Metamaterial reconfigurable dual-polarized antenna array and dynamic beam forming method thereof

Abstract

The invention provides a metamaterial reconfigurable dual-polarized antenna array and a dynamic beam forming method thereof. The dual-polarized antenna array is composed of a metamaterial resonance unit array, a reconfigurable polarization network, an intelligent control unit and an adaptive beam forming algorithm. The metamaterial unit structure with the dual-polarization characteristic is designed, and continuous adjustment of the polarization state and the radiation characteristic is achieved by integrating the liquid metal microchannel and the variable capacitance diode. A hierarchical control framework and a deep reinforcement learning algorithm are adopted, self-adaptive optimization control over an antenna array is achieved, and the characteristics of high-precision beam forming, quick response and high energy efficiency are provided while low system complexity is kept. Flexible regulation and control of the polarization state, high-precision beam forming, low system complexity, quick response, high energy efficiency and wide dynamic range can be realized at the same time, so that the requirement of a future communication system on a high-performance antenna is met.

Description

Technical Field

[0001] The present invention relates to the field of wireless communication technologies, and particularly to a metamaterial reconfigurable dual-polarized antenna array and its dynamic beamforming method. Background Art

[0002] Modern wireless communication systems require antennas to have the capabilities of multi-band operation, multi-polarization characteristics, reconfigurable functions, and intelligent beam control to adapt to the increasingly complex electromagnetic environment and diverse application scenarios. As an artificially designed electromagnetic material, metamaterials have electromagnetic characteristics that natural materials do not possess, such as negative refractive index, unconventional wavefront regulation, etc. These unique properties enable metamaterials to show great potential in the field of high-performance antenna design.

[0003] In the prior art, there are solutions such as PIN diode-based reconfigurable antenna arrays, liquid crystal material-based phased array antennas, and MEMS switch-based reconfigurable metasurface beamforming systems. Among them, the PIN diode-based reconfigurable antenna array integrates PIN diodes in the antenna radiation element or the feeding network and uses voltage to control its on and off states to achieve the switching of the antenna operating mode. The existing solutions adopt a dual-band design and can switch between frequency bands such as 2.4 GHz and 5.8 GHz. However, due to the ability to only achieve discrete state conversion, its flexibility and continuous regulation capabilities are limited.

[0004] The liquid crystal material-based phased array antenna utilizes the dielectric anisotropy characteristics of liquid crystal materials, and controls the arrangement of liquid crystal molecules by applying an external electric field, thereby changing the electromagnetic wave propagation characteristics and achieving continuously adjustable beam scanning. Beam control of ±60° can be achieved within the Ka band. However, since the response speed of liquid crystal materials is usually in the millisecond level, the real-time regulation performance is insufficient.

[0005] The MEMS switch-based reconfigurable metasurface beamforming system integrates microelectromechanical system (MEMS) switches into the metasurface structure and adjusts the electromagnetic response of the metasurface by controlling the switch states to achieve beam directional reflection. However, there are certain limitations in the reliability and service life of MEMS devices, and at the same time, the complex control circuit also increases the overall implementation difficulty of the system.

[0006] Meanwhile, the traditional solutions generally have the following problems: Insufficient polarization flexibility: Most existing reconfigurable antenna systems can only support a single polarization mode or limited polarization state switching, making it difficult to adapt to complex and changing electromagnetic environments. Especially in multipath propagation environments, polarization mismatch leads to serious signal attenuation. Limited beamforming accuracy: Although traditional phased arrays can achieve beam scanning, phase quantization errors and inter-element coupling effects limit the beamforming accuracy and sidelobe level control ability. High system complexity: Existing reconfigurable antenna systems usually require complex control circuits and feeding networks. Each radiation element requires an independent control line, and the system complexity increases exponentially with the array scale, greatly increasing the cost and power consumption. Summary of the Invention

[0007] Aiming at the deficiencies of the existing technology, the purpose of the present invention is to provide a metamaterial reconfigurable dual-polarization antenna array and its dynamic beamforming method to solve the problems proposed in the above background technology. The present invention can simultaneously achieve flexible regulation of polarization states, high-precision beamforming, low system complexity, fast response, high energy efficiency, and wide dynamic range to meet the requirements of future communication systems for high-performance antennas.

[0008] To achieve the above purpose, the present invention is realized through the following technical solutions: A metamaterial reconfigurable dual-polarization antenna array, characterized in that it includes a metamaterial resonance unit array, a reconfigurable polarization network, a varactor diode tuning network, a wireless power supply and control system, an adaptive beamforming algorithm, and system integration and packaging. The metamaterial resonance unit array is a two-dimensional or three-dimensional array composed of multiple reconfigurable metamaterial units, and each unit can independently regulate the amplitude, phase, and polarization characteristics of electromagnetic waves. The reconfigurable polarization network consists of liquid metal microchannels and micro piezoelectric pumps, and by controlling the distribution of liquid metal in the microchannels, continuous adjustment of the polarization characteristics of the units can be achieved. The varactor diode tuning network is integrated into the metamaterial unit, and by adjusting the bias voltage to control the capacitance value of the varactor diode, precise regulation of the resonance frequency and phase response of the unit can be achieved. The wireless power supply and control system uses magnetic coupling resonance wireless power supply technology and low-power Bluetooth communication to achieve wireless power supply and control of the antenna units. The adaptive beamforming algorithm is based on a deep reinforcement learning framework, combined with electromagnetic field theory and statistical signal processing methods, to achieve real-time optimization and regulation of the antenna array parameters. The system integration and packaging adopt a heterogeneous integration scheme combining multilayer PCB technology and 3D printing technology.

[0009] Furthermore, the metamaterial resonance unit adopts a three-layer structure: the top layer is a metal resonator pattern, the middle layer is a dielectric substrate, and the bottom layer is a ground plane. The metal resonator adopts an improved "cross" structure, where the horizontal and vertical arms respectively correspond to two orthogonal polarization directions. Varactor diodes are integrated into the horizontal and vertical arms of the metal resonator respectively. By adjusting their bias voltages, the resonance frequency and phase response can be changed. Varactor diodes D1 and D2 respectively control the response characteristics of horizontal and vertical polarizations.

[0010] Furthermore, the design parameters of the metamaterial unit are optimized through electromagnetic simulation. The specific parameters are as follows:

[0011] Unit period: a = 0.2λ, where λ is the wavelength corresponding to the operating frequency;

[0012] Resonator arm length: L = 0.15λ, resonator arm width: w = 0.02λ;

[0013] Dielectric substrate thickness: h = 0.04λ;

[0014] Dielectric substrate material: Rogers RO4350B, εr = 3.66, tanδ = 0.0037;

[0015] Liquid metal material: gallium-indium alloy, conductivity σ = 3.4×10 6 S / m;

[0016] Varactor diode: Skyworks SMV1405, capacitance range: 0.63 pF to 2.67 pF.

[0017] Furthermore, the reconfigurable polarization network consists of microchannels made of PDMS material, liquid metal, and a micro piezoelectric pump. The microchannels connect the four quadrants of the resonator to form a "#" shape structure. By controlling the distribution position of the liquid metal in different channels, the response characteristics of the metamaterial unit to different polarized electromagnetic waves can be changed. When the liquid metal fills the horizontal channels and the vertical channels are empty, the unit exhibits horizontal polarization characteristics; when the liquid metal fills the vertical channels and the horizontal channels are empty, the unit exhibits vertical polarization characteristics; when the distribution ratio of the liquid metal in the two-direction channels changes, any linear polarization state can be achieved.

[0018] Furthermore, the metamaterial units are arranged in a two-dimensional planar array and adopt a hybrid integration architecture: each 3×3 sub-array shares a control module, and multiple sub-arrays form a complete antenna system. The array design parameters are as follows:

[0019] Array size: 12×12, a total of 144 units;

[0020] Sub-array size: 3×3, a total of 9 units;

[0021] Element spacing: d = 0.5λ, satisfying the condition of no grating lobes;

[0022] Array arrangement: Planar rectangular array;

[0023] Feeding method: Hybrid feeding network combining microstrip line and probe.

[0024] Furthermore, the control system adopts a hierarchical architecture, including three levels:

[0025] Bottom layer: Element control layer, responsible for controlling the bias voltage of the varactor diode and the distribution of liquid metal of a single metamaterial element;

[0026] Middle layer: Sub-array control layer, responsible for coordinating the working states of sub-arrays composed of 9 elements;

[0027] Top layer: System control layer, responsible for executing the global beamforming algorithm and performance optimization. The control system hardware consists of a main controller and multiple sub-controllers;

[0028] The main controller communicates with the sub-controllers through a wired interface. The sub-controllers control the bias voltage of the varactor diode through a digital-to-analog converter and control the working state of the micro piezoelectric pump through a pulse width modulation signal.

[0029] Furthermore, this dual-polarization antenna array adopts the magnetic coupling resonance wireless power supply technology. The transmitting end is set on the back of the array and consists of an LC resonance circuit and a driving circuit; the receiving end is integrated in each sub-array control module and consists of a receiving coil, a rectifying circuit and a power management chip. The communication adopts the low-power Bluetooth protocol, and a star network topology is formed between the sub-array control module and the main controller.

[0030] A dynamic beamforming method for the above antenna array, including the following steps:

[0031] S1. Environment perception and modeling: Obtain the current electromagnetic environment information through channel sounding, including the positions of interference sources, multipath characteristics, and user distributions, and establish a real-time model of the electromagnetic environment;

[0032] S2. Beamforming target definition: According to application requirements, define the optimization targets of beamforming, including maximizing the signal-to-noise ratio, minimizing interference, and covering multiple users;

[0033] S3. State space definition: Map the configuration parameters of the metamaterial elements to the state space to form high-dimensional control variables. The configuration parameters of the metamaterial elements are the capacitance value of the varactor diode and the distribution of liquid metal;

[0034] S4. Reward function design: Design a reward function reflecting the beamforming effect, comprehensively considering factors such as main lobe gain, sidelobe level, polarization matching degree, and system power consumption;

[0035] S5. Reinforcement learning process: The deep Q-network algorithm is adopted to continuously optimize the control strategy through interaction with the environment, realizing iterative improvement of beamforming performance;

[0036] S6. Fast convergence strategy: Combining electromagnetic theory knowledge, heuristic rules are designed to accelerate learning convergence and improve the system response speed.

[0037] Furthermore, it also includes a metamaterial beamforming algorithm based on multi-modal reinforcement learning. In this algorithm, a core mathematical model is established. The core mathematical model includes a metamaterial unit electromagnetic response model and an array far-field radiation model. The metamaterial unit electromagnetic response model defines the dual-polarization coupling Jones matrix of the reconfigurable metamaterial unit:

[0038]

[0039] The array far-field radiation model improves the space-frequency-polarization joint radiation equation:

[0040]

[0041] where Γ m,n (t) is the environmental perturbation compensation factor, satisfying:

[0042]

[0043] (H m,n is the channel impulse response).

[0044] Furthermore, it also includes state space design, action space optimization, and a multi-objective reward function. The state space design is:

[0045]

[0046] -C t ∈R M×N : varactor diode capacitance matrix

[0047] -L t ∈{0,1} M×N : liquid metal on-off state

[0048] Polarization parameters (A xx , A yy , φ xx , φ yy )

[0049] Environmental perturbation gradient

[0050] The action space optimization includes: proposing a hierarchical action coding mechanism:

[0051]

[0052] Capacitance adjustment: a C ∈[-0.1, 0.1] M×N (Relative change)

[0053] Liquid metal control: a L ∈{0, 1} M×N (0 = disconnected, 1 = connected)

[0054] Parameter update:

[0055] The multi-objective reward function is:

[0056]

[0057] Innovative weight strategy: (Time-varying adaptive weight).

[0058] Advantages of the present invention:

[0059] 1. Through the innovative design of the liquid metal microfluidic network, the present invention realizes continuous adjustment of the polarization state, supports not only horizontal / vertical linear polarization, but also any linear polarization angle and circular / elliptical polarization, with a polarization regulation range of 360°, which is 200% higher than the traditional scheme.

[0060] 2. Based on the phase fine regulation of varactor diodes and the optimization of the deep reinforcement learning algorithm, the beam pointing accuracy of the present invention reaches ±0.5°, and the sidelobe level can be controlled below -25 dB, improving the shaping accuracy by 40% compared with the traditional phased array.

[0061] 3. The present invention adopts a sub-array control architecture and wireless power supply technology, reducing the number of control lines by 85%, significantly reducing the system complexity, while maintaining the fine control ability at the unit level; the response time of the liquid metal microfluidic network is on the order of 10 ms, and the response time of the varactor diode tuning network is on the microsecond level. Combining with the hierarchical control architecture, it realizes the unity of fast response and wide regulation range.

[0062] 4. Compared with the traditional PIN diode scheme, the energy consumption of the varactor diode tuning network of the present invention is reduced by 70%, and the overall system efficiency is increased by 35%; adopting mature PCB technology and 3D printing technology, it avoids complex micro-nano processing processes, reduces the manufacturing cost, and improves the mass production possibility; based on the beam shaping algorithm of deep reinforcement learning, the system has a strong environmental adaptability, and can automatically optimize the working parameters according to the changes of the electromagnetic environment, with the adaptability increased by 150%.

[0063] 5. The present invention can simultaneously achieve multiple functions of frequency tuning, polarization control, and beamforming using a single system. The operating frequency band can cover 3 GHz - 6 GHz (67% relative bandwidth), far exceeding the 10 - 20% bandwidth of traditional solutions. Through heterogeneous integration design, the system volume is reduced by 30% and the weight is reduced by 25%, making it more suitable for applications in space - limited scenarios.

[0064] 6. Through the non - mechanical movement design of the liquid - metal microfluidic network and varactor diodes, the present invention significantly improves the system reliability and service life, and the theoretical service life is increased to more than 100,000 hours. Moreover, the present invention is superior to the prior art in key performance indicators, especially having significant advantages in polarization flexibility, beamforming accuracy, and system complexity, providing a new technical path for future high - performance communication systems. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] Figure 1 It is a schematic diagram of the metamaterial resonance unit structure in the present invention;

[0066] Figure 2 It is a schematic diagram of the reconfigurable polarization network structure;

[0067] Figure 3 It is a schematic diagram of the metamaterial antenna array structure;

[0068] Figure 4 It is a hierarchical architecture diagram of the control system;

[0069] Figure 5 It is a block diagram of the wireless power supply and communication system;

[0070] Figure 6 It is a flowchart of the dynamic beamforming algorithm;

[0071] Figure 7 It is a diagram of the system performance test results;

[0072] Figure 8 It is the radiation pattern in different operating modes;

[0073] Figure 9 It is a performance comparison between the present invention and the prior art;

[0074] Figure 10 It is a flowchart of the algorithm in the present invention;

[0075] Figure 11 It is a generated simulation comparison curve graph in the embodiment of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0076] To make the technical means, creative features, achieving purposes, and functions of the present invention easy to understand, the present invention will be further described below in conjunction with specific embodiments.

[0077] Please refer to Figures 1 to 11 , the present invention provides the following technical solutions:

[0078] Embodiment 1

[0079] The metamaterial reconfigurable dual-polarized antenna array consists of the following parts:

[0080] a) Metamaterial resonance unit array: A two-dimensional or three-dimensional array composed of multiple reconfigurable metamaterial units, each unit capable of independently regulating the amplitude, phase, and polarization characteristics of electromagnetic waves.

[0081] b) Reconfigurable polarization network: Composed of liquid metal microchannels and micro piezoelectric pumps, the continuous tunability of the unit polarization characteristics is achieved by controlling the distribution of liquid metal in the microchannels.

[0082] c) Varactor diode tuning network: Integrated into the metamaterial unit, the capacitance value of the varactor diode is controlled by adjusting the bias voltage to achieve precise regulation of the unit resonance frequency and phase response.

[0083] d) Wireless power supply and control system: Adopting magnetic coupling resonance wireless power supply technology and low-power Bluetooth communication to achieve wireless power supply and control of the antenna unit, reducing the complexity of the feeding network.

[0084] e) Adaptive beamforming algorithm: Based on the deep reinforcement learning framework, combined with electromagnetic field theory and statistical signal processing methods, to achieve real-time optimization and regulation of the antenna array parameters.

[0085] f) System integration and packaging: Adopting a heterogeneous integration scheme combining multi-layer PCB technology and 3D printing technology to achieve miniaturization, lightweight, and high reliability of the system.

[0086] This embodiment describes the structure of the above-mentioned metamaterial reconfigurable dual-polarized antenna array as follows:

[0087] a) Metamaterial resonance unit design

[0088] In this embodiment, a new type of dual-polarized metamaterial resonance unit is designed, and its core structure is as Figure 1 shown. The unit adopts a three-layer structure: the top layer is a metal resonator pattern, the middle layer is a dielectric substrate, and the bottom layer is a ground plane. The metal resonator adopts an improved "cross" structure, where the horizontal and vertical arms respectively correspond to two orthogonal polarization directions. Varactor diodes are integrated into the horizontal and vertical arms of the metal resonator respectively, and the resonance frequency and phase response can be changed by adjusting their bias voltages. Varactor diodes D1 and D2 respectively control the response characteristics of horizontal and vertical polarizations. In addition, the four quadrant regions of the resonator are connected by a reconfigurable polarization network composed of liquid metal microchannels.

[0089] The design parameters of the metamaterial unit are optimized through electromagnetic simulation, and the specific parameters are as follows:

[0090] · Unit period: a = 0.2λ (λ is the wavelength corresponding to the operating frequency)

[0091] · Resonator arm length: L = 0.15λ · Resonator arm width: w = 0.02λ

[0092] · Dielectric substrate thickness: h = 0.04λ

[0093] · Dielectric substrate material: Rogers RO4350B (εr = 3.66, tanδ = 0.0037)

[0094] · Liquid metal material: Gallium-indium alloy (conductivity σ = 3.4×10^6 S / m)

[0095] · Varactor diode: Skyworks SMV1405 (capacitance range: 0.63 pF to 2.67 pF)

[0096] b) Reconfigurable polarization network

[0097] The reconfigurable polarization network is realized by microfluidic technology, as Figure 2 shown. The network consists of microchannels made of PDMS material (polydimethylsiloxane), liquid metal, and a micro piezoelectric pump. The microchannels connect the four quadrants of the resonator, forming a "#" shaped structure. By controlling the distribution position of the liquid metal in different channels, the response characteristics of the metamaterial unit to different polarized electromagnetic waves can be changed. When the liquid metal fills the horizontal channels and the vertical channels are empty, the unit exhibits horizontal polarization characteristics; when the liquid metal fills the vertical channels and the horizontal channels are empty, the unit exhibits vertical polarization characteristics; when the distribution ratio of the liquid metal in the two-direction channels changes, any linear polarization state can be achieved; in particular, when a phase delay is introduced in a specific channel, circular polarization or elliptical polarization states can also be achieved.

[0098] c) Array structure design

[0099] The metamaterial units are arranged in a two-dimensional planar array, as Figure 3 shown. To reduce the coupling effect between units and improve the system integration, a hybrid integration architecture is adopted: each 3×3 sub-array shares a control module, and multiple sub-arrays form a complete antenna system. The array design parameters are as follows:

[0100] · Array size: 12×12 (144 units)

[0101] · Sub-array size: 3×3 (9 units)

[0102] · Unit spacing: d = 0.5λ (meeting the non-grating lobe condition)

[0103] · Array arrangement: Planar rectangular array

[0104] · Feeding method: Hybrid feeding network combining microstrip line and probe

[0105] d) Control system architecture The control system adopts a hierarchical architecture, as Figure 4 shown, including three levels:

[0106] 1. Bottom layer: Unit control layer, responsible for controlling the varactor diode bias voltage and liquid metal distribution of a single metamaterial unit

[0107] 2. Middle layer: Sub-array control layer, responsible for coordinating the working states of sub-arrays composed of 9 units

[0108] 3. Top layer: System control layer, responsible for executing the global beamforming algorithm and performance optimization The control system hardware consists of a main controller (based on the ARM Cortex-M4 kernel) and multiple sub-controllers (based on low-power MCUs).

[0109] The main controller communicates with the sub-controllers through a wired interface (SPI or I2C). The sub-controllers control the varactor diode bias voltage through a digital-to-analog converter and control the working state of the micro piezoelectric pump through a pulse width modulation signal.

[0110] e) Wireless power supply and communication unit

[0111] To reduce the system complexity, the present invention adopts the magnetic coupling resonance wireless power supply technology, as Figure 5 shown. The transmitting end is arranged on the back of the array and consists of an LC resonance circuit and a driving circuit; the receiving end is integrated in each sub-array control module and consists of a receiving coil, a rectifying circuit and a power management chip.

[0112] The communication adopts the low-power Bluetooth protocol, and a star network topology is formed between the sub-array control module and the main controller to achieve efficient transmission of configuration instructions.

[0113] Embodiment 2

[0114] This embodiment also provides a dynamic beamforming method based on deep reinforcement learning, which combines electromagnetic field theory and statistical signal processing technology to achieve adaptive optimization control of antenna array parameters. The method includes the following steps:

[0115] 1. Environment perception and modeling: Obtain the current electromagnetic environment information through channel detection, including the positions of interference sources, multipath characteristics, user distribution, etc., and establish a real-time model of the electromagnetic environment.

[0116] 2. Beamforming objective definition: According to application requirements, define the optimization objectives of beamforming, such as maximizing the signal-to-noise ratio, minimizing interference, covering multiple users, etc.

[0117] 3. State space definition: Map the configuration parameters of the metamaterial unit (varactor diode capacitance value and liquid metal distribution) to the state space to form high-dimensional control variables.

[0118] 4. Reward function design: Design a reward function that reflects the beamforming effect, comprehensively considering factors such as main lobe gain, side lobe level, polarization matching degree, and system power consumption.

[0119] 5. Reinforcement learning process: Adopt the Deep Q-Network (DQN) algorithm to continuously optimize the control strategy through interaction with the environment, and achieve iterative improvement of beamforming performance.

[0120] 6. Fast convergence strategy: Combine electromagnetic theory knowledge to design heuristic rules to accelerate learning convergence and improve the system response speed.

[0121] Metamaterial Beamforming Algorithm Based on Multimodal Reinforcement Learning

[0122] 1. Innovation in Core Mathematical Model

[0123] 1.1 Electromagnetic Response Model of Metamaterial Unit

[0124] Define the dual-polarization coupling Jones matrix of the reconfigurable metamaterial unit:

[0125]

[0126] In this embodiment:

[0127] - Introduce a dynamic phase compensation term and a cross-polarization coupling term

[0128] - The time-varying parameters are jointly regulated by varactor diodes (capacitance C m,n ) and liquid metals (inductance L m,n )

[0129] 1.2 Array Far-Field Radiation Model

[0130] Improved space-frequency-polarization joint radiation equation:

[0131]

[0132] where Γ m,n (t) is the environmental perturbation compensation factor, satisfying:

[0133]

[0134] (Hm,n (for the channel impulse response)

[0135] 2. Innovation of the Reinforcement Learning Framework

[0136] 2.1 State Space Design

[0137]

[0138] - C t ∈ R M×N : varactor diode capacitance matrix - L t ∈ {0, 1} M×N : on - off state of liquid metal Polarization parameters (A xx , A yy , φ xx , φ yy )

[0139] Environmental disturbance gradient

[0140] 2.2 Action Space Optimization

[0141] A hierarchical action coding mechanism is proposed:

[0142]

[0143] - Capacitance adjustment: a C ∈ [-0.1, 0.1] M×N (relative change)

[0144] - Liquid metal control: a L ∈ {0, 1} M×N (0 = off, 1 = on) - Polarization parameter update:

[0145] 2.3 Multi - objective Reward Function

[0146]

[0147]

[0148] Innovation weight strategy: (time - varying adaptive weight)

[0149] 3. Algorithm Flow

[0150] Hybrid learning architecture:

[0151] - Scenario - adaptive switching between centralized DQN (suitable for static scenarios) and distributed MADDPG (suitable for dynamic multi - users)

[0152] - Innovative network structure:

[0153]

[0154] This code defines the structure of a hybrid neural network model, HybridNN. This model combines:

[0155] · Graph Attention Layer: Used to understand the spatial structure or relationships of the input.

[0156] · LSTM layer (nn.LSTM): Used to process sequence information, especially the temporal features related to the "polarization state".

[0157] · Dual Heads MLP: Used to implement the Actor-Critic architecture and output both action decisions (policies) and value evaluations simultaneously.

[0158] This structure indicates that the model may be designed to solve a complex task that involves both spatial correlations and time series dependencies and requires decision-making through reinforcement learning.

[0159] 2. Three-dimensional performance evaluation system:

[0160] - Electromagnetic performance (gain / sidelobe)

[0161] - Communication quality (BER / throughput)

[0162] - Energy consumption efficiency (Watt / bit)

[0163] Evaluation metrics:

[0164]

[0165] 4.1 Comparison with existing technologies

[0166] Technical indicators Traditional methods (such as SMI) The present invention Response time > 100 ms <5ms Polarization matching degree 0.6-0.8 0.92+0.05 Multi - user interference suppression ratio 15 dB 28 dB Power consumption 40W 22W

[0167] The generated experimental data is as Figure 11 shown.

[0168] Example 3

[0169] This example also provides a beamforming algorithm based on the dynamic regulation of intelligent metasurfaces, which combines polarization regulation and spatial beamforming to achieve joint optimization under the guidance of deep reinforcement learning and heuristic rules, thereby improving the anti-interference and spectrum utilization performance of wireless communication systems. The following elaborates in detail the mathematical model, state-action definition, reward design, optimization objective, and overall closed-loop optimization process of this algorithm.

[0170] 1. Construction of the mathematical model

[0171] 1.1 Unit response model of the metamaterial array

[0172] For an intelligent metasurface composed of M×N units, the electromagnetic response of each unit is described by the Jones matrix. Let the response of the (m,n) unit at frequency ω be:

[0173]

[0174] where A ij represents the amplitude response, φ ij represents the phase response, and i,j∈{x,y} correspond to different polarization components.

[0175] 1.2 Array far-field radiation pattern

[0176] The far-field radiation pattern F(θ,φ) of the entire metasurface array can be expressed as the superposition of all unit responses. Considering the incident field E inc

[0177] and the corresponding phase compensation:

[0178]

[0179] where:

[0180] · is the free-space wave number,

[0181] ·d x ,d y are the spacings of the array units in the x and y directions respectively,

[0182] ·θ and φ are the zenith angle and azimuth angle of the radiation direction respectively.

[0183] 2. Definition of the joint optimization problem

[0184] To achieve polarization control and beamforming simultaneously, this embodiment constructs a joint optimization problem whose objective is to optimize the reflection coefficient of the IRS (intelligent metasurface) to maximize the main lobe gain while suppressing the sidelobe interference, and to achieve the optimal polarization matching under the conditions of power consumption and energy constraints.

[0185] The specific objective function is defined as:

[0186]

[0187] where the expression of SINR is:

[0188]

[0189] ·h total =Hd +H r ΦH i denotes the total channel,

[0190] · Φ is the IRS reflection coefficient matrix, and its block form is

[0191]

[0192] where the elements of each sub - matrix are given by

[0193]

[0194] and satisfy the energy conservation constraint:

[0195] |β HH | 2 +|β HV | 2 ≤1,|β VH | 2 +|β VV | 2 ≤1,θ pq ∈[0,2π)

[0196] In addition, in this embodiment, an energy efficiency metric is introduced in the joint optimization:

[0197]

[0198] where R represents the communication rate, P t and P IRS are the transmit power and the IRS power consumption respectively.

[0199] 3. State - Action and Reward Design under the Reinforcement Learning Framework

[0200] To achieve adaptive optimization, the present invention defines a state space S, an action space A, and a reward function R under the reinforcement learning framework:

[0201] 3.1 Definition of the State Space

[0202] The state space S is composed of the parameter states of each meta - surface unit:

[0203] S={s t ∣s t =[C 1,1 ,C 1,2 ,…,C M,N ,L 1,1 ,L 1,2 ,…,L M,N}

[0204] Where:

[0205] ·Cm,n represents the capacitance value of the variable capacitance element in the (m,n) unit,

[0206] ·L m,n represents a parameter related to the distribution state of the liquid metal (or other variables related to polarization control).

[0207] 3.2 Action Space Definition

[0208] The action space AA is the adjustment amount of the above state parameters:

[0209] A = {a t | a t = [ΔC 1,1 , …, ΔC M,N , ΔL 1,1 , …, ΔL M,N}

[0210] where ΔC m,n and ΔL m,n respectively represent the adjustment of the parameters of the (m,n) unit.

[0211] 3.3 Reward Function Design

[0212] The reward function R(s t , a t ) comprehensively considers the main lobe gain, sidelobe suppression, polarization matching degree, and energy consumption cost, and its design is as follows:

[0213] R(s t , a t ) = w1·G main - w2·G side + w3·P match - w4·P power

[0214] where:

[0215] ·G main represents the main lobe gain;

[0216] ·G side represents the sidelobe level, and usually the lower the better;

[0217] ·P match represents the polarization matching degree;

[0218] ·P power represents the system power consumption;

[0219] ·w i are positive weight coefficients used to balance various indicators.

[0220] 4. Optimization Process and Closed-loop Regulation

[0221] The present invention adopts a hybrid optimization method combining deep reinforcement learning (such as Deep Deterministic Policy Gradient DDPG) with heuristic rules to form a closed-loop dynamic regulation process. The overall process mainly includes the following links:

[0222] 1. Environmental perception

[0223] Obtain real-time environmental data through channel detection, establish a system state model, and construct a state vector sts_t.

[0224] 2. State evaluation and target determination

[0225] Determine the target parameters of beamforming and polarization regulation according to the current state and user requirements.

[0226] 3. Action selection and parameter update

[0227] Use the policy network to select the optimal action ata_t, and map the action to the IRS configuration parameters. Further refine the output through heuristic rules (such as phase compensation, array weighting, and polarization matching rules), and update the IRS unit parameters.

[0228] 4. Performance evaluation and reward feedback

[0229] Calculate performance metrics such as the current SINR and spectral efficiency of the system, and feedback them to the policy network according to the reward function to adjust and optimize the policy.

[0230] 5. Closed-loop adaptive update

[0231] When the system performance does not reach the preset target, enter the optimization loop; when the target is reached, save the optimal configuration and restart the regulation process when the environment changes.

[0232] Example 4

[0233] This embodiment also provides the following solution: In the design of the metamaterial resonance unit, a multi-resonant ring structure can be used to replace the "cross" resonator. Specifically, three nested split-ring resonators (SRRs) are used, and varactor diodes are integrated on each ring. The frequency and phase are regulated by controlling the resonance characteristics of each ring. The polarization network uses a PIN diode switch array to replace the liquid metal microchannel, and the polarization characteristics are regulated by controlling the switch state to change the current path. The advantage of this alternative solution is that the manufacturing process is more mature and does not require a complex microfluidic structure; the disadvantage is that the polarization regulation shows a discrete state, with lower flexibility than the liquid metal solution, and the loss introduced by the PIN diode is relatively large, reducing the energy efficiency.

[0234] The key parameters of alternative solution 1 are as follows:

[0235] · Resonant ring diameter: The outer ring D1 = 0.18λ, the middle ring D2 = 0.12λ, and the inner ring D3 = 0.08λ

[0236] · Ring width: w = 0.01λ

[0237] · PIN diode: BAR64 - 02V (switching time < 100ns)

[0238] · Control circuit: Each unit requires 8 control lines (an increase of 5 compared to the original 3 - line solution).

[0239] Embodiment 5

[0240] In this embodiment, in terms of the beamforming algorithm, a genetic algorithm combined with a fast electromagnetic calculation model can be used to replace the deep reinforcement learning method. Specifically, an equivalent circuit model of the metamaterial unit is established, and based on this model, the radiation characteristics of the array are calculated quickly. Then, the genetic algorithm is used to search for the optimal parameter configuration.

[0241] This scheme defines the chromosome encoding to represent the metamaterial unit parameters, and continuously evolves through operations such as crossover and mutation. With a specific objective function (such as main lobe gain, side lobe level, etc.) as the fitness evaluation criterion, the optimal solution is iteratively searched.

[0242] The key parameter settings of the genetic algorithm are as follows:

[0243] Population size: 200

[0244] Crossover probability: 0.8

[0245] Mutation probability: 0.1

[0246] Maximum number of iterations: 500

[0247] Fitness function: F = α·G_main - β·G_side + γ·P_match - δ·P_power

[0248] The advantage of this alternative scheme is that the algorithm has good stability and is not easily trapped in local optima; the disadvantage is that the computational complexity is relatively high, the real - time performance is slightly poor, and the adaptive ability to the dynamic environment is not as good as the reinforcement learning method. The schemes provided in Embodiment 4 and Embodiment 5 can both achieve the technical objectives of the present invention to a certain extent, but there are differences from the optimal scheme of the present invention in terms of performance, complexity, or cost. They can be used as technical alternative solutions of the present invention and are applicable to different application scenarios and manufacturing conditions

[0249] The above has shown and described the basic principles, main features, and advantages of the present invention. For those skilled in the art, it is obvious that the present invention is not limited to the details of the above - mentioned exemplary embodiments, and without departing from the spirit or basic features of the present invention, the present invention can be implemented in other specific forms.

[0250] In addition, it should be understood that although this specification is described in terms of embodiments, not every embodiment only contains an independent technical solution. This narrative manner of the specification is only for clarity. Those skilled in the art should regard the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. Reconfigurable dual-polarized antenna array made of metamaterials, characterized in that: It includes a metamaterial resonance unit array, a reconfigurable polarization network, a varactor diode tuning network, a wireless power supply and control system, an adaptive beamforming algorithm, and system integration and packaging. The metamaterial resonance unit array is a two-dimensional or three-dimensional array composed of multiple reconfigurable metamaterial units, and each unit can independently control the amplitude, phase, and polarization characteristics of electromagnetic waves. The reconfigurable polarization network consists of liquid metal microchannels and micro piezoelectric pumps. By controlling the distribution of liquid metal in the microchannels, continuous tunability of the unit polarization characteristics is achieved. The varactor diode tuning network is integrated into the metamaterial unit. By adjusting the bias voltage to control the capacitance value of the varactor diode, precise control of the unit resonance frequency and phase response is realized. The wireless power supply and control system uses magnetic coupling resonance wireless power supply technology and low-power Bluetooth communication to achieve wireless power supply and control of the antenna unit. The adaptive beamforming algorithm is based on a deep reinforcement learning framework, combining electromagnetic field theory and statistical signal processing methods to achieve real-time optimization and control of the antenna array parameters. The system integration and packaging adopt a heterogeneous integration scheme that combines multi-layer PCB technology and 3D printing technology.

2. The metamaterial reconfigurable dual-polarized antenna array according to claim 1, wherein The metamaterial resonance unit adopts a three-layer structure: the top layer is a metal resonator pattern, the middle layer is a dielectric substrate, and the bottom layer is a ground plane. The metal resonator adopts an improved "cross" structure, where the horizontal and vertical arms respectively correspond to two orthogonal polarization directions. Varactor diodes are integrated into the horizontal and vertical arms of the metal resonator. By adjusting their bias voltages, the resonance frequency and phase response can be changed. Varactor diodes D1 and D2 respectively control the response characteristics of horizontal and vertical polarizations.

3. The metamaterial reconfigurable dual-polarized antenna array according to claim 2, wherein: The design parameters of the metamaterial unit are optimized through electromagnetic simulation, and the specific parameters are as follows: Unit period: a = 0.2λ, where λ is the wavelength corresponding to the operating frequency; Resonator arm length: L = 0.15λ, resonator arm width: w = 0.02λ; Dielectric substrate thickness: h = 0.04λ; Dielectric substrate material: Rogers RO4350B, εr = 3.66, tanδ = 0.0037; Liquid metal material: gallium-indium alloy, electrical conductivity σ = 3.4×10 6 S / m; Varactor diode: Skyworks SMV1405, capacitance range: 0.63 pF to 2.67 pF.

4. The metamaterial reconfigurable dual-polarized antenna array according to claim 1, wherein: The reconfigurable polarization network consists of microchannels made of PDMS material, liquid metal, and micro piezoelectric pumps. The microchannels connect the four quadrants of the resonator to form a "#" shape structure. By controlling the distribution position of liquid metal in different channels, the response characteristics of the metamaterial unit to different polarized electromagnetic waves are changed. When the liquid metal fills the horizontal channels and the vertical channels are empty, the unit exhibits horizontal polarization characteristics. When the liquid metal fills the vertical channels and the horizontal channels are empty, the unit exhibits vertical polarization characteristics. When the distribution ratio of liquid metal in the two-direction channels changes, any linear polarization state can be achieved.

5. The metamaterial reconfigurable dual-polarized antenna array according to claim 1, characterized in that: The metamaterial units are arranged in a two-dimensional planar array and adopt a hybrid integration architecture: each 3×3 sub-array shares a control module, and multiple sub-arrays form a complete antenna system. The array design parameters are as follows: Array size: 12×12, a total of 144 units; Sub - array size: 3×3, with a total of 9 units; Element spacing: d = 0.5λ, satisfying the condition of no grating lobes; Array arrangement: Planar rectangular array; Feeding method: Hybrid feeding network combining microstrip lines and probes.

6. The metamaterial reconfigurable dual-polarized antenna array according to claim 1, characterized in that, The control system adopts a hierarchical architecture, including three levels: Bottom layer: Element control layer, responsible for controlling the varactor diode bias voltage and liquid metal distribution of a single metamaterial element; Middle layer: Sub - array control layer, responsible for coordinating the working states of the sub - array composed of 9 units; Top layer: System control layer, responsible for the execution of the global beamforming algorithm and performance optimization. The control system hardware consists of a main controller and multiple sub - controllers; The main controller communicates with the sub - controllers through a wired interface. The sub - controllers control the varactor diode bias voltage through a digital - to - analog converter and control the working state of the micro - piezoelectric pump through a pulse - width modulation signal.

7. The metamaterial reconfigurable dual-polarized antenna array according to claim 1, characterized in that: This dual - polarized antenna array adopts the magnetic - coupling resonance - based wireless power supply technology. The transmitting end is set on the back of the array and consists of an LC resonance circuit and a driving circuit; the receiving end is integrated in each sub - array control module and consists of a receiving coil, a rectifying circuit, and a power management chip. The communication uses the low - power Bluetooth protocol, and a star - shaped network topology is formed between the sub - array control module and the main controller.

8. A dynamic beamforming method for the antenna array according to claim 1, characterized in that, Including the following steps: S1. Environmental perception and modeling: Obtain the current electromagnetic environment information through channel sounding, including the positions of interference sources, multipath characteristics, and user distributions, and establish a real - time model of the electromagnetic environment; S2. Beamforming target definition: According to application requirements, define the optimization targets of beamforming, including maximizing the signal - to - noise ratio, minimizing interference, and covering multiple users; S3. State - space definition: Map the configuration parameters of the metamaterial elements to the state - space to form high - dimensional control variables. The configuration parameters of the metamaterial elements are the varactor diode capacitance value and the liquid metal distribution; S4. Reward function design: Design a reward function reflecting the beamforming effect, comprehensively considering factors such as main - lobe gain, side - lobe level, polarization matching degree, and system power consumption; S5. Reinforcement learning process: Adopt the deep Q - network algorithm to continuously optimize the control strategy through interaction with the environment, and achieve iterative improvement of beamforming performance; S6. Fast - convergence strategy: Combine electromagnetic theory knowledge to design heuristic rules to accelerate learning convergence and improve the system response speed.

9. The dynamic beamforming method according to claim 8, wherein: It also includes a metamaterial beamforming algorithm based on multi - modal reinforcement learning. In this algorithm, a core mathematical model is established. The core mathematical model includes a metamaterial element electromagnetic response model and an array far - field radiation model. The metamaterial element electromagnetic response model defines the dual - polarized coupling Jones matrix of the reconfigurable metamaterial element: The array far - field radiation model improves the space - frequency - polarization joint radiation equation: where Γ m,n (t) is the environmental disturbance compensation factor, satisfying: (H m,n is the channel impulse response).

10. The dynamic beamforming method according to claim 8, wherein: It also includes state - space design, action - space optimization, and a multi - objective reward function. The state - space design is as follows: -C t ∈R M×N : varactor diode capacitance matrix -L t ∈ {0, 1} M×N : On / Off state of liquid metal - Polarization parameters (A xx , A yy , φ xx , φ yy ) - Environmental disturbance gradient The action - space optimization includes: Proposing a hierarchical action coding mechanism: Capacitance adjustment: a C ∈[-0.1, 0.1] M×N (Relative change amount) Liquid metal control: a L ∈ {0, 1} M×N (0 = off, 1 = on) Parameter update: The multi - objective reward function is: Innovative weight strategy: (Time-varying adaptive weight).

Images