Robust beamforming method and device for unmanned aerial vehicle carrying intelligent reflecting surface based on wahba attitude reconstruction under wind disturbance

Abstract

The application discloses a robust beamforming method and device based on Wahba attitude reconstruction under wind disturbance, which adopts a two-stage scheme of "off-line static design + on-line dynamic compensation". In the off-line stage, the problem of maximum energy efficiency is solved under the nominal attitude to obtain the analog / digital beamforming matrix of the base station and the static phase shift matrix of the RIS. In the on-line stage, the instantaneous attitude of the unmanned aerial vehicle is reconstructed through the singular value decomposition (SVD) closed-form solution of the Wahba problem by using the multi-source sensor observation vector per time slot; then, the direction deviation of the incident and reflection links is calculated, and the compensation phase of each unit of the RIS is calculated based on the geometric relationship; finally, the compensation phase shift diagonal matrix is generated according to the fusion compensation phase, and the compensation phase shift diagonal matrix is multiplied by the static phase shift matrix to form the total phase shift matrix of the current time slot to correct the beam pointing. The application does not need training data, has low computational complexity, can effectively compensate the beam mismatch caused by wind disturbance, and improves the robustness and energy efficiency of the communication system.

Description

Technical Field

[0001] This invention relates to the fields of unmanned aerial vehicle (UAV) assisted wireless communication, marine Internet of Things (IoT), reconfigurable intelligent surface (RIS), hybrid beamforming, and robust beamforming, and in particular to a beamforming technology for dealing with UAV attitude jitter. Background Technology

[0002] With the rapid development of marine resource development and environmental monitoring applications, the demand for all-weather, high-bandwidth marine Internet of Things (IoT) is becoming increasingly urgent. Utilizing unmanned aerial vehicles (UAVs) equipped with intelligent reflective surfaces (RIS) to construct aerial relay platforms and provide millimeter-wave communication services to maritime users has become an important technological approach. In the millimeter-wave band, precise narrow beamforming is key to achieving high array gain and reliable communication.

[0003] However, when performing missions, especially in maritime environments, drones are subject to continuous sea wind disturbances. These disturbances trigger high-frequency fine-tuning of the drone's flight control system, resulting in minute angular jitter of 0.5°–3° and 10–50 Hz. Since the onboard RIS is fixed to the drone's body, this minute attitude jitter directly alters the direction of the narrow beam, causing beam mismatch with the target user and severely degrading the cascaded channel quality, thereby significantly reducing the performance of the communication system.

[0004] In existing technologies, the main approaches to address this problem include: 1) improving flight control algorithms to enhance attitude stability, but their control cycle (approximately 20ms) is much longer than the response time (sub-millisecond level) required for beam compensation, making it difficult to meet real-time requirements; 2) designing anti-interference beamforming schemes on the communication side, but usually simplifying jitter to random angle shifts, failing to accurately characterize the physical coupling relationship between wind disturbance and flight control response, resulting in limited compensation accuracy; 3) adopting joint design of communication and flight control or learning-based methods, but these methods often rely on a large amount of training data, which limits their practical deployment. Summary of the Invention

[0005] The present invention aims to solve the technical problem in the prior art where wind disturbance causes drone attitude jitter, which in turn leads to millimeter-wave narrow beam mismatch and reduced communication quality.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a robust beamforming method for UAVs equipped with intelligent reflective surfaces under wind disturbance based on Wahba attitude reconstruction, applied to a wireless communication system assisted by an intelligent reflective surface on a UAV. The system includes a base station equipped with an antenna array and employing an analog / digital hybrid beamforming architecture, a UAV equipped with a passive intelligent reflective surface, and multiple users. The intelligent reflective surface is fixedly connected to the UAV body. The method includes an offline static stage and an online dynamic stage. Offline static phase: Establish an inertial coordinate system and a UAV body coordinate system. Determine the nominal attitude matrix based on the statistical average of the attitude jitter angles under wind disturbance. Calculate the base station to smart reflective surface channel, the smart reflective surface to each user channel, and the base station to each user direct connection channel under the nominal attitude. Under the nominal attitude matrix, an energy efficiency maximization beamforming optimization problem is established and solved, and the simulated beamforming matrix, digital beamforming matrix and intelligent reflective surface static phase shift matrix are output, which remain unchanged during the communication service period. During the online dynamic phase, execution is performed cyclically, one time slot at a time. Obtain multiple pairs of unit direction observation vectors and their corresponding confidence weights. The unit direction observation vectors are unit vectors that express the same physical direction in the inertial coordinate system and the body coordinate system respectively, and the observation directions are not collinear. A pose matrix is ​​constructed by weighting the observation vectors, a singular value decomposition is performed on it, and a reconstructed pose matrix is ​​obtained based on the closed-form solution of the Wahba problem. The closed-form solution includes a diagonal correction factor to ensure that the reconstructed pose matrix belongs to a three-dimensional special orthogonal group. Based on the reconstructed attitude matrix, the nominal attitude matrix, and the line-of-sight direction vectors in the position coordinates or inertial coordinate systems of the base station, the UAV, and each user, the incident link direction deviation from the base station to the intelligent reflective surface and the reflection link direction deviation from the intelligent reflective surface to each user are calculated respectively. Based on the geometric inner product relationship between the position vector of each reflective unit of the intelligent reflective surface in the body coordinate system and the direction deviation, the compensation phase of each reflective unit for each user is calculated in a closed loop. A unified compensation phase is obtained by fusing the compensation phases of multiple users, a compensation phase shift diagonal matrix is ​​generated, and it is multiplied with the offline static phase shift matrix to form the total phase shift matrix of the intelligent reflective surface in the current time slot. The matrix multiplication is equivalent to adding the static phase angle of each reflective unit to the compensation phase angle. Downlink communication in the current time slot is performed using the analog beamforming matrix, the digital beamforming matrix, and the total phase shift matrix of the smart reflective surface in the current time slot.

[0007] Furthermore, the nominal attitude matrix is ​​set according to the statistical average value of the attitude jitter angle under wind disturbance, and the attitude uncertainty is characterized by a set based on the Frobenius norm. The set is defined as a subset of a three-dimensional special orthogonal group centered on the nominal attitude matrix and with the attitude jitter radius as the norm sphere. In the online stage, the deterministic instantaneous attitude estimate of the current time slot is obtained through the closed-form solution of the Wahba problem. The deterministic instantaneous attitude estimate replaces the time-slot-by-time processing of the attitude uncertainty set in the original robust joint optimization problem, and deterministic geometric phase compensation is performed for the current time slot. Thus, the semi-infinite worst-case constraint in the original robust joint optimization problem is transformed into a deterministic online compensation problem based on reconstructed instantaneous attitude.

[0008] Furthermore, the energy efficiency maximization beamforming optimization problem aims to maximize system energy efficiency, where system energy efficiency is the ratio of system power efficiency to rate of power consumption. Constraints include a service quality constraint that the signal-to-interference-plus-noise ratio (SNR) of each user is not lower than a threshold, a power constraint that the total transmit power is not greater than the maximum transmit power, a constant mode constraint for simulated beamforming, and a constant mode constraint for the intelligent reflector surface. The problem is solved using fractional programming and alternating optimization iterations, including: introducing a Dinkelbach energy efficiency parameter to transform the fractional objective into a subtractive form; introducing an auxiliary variable to transform the rate function into an equivalent lower bound, and updating the auxiliary variable in a closed loop when the beamforming variable is fixed; and fixing the simulated beamforming... The digital beamforming subproblem is transformed into a convex optimization problem and solved by fixing the digital beamforming matrix and the static phase shift matrix of the intelligent reflective surface. The simulation beamforming matrix is ​​updated by a successive convex approximation first-order expansion and the closed phase projection rule. The simulation beamforming matrix is ​​updated by a successive convex approximation first-order expansion and the closed phase projection rule. When the energy efficiency change rate of two adjacent rounds is less than the preset convergence threshold or the maximum number of iterations is reached, the iteration stops and the final simulation beamforming matrix, digital beamforming matrix and intelligent reflective surface static phase shift matrix are output.

[0009] Furthermore, the offline static phase also includes a power back-substitution mechanism: when the iteratively updated solution does not meet the signal-to-interference-plus-noise ratio (SINR) constraint, the digital beamforming subproblem is resolved and the transmit power is increased until the SINR threshold is met or the maximum transmit power is reached.

[0010] Furthermore, the calculation of the direction deviation includes: calculating the line-of-sight vector from the inertial frame to the machine system under the nominal attitude and the line-of-sight vector from the inertial frame to the machine system under the instantaneous attitude; converting the nominal and instantaneous line-of-sight vectors into unit direction vectors through unit direction mapping, and subtracting them to obtain the incident link direction deviation and the reflection link direction deviation; when the attitude deviation is within a small angle range of ±10°, the elevation angle deviation and azimuth angle deviation are directly calculated from the line-of-sight vector deviation using a linearized approximation formula, avoiding inverse trigonometric function calculations.

[0011] Furthermore, the fusion of multi-user compensation phases adopts an equal-weighted averaging criterion; or a weighted non-uniform fusion criterion with the channel quality of each user as the weight.

[0012] Furthermore, the unit direction observation vector is derived from global navigation satellite system carrier phase measurement, airborne inertial measurement unit, or visual odometry, and the confidence weight is set according to the measurement accuracy or signal-to-noise ratio of the corresponding sensor.

[0013] This invention also provides a robust beamforming device for a UAV equipped with an intelligent reflective surface based on Wahba attitude reconstruction under wind disturbance, comprising: The offline optimization module is used to establish the inertial coordinate system and the UAV body coordinate system, determine the nominal attitude matrix and calculate the channel under the nominal attitude, solve the nominal attitude energy efficiency maximization problem offline, and output the simulated beamforming matrix, digital beamforming matrix and intelligent reflective surface static phase shift matrix, which remain unchanged during the communication service period. The attitude reconstruction module is used to acquire multiple pairs of unit direction observation vectors and confidence weights in each time slot, construct the Wahba attitude matrix and perform singular value decomposition on it, and reconstruct the instantaneous attitude matrix based on the closed decomposition with diagonal correction factor. The orientation deviation calculation module is used to calculate the orientation deviation of the incident link from the base station to the smart reflective surface and the orientation deviation of the reflected link from the smart reflective surface to each user based on the reconstructed attitude matrix, the nominal attitude matrix, and the line-of-sight orientation vectors in the position coordinates or inertial coordinate system of the base station, the UAV, and each user. The phase compensation and fusion module is used to calculate the compensation phase in a closed loop based on the geometric inner product relationship between the position vector of each reflection unit of the smart reflective surface and the direction deviation, fuse the compensation phase of multiple users to generate a compensation phase shift diagonal matrix, and multiply it with the static phase shift matrix to form the total phase shift matrix of the smart reflective surface in the current time slot. The matrix multiplication is equivalent to adding the static phase angle of each reflection unit to the compensation phase angle, and outputting it to the smart reflective surface controller.

[0014] Furthermore, the device is deployed on the base station side, where the base station completes offline optimization and online compensation calculations and then sends the total phase shift matrix of the intelligent reflective surface to the intelligent reflective surface controller on the UAV via the control link; or, the device is deployed on the UAV side, where the onboard computing unit completes attitude reconstruction and phase compensation calculations locally.

[0015] The present invention also provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program, when executed by a processor, implements the above-described robust beamforming method based on Wahba attitude reconstruction under wind disturbance.

[0016] The present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the robust beamforming method based on Wahba attitude reconstruction under wind disturbance described above.

[0017] Compared with the prior art, the beneficial effects of the present invention are: 1. Strong robustness to wind disturbance: The instantaneous attitude is reconstructed in real time through Wahba-SVD closed-form decomposition, which transforms the attitude uncertainty into a deterministic attitude estimate for the current time slot. Then, phase compensation is performed through pure geometric closed-form calculation, which can quickly correct the beam pointing and reduce beam mismatch and channel degradation caused by attitude jitter.

[0018] 2. Deployment-friendly: The online compensation phase relies solely on sensor observations and deterministic geometric calculations, requiring no offline training data. This avoids the complexity of data collection and model training, resulting in a low barrier to engineering deployment.

[0019] 3. Good scalability: The closed-loop phase compensation method of the present invention has low computational complexity and is applicable to passive RIS. Under the passive RIS power consumption model adopted, increasing the RIS scale can improve the reflection gain without introducing additional active emission power consumption of the reflection unit, thus having good scalability.

[0020] 4. Low computational complexity and high real-time performance. In the online phase, only three deterministic calculations are performed per time slot: "attitude reconstruction - direction deviation calculation - closed phase compensation". No iterative optimization is required, and it has the potential to achieve millisecond-level beam compensation with low complexity. Attached Figure Description

[0021] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the UAV-assisted marine IoT communication system under sea breeze disturbance according to the present invention; Figure 2 This is a flowchart of the offline static beamforming design process of the present invention; Figure 3 This is a flowchart of the online attitude reconstruction and phase compensation process of the present invention; Figure 4 This is a flowchart of the real-time geometric phase compensation algorithm based on attitude reconstruction of the present invention; Figure 5 This is a diagram showing the functional modules of the device of the present invention; Figure 6 The figure shows the simulation results of the energy efficiency of the present invention as a function of maximum transmission power. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] System Architecture and Signal Model This invention is applied to a marine IoT downlink communication system equipped with RIS for use by unmanned aerial vehicles (UAVs) under wind disturbance conditions. For example... Figure 1 As shown, the system includes: equipped with root antenna and A base station with a single RF chain and employing a uniform array and analog / digital hybrid beamforming architecture; a drone equipped with a passive RIS (Radio Resonance Array), where the RIS is fixed to the airframe. The system consists of a uniform planar array of cells and K sea surface users. Communication between the base station and the sea surface users is via a cascaded reflection link ("base station → RIS → user") and a direct line-of-sight (LoS) link.

[0024] Establish an East-North-Up (ENU) inertial coordinate system With the UAV body coordinate system . No. Time slots are arrive The attitude rotation matrix is ​​denoted as ,satisfy and Wind disturbances cause the flight controller's PID high-frequency fine-tuning to produce slight angular jitter in the airframe. The time-slot variation and alteration of the machine system's line-of-sight vector projection cause narrow beam pointing deviation and cascaded channel perturbation.

[0025] Wind disturbance is converted into airframe drag torque through aerodynamic action. Under the combined action of drag torque and flight control torque, discrete attitude dynamics drive angular velocity evolution, which is then mapped to time-slot updates of the attitude rotation matrix. Attitude rotation causes coordinate transformation of the line-of-sight vector within the airframe, directly altering the beam pointing observed by the airborne array.

[0026] In the The first time slot, the first The received signal for each sea surface user is: in, To send to the Data symbols for each user; Simulate the beamforming matrix for the base station; Digital beamforming matrix The List; For variance Additive complex Gaussian white noise.

[0027] Subject to posture shaking Impact, the first The equivalent concatenated channel vector for each user is: in, The time-varying channel matrix from the base station to the UAV, with its steering vector coupling The resulting rotational deviation, For drones and the first The channel response vectors between users are both... The function; This is the RIS reflection phase shift diagonal matrix; For base station and the The direct LoS channel vectors between users are considered constant because the base station and the users on the sea surface are relatively stationary.

[0028] The signal-to-interference-plus-noise ratio (SINR) of the kth user is: In the formula, the numerator is the desired signal power, and the denominator is the sum of interference and noise power; the input is the equivalent cascaded channel and hybrid beamforming variables, and the output is the scalar SINR, which is the basis for constraint C1 verification and rate calculation, corresponding to steps S3 and S11.

[0029] The system's achievable rate is the sum of the rates of all users: In the formula, For the first The downlink transmission rate for each user (in bps), B is the system bandwidth (Hz); the input is the SINR of each user, and the output is... The total system rate constitutes the numerator of the energy efficiency target, corresponding to step S3.

[0030] Two-stage variables, attitude uncertainty set and robust energy efficiency maximization problem This invention divides the variables to be designed into two categories: static variables, including... , With RIS static phase shift matrix The design is performed offline and remains unchanged throughout the communication service period; the online variable is the compensation phase vector. With the compensation phase shift diagonal matrix Each time slot is calculated in real time using a purely geometric closed-form formula. The total phase shift matrix of the RIS for the current time slot is: .

[0031] Attitude jitter is characterized using an uncertainty set based on the Frobenius norm: in, This is the nominal attitude matrix; The attitude jitter radius reflects the combined effect of sea wind intensity and hovering accuracy. Geometrically, it represents the true attitude. lie in manifold Centered on Within the norm sphere of radius; the input is attitude jitter statistics, and the output is the range of attitude uncertainty, corresponding to step S2.

[0032] With the goal of maximizing system energy efficiency, a robust optimization problem (P1) is established by jointly optimizing the two-stage hybrid beamforming and RIS phase shift: in, For system energy efficiency; For the first Transmit power of each user; This represents the total power consumption of the system's hardware circuitry. ,in , , , These represent baseband power consumption, power consumption of a single active antenna, power consumption of a single RF link, and power consumption of a single RIS reflector unit (for passive RIS, the power consumption is...). ); The SINR threshold; Let C be the maximum transmit power of the base station. Constraint C1 is a robust quality of service constraint, requiring... The worst-case SINR for each user under any internal posture is no less than C2 is the total transmit power constraint; C3 is the simulated beamforming constant mode constraint; C4 is the RIS constant mode constraint.

[0033] The aforementioned problem (P1) is non-convex, strongly coupled, and contains semi-infinite constraints, making direct solutions difficult. Therefore, this invention decomposes (P1) into an offline static stage and an online dynamic stage.

[0034] Offline static phase Step S1: Obtain system parameters and nominal location information. Input: Base station, UAV, and RIS array parameters ( , , wait), User location Maximum transmit power With power consumption parameters ( , , , SINR threshold Including wind disturbance statistics; Processing: Summarizing into a set of parameters required for offline design; Output: System parameter set and wind disturbance statistics.

[0035] Step S2: Establish a coordinate system and determine the nominal attitude. Establish an inertial coordinate system. With body coordinate system The nominal attitude matrix is ​​determined by statistically averaging the attitude jitter angles. And calculate the channel of each link under the nominal attitude. , and Output: With the nominal channel.

[0036] Step S3: Solve the nominal attitude energy efficiency maximization problem offline. In the nominal attitude... Next, the k The nominal equivalent channel for each user is: In the formula, , The base station to UAV channel matrix and the UAV to the first UAV channel matrix under nominal attitude are given. User channel vector (input); This is the static phase shift matrix of RIS; Let be the nominal equivalent channel of the output. ,make For the first The equivalent transmit beam for each user is then labeled with the nominal SINR as follows: , For the desired signal power, For interference and noise power, nominal rate .

[0037] The offline static design problem is represented as (P2): In the formula, the optimization variable is a static matrix. , , ,Target The system energy efficiency is defined by the nominal attitude. Compared to (P1), C1 degenerates into the nominal attitude SINR constraint, thus eliminating the semi-infiniteness. (P2) remains a non-convex problem, with difficulties arising from the fractional objective, the coupling of SINR terms, and the constant modulus constraint. Therefore, it is solved iteratively using fractional programming and alternating optimization according to S3-3 to S3-8. This problem remains non-convex and is solved using fractional programming and alternating optimization iteratively. S3-1 Initialization: Randomly initialize the variables that satisfy the constant modulus constraints C3 and C4. and It is initialized based on the nominal channel using the Maximum Ratio Transmission (MRT) or Zero Forcing (ZF) criteria. Set the iteration number Input: Nominal channel and system parameters; Output: Feasible initial point .

[0038] S3-2 Nominal equivalent channel and problem (P2) establishment.

[0039] S3-3 Dinkelbach fractional programming transformation: Introducing energy efficiency parameters This transforms the fractional objective into a subtraction form. After each round of iterations, calculate the solutions for each user based on the current solution. and And update η^(t+1) = In the formula, the input is the first... Wheel rate and equivalent beam The output is When the objective value converges to near zero, the current solution satisfies the first-order optimality condition of the energy efficiency fractional problem.

[0040] S3-4 Logarithmic Fractional Programming Auxiliary Variables: Introducing Non-negative Auxiliary Variables With multiple auxiliary variables Using logarithmic fractional programming transformation to Write down the equivalent lower bound. (Fixed) , , When the auxiliary variable has a closed-form optimal solution: In the formula, Get the current nominal SINR; It is the ratio of weighted channel gain to interference noise power.

[0041] S3-5 Digital Beamforming Bump Problem: Fixed and renew When, the objective function is about A concave quadratic function. Using the current point as an example for C1. By approximating the desired signal power with a first-order lower bound, the digital beamforming problem is transformed into a standard second-order conical programming / convex quadratic programming problem to be solved.

[0042] S3-6 Simulated Beamforming Closed Phase Update: Fixed and renew We employ a successive convex approximation of a first-order expansion. The m Each element is updated according to the closed-phase rule as follows: , In the formula, To take the complex argument; coefficient Make the update result satisfy the constant modulus constraint C3; Represents the vector of the first 1 element. Input: and , Output: .

[0043] S3-7 RIS Static Phase Shift Closed-Loop Update: Fixed and renew Using a successive convex approximation and closed-phase projection strategy similar to S3-6, the RIS reflection coefficient is updated according to the closed-phase: In the formula, the modulus of the updated result is always 1, which automatically satisfies the constant modulus constraint C4; For the reflection unit index. Input: and , Output: .

[0044] S3-8 Power Back-Substitution Mechanism: If the updated solution does not satisfy SINR constraint C1, then the digital beamforming problem is resolved and the transmit power is appropriately increased until it is satisfied. ( (or reach maximum transmission power) .

[0045] S3-9 Convergence Criterion and Output: When the energy efficiency of two adjacent rounds meets the requirements... ( (For the convergence threshold) or the number of iterations reached Stop iteration and output when the time comes. , and It remains unchanged during the online dynamic phase.

[0046] Online dynamic phase The deterministic reconstruction posture is obtained through steps S4 to S6. Afterwards, the set of uncertainties The online phase is relaxed, and online design only needs to make deterministic compensation for the refactoring posture.

[0047] Step S4: In the current time slot n Acquired by airborne multi-source sensors For a three-dimensional unit direction observation vector { ( ) and the corresponding confidence weights .in, For the first The physical directions in the inertial coordinate system unit vector in For the same physical direction in the body coordinate system Unit vector in; logarithm of observations Furthermore, the observation directions are not collinear; The accuracy or signal-to-noise ratio of the corresponding sensor is determined. The multi-source sensor may be, for example, a GNSS carrier phase measurement device, an IMU, or a visual odometry system.

[0048] Step S5: Construct the Wahba attitude matrix. Sum the outer products of each pair of observation vectors using confidence weights: The pose estimation problem is characterized by the Wahba weighted least squares objective, i.e., obtaining the pose matrix. This problem is equivalent to minimizing the weighted squared error between inertial frame observations and rotated machine frame observations. Trace maximization problem under constraints: In the formula, Represents the trace of a matrix. This represents the set of three-dimensional rotation matrices that satisfy orthogonality constraints and have a determinant of 1; the input to this expression is the attitude matrix. The output is the optimal attitude matrix. The physical meaning is to select the rotation that best matches the observed information among all legal rotations.

[0049] Step S6: Singular value decomposition closed-form reconstruction of instantaneous attitude. Perform singular value decomposition Reconstruct the instantaneous attitude matrix using the closed-form formula: Among them, the diagonal correction factor ensure That is, it satisfies both orthogonality constraints and determinant constraints, thus avoiding the inclusion of reflection solutions with a determinant of -1.

[0050] Step S7: Calculate the nominal and instantaneous beam directions. Calculate the line-of-sight vector of the nominal camera system separately. In the formula, Inertial coordinate system The Loss vector in the middle, For the first The location of each user (stationary). For the first The location of time-slot drones The nominal attitude matrix is ​​represented by the horizontal bar symbol. Indicates the nominal value; With instantaneous system line-of-sight vector , In the formula, the upper sharp corner symbol This represents the real-time estimate based on the reconstructed attitude; the input to this equation is the Loss-of-Speed ​​(LoS) vector of the inertial frame. With reconstructed posture The output is the instantaneous Loss vector of the system. The physical meaning is the first under wind-disturbed attitude. The actual pointing vector of each user relative to the airborne array; The nominal / instantaneous elevation and azimuth angles are extracted. Similarly, the corresponding vectors and angles for the base station direction are obtained.

[0051] Step S8: Calculate the dual-link directional deviation. Substitute the nominal and instantaneous directions into the unit direction mapping. In the formula, It is a three-dimensional unit direction vector; the input of this mapping is the elevation angle and the azimuth angle, and the output is the unit direction vector in the body coordinate system. Its physical meaning is a bridge for the conversion from the angle domain to the vector domain. The difference is then calculated to obtain the incident link direction deviation. Deviation from the reflection link direction vector of each user .

[0052] When the attitude deviation is within a small angular range (usually no more than ±10°), a linearized fast path can be used to directly calculate the angle deviation, avoiding the need for trigonometric function inversion. in, For Loss vector deviation, , , These are its three orthogonal components; the inputs of the above two equations are the nominal and instantaneous Loss vectors, and the output is a linear approximation of the angle deviation, which physically means that algebraic operations replace trigonometric inverse function operations under small angle conditions.

[0053] Step S9: Closed-loop calculation of the compensation phase for each element. RIS m The positions of the reflective units in the body coordinate system are: After the attitude shift, the first m The unit for the first k The compensation phase for each user is: In the formula, For the first Time Slot RIS m The first reflective unit targets the first The compensation phase for each user; the physical meaning of this formula is: the compensation phase of the m-th element is equal to the phase corresponding to the projection of the sum of the incident link direction deviation and the reflected link direction deviation onto the element's position, i.e., the distribution of the path difference caused by the dual-link pointing offset along the array aperture. The compensation phase of each element can be calculated independently in a closed-form formula, without the need for simultaneous solutions.

[0054] Step S10: Fuse and generate the current time slot phase shift matrix. Since the RIS reflection unit is shared by all K users, the individual compensation requirements of each user need to be merged into a unified RIS phase shift adjustment. An equal-weighted averaging criterion is adopted: In the formula, For the first Time slot number The unified compensation phase after the fusion of each reflection unit; the input of this formula is The unit-by-unit compensation phase for each user outputs a fused compensation phase. Physically, this means balancing the compensation requirements of each user on a shared aperture, keeping the phase error of the cascaded link low across all users.

[0055] Alternatively, a weighted non-uniform fusion method, with each user's channel quality as the weight, can be used to adapt to the asymmetry of channel conditions among users.

[0056] The online compensation phase shift diagonal matrix for the nth time slot is generated by fusion compensation phase: In the formula, The imaginary unit, To compensate for the phase shift diagonal matrix, its first... The diagonal element is the first The unit mode complex phase shift of each reflecting unit; the input of this equation is Each phase is fused and compensated, and the output is a compensated phase shift diagonal matrix.

[0057] Multiplying the online compensation phase shift matrix by the offline static phase shift matrix yields the total phase shift matrix applied to the RIS in the current time slot: In the formula, This is the RIS static phase shift matrix output during the offline phase, where matrix multiplication is equivalent to adding the static phase angle of each reflection unit to the compensation phase angle, which remains unchanged during the communication service period; For the first The total phase shift matrix of the RIS in the time slot.

[0058] Step S11: Using the offline output of the simulated beamforming matrix Digital beamforming matrix Phase shift matrix of the current time slot Perform downlink communication in the current time slot, then proceed to the next time slot and return to step S4.

[0059] Device Implementation like Figure 5 As shown, the present invention also provides a beamforming apparatus for implementing the above method, comprising: The offline optimization module is used to perform steps S2–S3: establishing an inertial coordinate system. With body coordinate system Determine the nominal attitude matrix It also calculates the channel under nominal attitude, solves the nominal attitude energy efficiency maximization problem offline, and outputs the simulated beamforming matrix. 、 Digital beamforming matrix With RIS static phase shift matrix And it remains unchanged during the communication service period.

[0060] The attitude reconstruction module is used to perform steps S4–S6: acquiring attitude data in each time slot. Observation vector of direction and confidence weight Construct the Wahba pose matrix Then, singular value decomposition is performed on it, and the instantaneous attitude matrix is ​​reconstructed in closed form according to the above Wahba-SVD closed formula. .

[0061] The orientation deviation calculation module is used to perform steps S7–S8: based on the reconstructed attitude. With nominal attitude Calculate the nominal and instantaneous system Loss vectors and extract the corresponding beam angles to obtain the incident link direction vector deviation. Deviation from the reflection link direction vector of each user .

[0062] The phase compensation and fusion module is used to perform steps S9–S10: calculate the compensation phase of each reflection unit for each user according to the closed geometric relationship, and fuse the compensation phases of multiple users to generate a compensation phase shift diagonal matrix. and with the offline static phase shift matrix Multiplying them yields the RIS phase shift matrix for the current time slot. And output to the RIS controller.

[0063] The device can be centrally deployed on the base station side. After the base station completes all offline optimization and online compensation calculations, it sends the RIS phase shift matrix Φ[n] to the RIS controller on the UAV via the control link. Alternatively, the module can also be deployed on the UAV side (e.g., onboard computing unit) to complete attitude reconstruction and phase compensation calculations locally.

[0064] The present invention also provides a computer-readable storage medium and an electronic device. The computer-readable storage medium stores a computer program, which, when executed by a processor, implements the robust beamforming method under wind disturbance described in the steps above. The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the above method. The electronic device is, for example, a baseband processing unit on a base station side or an airborne embedded computing platform.

[0065] Simulation verification In a specific embodiment of the present invention, the system consists of a single base station, K It consists of 8 single-antenna sea surface users and a single drone equipped with a passive RIS (Radio Retrieval System). Base station configuration. root antenna and One radio frequency chain; RIS has One passive reflector unit. Carrier frequency. GHz, system bandwidth B = 100 MHz. Maximum total transmit power of the base station. dBm. Drone hovering altitude 100 m. Sea breeze model using Dryden gusts.

[0066] Simulation results show (see) Figure 6 At maximum transmission power At dBm: without jitter (σ=0°), the energy efficiency of this invention is approximately 102.87 Mbit / J; at σ=1.5°, the energy efficiency is approximately 101.01 Mbit / J, significantly better than the comparative scheme; at σ=3°, the energy efficiency is approximately 96.18 Mbit / J, still maintaining high energy efficiency performance. This verifies the effective suppression capability of this invention against cascaded channel degradation under wind disturbance conditions.

[0067] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A robust beamforming method for UAVs equipped with intelligent reflective surfaces based on Wahba attitude reconstruction under wind disturbance, characterized in that, A wireless communication system for use with a drone equipped with a smart reflective surface is provided. The system includes a base station equipped with an antenna array and employing an analog / digital hybrid beamforming architecture, a drone equipped with a passive smart reflective surface, and multiple users. The smart reflective surface is fixedly connected to the drone's body. The method includes an offline static phase and an online dynamic phase. Offline static phase: Establish an inertial coordinate system and a UAV body coordinate system. Determine the nominal attitude matrix based on the statistical average of the attitude jitter angles under wind disturbance. Calculate the base station to smart reflective surface channel, the smart reflective surface to each user channel, and the base station to each user direct connection channel under the nominal attitude. Under the nominal attitude matrix, an energy efficiency maximization beamforming optimization problem is established and solved, and the simulated beamforming matrix, digital beamforming matrix and intelligent reflective surface static phase shift matrix are output, which remain unchanged during the communication service period. During the online dynamic phase, execution is performed cyclically, one time slot at a time. Obtain multiple pairs of unit direction observation vectors and their corresponding confidence weights. The unit direction observation vectors are unit vectors that express the same physical direction in the inertial coordinate system and the body coordinate system respectively, and the observation directions are not collinear. A pose matrix is ​​constructed by weighting the observation vectors, a singular value decomposition is performed on it, and a reconstructed pose matrix is ​​obtained based on the closed-form solution of the Wahba problem. The closed-form solution includes a diagonal correction factor to ensure that the reconstructed pose matrix belongs to a three-dimensional special orthogonal group. Based on the reconstructed attitude matrix, the nominal attitude matrix, and the line-of-sight direction vectors in the position coordinates or inertial coordinate systems of the base station, the UAV, and each user, the incident link direction deviation from the base station to the intelligent reflective surface and the reflection link direction deviation from the intelligent reflective surface to each user are calculated respectively. Based on the geometric inner product relationship between the position vector of each reflective unit of the intelligent reflective surface in the body coordinate system and the direction deviation, the compensation phase of each reflective unit for each user is calculated in a closed loop. A unified compensation phase is obtained by fusing the compensation phases of multiple users, a compensation phase shift diagonal matrix is ​​generated, and it is multiplied with the offline static phase shift matrix to form the total phase shift matrix of the intelligent reflective surface in the current time slot. The matrix multiplication is equivalent to adding the static phase angle of each reflective unit to the compensation phase angle. Downlink communication in the current time slot is performed using the analog beamforming matrix, the digital beamforming matrix, and the total phase shift matrix of the smart reflective surface in the current time slot.

2. The method according to claim 1, characterized in that, The nominal attitude matrix is ​​set according to the statistical average of the attitude jitter angle under wind disturbance. The attitude uncertainty is characterized by a set based on the Frobenius norm. The set is defined as a subset of a three-dimensional special orthogonal group centered on the nominal attitude matrix and with the attitude jitter radius as the norm sphere. In the online stage, the deterministic instantaneous attitude estimate of the current time slot is obtained by the closed-form solution of the Wahba problem, which replaces the time-slot processing of the attitude uncertainty set, and deterministic geometric phase compensation is performed for the current time slot.

3. The method according to claim 1, characterized in that, The energy efficiency maximization beamforming optimization problem aims to maximize system energy efficiency, which is the ratio of system power efficiency to total power consumption. Constraints include a service quality constraint that the signal-to-interference-plus-noise ratio (SNR) of each user is not lower than a threshold, a power constraint that the total transmit power is not greater than the maximum transmit power, a constant mode constraint for simulated beamforming, and a constant mode constraint for the intelligent reflector surface. The problem is solved using fractional programming and alternating optimization iterations. This includes introducing auxiliary variables to transform the rate function into an equivalent lower bound, and alternately updating the digital beamforming matrix, simulated beamforming matrix, and intelligent reflector surface static phase shift matrix during iteration. When the rate of change of energy efficiency between two adjacent iterations is less than a preset convergence threshold or the maximum number of iterations is reached, iteration stops, and the final simulated beamforming matrix, digital beamforming matrix, and intelligent reflector surface static phase shift matrix are output.

4. The method according to claim 1, characterized in that, The offline static phase also includes a power back-substitution mechanism: when the solution after iterative update does not meet the signal-to-interference-plus-noise ratio (SINR) constraint, the digital beamforming subproblem is resolved and the transmit power is increased until the SINR threshold is met or the maximum transmit power is reached.

5. The method according to claim 1, characterized in that, The calculation of the directional deviation includes: calculating the line-of-sight vector from the inertial frame to the machine system under the nominal attitude and the line-of-sight vector from the inertial frame to the machine system under the instantaneous attitude; converting the nominal and instantaneous line-of-sight vectors into unit direction vectors through unit direction mapping, and subtracting them to obtain the incident link directional deviation and the reflection link directional deviation; when the attitude deviation is within a small angle range of ±10°, the elevation angle deviation and azimuth angle deviation are directly calculated from the line-of-sight vector deviation using a linearized approximation formula.

6. The method according to claim 1, characterized in that, The fusion of multi-user compensation phases adopts an equal-weighted average criterion; or a weighted non-uniform fusion criterion with the channel quality of each user as the weight.

7. The method according to claim 1, characterized in that, The unit direction observation vector is derived from the carrier phase measurement of the global navigation satellite system, the airborne inertial measurement unit, or the visual odometry. The confidence weight is set according to the measurement accuracy or signal-to-noise ratio of the corresponding sensor.

8. A robust beamforming device with intelligent reflective surface for UAVs equipped with Wahba attitude reconstruction under wind disturbance, characterized in that, include: The offline optimization module is used to establish the inertial coordinate system and the UAV body coordinate system, determine the nominal attitude matrix and calculate the channel under the nominal attitude, solve the nominal attitude energy efficiency maximization problem offline, and output the simulated beamforming matrix, digital beamforming matrix and intelligent reflective surface static phase shift matrix, which remain unchanged during the communication service period. The attitude reconstruction module is used to acquire multiple pairs of unit direction observation vectors and confidence weights in each time slot, construct the Wahba attitude matrix and perform singular value decomposition on it, and reconstruct the instantaneous attitude matrix based on the closed decomposition with diagonal correction factor. The orientation deviation calculation module is used to calculate the orientation deviation of the incident link from the base station to the smart reflective surface and the orientation deviation of the reflected link from the smart reflective surface to each user based on the reconstructed attitude matrix, the nominal attitude matrix, and the line-of-sight orientation vectors in the position coordinates or inertial coordinate system of the base station, the UAV, and each user. The phase compensation and fusion module is used to calculate the compensation phase in a closed loop based on the geometric inner product relationship between the position vector of each reflection unit of the smart reflective surface and the direction deviation, fuse the compensation phase of multiple users to generate a compensation phase shift diagonal matrix, and multiply it with the static phase shift matrix to form the total phase shift matrix of the smart reflective surface in the current time slot, and output it to the smart reflective surface controller.

9. The apparatus according to claim 8, characterized in that, The device is deployed on the base station side, and after the base station completes offline optimization and online compensation calculations, it sends the total phase shift matrix of the intelligent reflective surface to the intelligent reflective surface controller on the UAV via the control link; or, the device is deployed on the UAV side, and the onboard computing unit completes attitude reconstruction and phase compensation calculations locally.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the robust beamforming method based on Wahba attitude reconstruction under wind disturbance as described in any one of claims 1 to 7.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the robust beamforming method based on Wahba attitude reconstruction under wind disturbance as described in any one of claims 1 to 7.

Images