wave wireless communication beam tracking method based on shaped beam according to claim 5, characterized in that: The step S4 adopts the fitting prediction algorithm based on the previous L t Estimated value of the channel departure angle of the frame And the maximum a posteriori estimate of the channel departure angle of the current frame The mean and standard deviation of the channel departure angle for the next frame are predicted, denoted as μ and ξ respectively.

7. The millimeter wave wireless communication beam tracking method based on shaped beam according to claim 6, characterized in that: In step S5, the beam center angle and beam width corresponding to the beam coverage range of the current frame are defined as θ c (t f ) and B(t f ), which is determined by the following expression:

Images